Designed Cost Function Research Articles

Full-waveform inversion by model extension: Theory, design, and optimization

We describe a new method, full-waveform inversion by model extension (FWIME), that recovers accurate acoustic subsurface velocity models from seismic data when conventional methods fail. We leverage the advantageous convergence properties of wave equation migration velocity analysis (WEMVA) with the accuracy and high-resolution nature of acoustic full-waveform inversion (FWI) by combining them into a robust mathematically consistent workflow with minimal need for user inputs. The novelty of FWIME resides in the design of a new cost function and a novel optimization strategy to combine the two techniques, making our approach more efficient and powerful than applying them sequentially. We observe that FWIME mitigates the need for accurate initial models and low-frequency long-offset data, which can be challenging to acquire. Our new objective function contains two components. First, we modify the forward mapping of the FWI problem by adding a data-correcting term computed with an extended demigration operator, whose goal is to ensure phase matching between predicted and observed data, even when the initial model is inaccurate. The second component, which is a modified WEMVA cost function, allows us to progressively remove the contributions of the data-correcting term throughout the inversion process. The coupling between the two components is handled by the variable projection method, which reduces the number of adjustable hyperparameters, thereby making our solution simple to use. We devise a model-space multiscale optimization scheme by reparameterizing the velocity model on spline grids to control the resolution of the model updates. We generate three cycle-skipped 2D synthetic data sets, each containing only one type of wave (transmitted, reflected, or refracted), and we analyze how FWIME successfully recovers accurate solutions with the same procedure for all three cases. In a second paper, we apply FWIME to challenging realistic examples where we simultaneously invert all wave modes.

Distributed sliding-mode cloud predictive formation control of networked multi-agent systems with application to air-bearing spacecraft simulators

This paper investigates the application of distributed sliding-mode cloud predictive control scheme in networked multi-agent control systems (NMACS). A sliding-mode cloud predictive control (SMCPC) scheme based on the cloud control system is designed. Combining the advantages of variable structure control (VSC) of sliding-mode control(SMC) and the advantages of cloud predictive control (CPC) in handling network delays, the network delay is actively compensated while ensuring the stability and output consistency of the NMACS. Also taking advantage of the small online computation of SMC, reduces the burden of CPC in processing big data and increases the computation rate. In addition, the scheme includes multi-step state prediction and output prediction as well as coordinated optimization among multiple agents. By optimizing the cost function design, the SMCPC scheme is guaranteed to achieve the desired control performance in multi-agent formation control. And the feasibility of the control scheme is verified by stability and consistency analysis. The applicability of the control scheme in practical applications is verified by air-bearing spacecraft simulator (ABS) formation control experiments.

Improved Particle Swarm Path Planning Algorithm with Multi-Factor Coupling in Forest Fire Spread Scenarios

In this paper, a solution based on an improved particle swarm algorithm is proposed for the path planning problem without a road network in forest fire rescue scenarios. The algorithm adopts an adaptive inertia weight and a dynamically updated learning factor strategy to enhance the global and local search capabilities of the algorithm. In terms of cost function design, the article considers three factors: path length, terrain slope, and obstacle avoidance ability to ensure the safety and effectiveness of the path. The experimental results show that: (1) the path planning algorithm based on improved particle swarm optimization can effectively avoid spreading wildfire and reach the designated target point with a good “detour” effect; (2) the path planned by the improved PSO algorithm performs better than the original PSO algorithm in terms of fitness evaluation and average slope; and (3) changes in the particle population, dimensions, and learning factors in the particle swarm optimization algorithm can affect the convergence of the final path. Increasing the particle dimensions can bring more reasonable and specific paths; decreasing the learning factor increases the convergence iterations, but also obtains a better path planning solution and higher fitness.

A New Optimization Control Policy for Fuzzy Vehicle Suspension Systems Under Membership Functions Online Learning

Active vehicle suspension systems play a crucial role in isolating vibration between the car body and pavement. In this article, based upon Takagi–Sugeno (T–S) fuzzy model, a novel optimization control algorithm is first proposed to furthest improve driving comfort level for fuzzy suspension systems compared with traditional existing approach. First, sufficient criteria for designing non parallel distribution compensation (non-PDC) fuzzy controller to improve driving comfort level and guarantee corresponding suspension constrained requirements are presented. Then, in the context of ensuring stability of studied systems, considering the feasible areas of membership functions (MFs) of non-PDC controller, a new MFs online optimization strategy is built for fuzzy suspension systems for the first time. By this proposed novel optimization algorithm, the values of controller MFs (CMFs) can be updated in real-time so as to obtain a better driving comfort level while improving suspension-constrained requirements. Sufficient criteria is obtained to guarantee the convergence of the designed cost function with the help of the Lyapunov stability theory. In the end, the usefulness, as well as superiority of the proposed online optimization strategy, are illustrated by contrastive verification.

Optimal [formula omitted] tracking control of nonlinear systems with zero-equilibrium-free via novel adaptive critic designs

In this paper, a novel adaptive critic control method is designed to solve an optimal H∞ tracking control problem for continuous nonlinear systems with nonzero equilibrium based on adaptive dynamic programming (ADP). To guarantee the finiteness of a cost function, traditional methods generally assume that the controlled system has a zero equilibrium point, which is not true in practical systems. In order to overcome such obstacle and realize H∞ optimal tracking control, this paper proposes a novel cost function design with respect to disturbance, tracking error and the derivative of tracking error. Based on the designed cost function, the H∞ control problem is formulated as two-player zero-sum differential games, and then a policy iteration (PI) algorithm is proposed to solve the corresponding Hamilton–Jacobi–Isaacs (HJI) equation. In order to obtain the online solution to the HJI equation, a single-critic neural network structure based on PI algorithm is established to learn the optimal control policy and the worst-case disturbance law. It is worth mentioning that the proposed adaptive critic control method can simplify the controller design process when the equilibrium of the systems is not zero. Finally, simulations are conducted to evaluate the tracking performance of the proposed control methods.

Economic optimization and predictive control for nonlinear systems using Lyapunov based nonlinear cost function design

Optimal control of physical processes depends on the formulation of objective functions and contemplation of appropriate predictive control. In this article, we propose a new framework, to blend the concepts of the Lyapunov-based economic model predictive control (Ly-EMPC) scheme with the novel cost-efficient profit function design. Furthermore, Lyapunov’s stability criteria is employed to verify the existence of duality and optimality for the designed economic stage costs. This method differs from the existing EMPC techniques by implementing Lagrange multipliers in the rotated cost function as a measure of stability. The closed-loop performance using the regularized Ly-EMPC is validated against a standard tracking objective-based nonlinear MPC for three benchmark examples through an econometrics gauge factor. Interestingly, in contrast to tracking NMPC, the proposed Ly-EMPC scheme excels in economic performance, while the formation of nonlinear economic stage cost is conceptually simpler, novel, and practically profitable.

Optimizing Representations of Multiple Simultaneous Attributes for Gait Generation Using Deep Learning.

Rich variations in gait are generated according to several attributes of the individual and environment, such as age, athleticism, terrain, speed, personal "style", mood, etc. The effects of these attributes can be hard to quantify explicitly, but relatively straightforward to sample. We seek to generate gait that expresses these attributes, creating synthetic gait samples that exemplify a custom mix of attributes. This is difficult to perform manually, and generally restricted to simple, human-interpretable and handcrafted rules. In this manuscript, we present neural network architectures to learn representations of hard to quantify attributes from data, and generate gait trajectories by composing multiple desirable attributes. We demonstrate this method for the two most commonly desired attribute classes: individual style and walking speed. We show that two methods, cost function design and latent space regularization, can be used individually or combined. We also show two uses of machine learning classifiers that recognize individuals and speeds. Firstly, they can be used as quantitative measures of success; if a synthetic gait fools a classifier, then it is considered to be a good example of that class. Secondly, we show that classifiers can be used in the latent space regularizations and cost functions to improve training beyond a typical squared-error cost.

Event-Triggered Robust Optimal Control for Robotic Manipulators with Input Constraints via Adaptive Dynamic Programming

This paper proposes an event-triggered robust tracking control method for robotic manipulators with asymmetrical input constraints based on adaptive dynamic programming (ADP). First, the original robust tracking control problem is transformed into an optimal control problem for a nominal system with the help of a novel cost function design. Next, an event-triggered optimal control approach is proposed to solve this optimization problem and obtain the optimal solution to the Hamilton-Jacobi-Bellman (HJB) equation. The classical ADP framework is then used to approximate the optimal cost function and controller. Unlike existing weight adaptive laws, an additional term is introduced for removing the requirement of initial stabilization. Based on the Lyapunov theory, the stability of the original robotic system and the convergence of weight estimation are both guaranteed. Finally, simulation results are presented to demonstrate the effectiveness of the proposed event-triggered optimal control method.

Reward (Mis)design for autonomous driving

This article considers the problem of diagnosing certain common errors in reward design. Its insights are also applicable to the design of cost functions and performance metrics more generally. To diagnose common errors, we develop 8 simple sanity checks for identifying flaws in reward functions. We survey research that is published in top-tier venues and focuses on reinforcement learning (RL) for autonomous driving (AD). Specifically, we closely examine the reported reward function in each publication and present these reward functions in a complete and standardized format in the appendix. Wherever we have sufficient information, we apply the 8 sanity checks to each surveyed reward function, revealing near-universal flaws in reward design for AD that might also exist pervasively across reward design for other tasks. Lastly, we explore promising directions that may aid the design of reward functions for AD in subsequent research, following a process of inquiry that can be adapted to other domains.

Improved Model Predictive Current Control With Series Structure for PMSM Drives

In order to improve the control performance of conventional finite-control-set model predictive control (FCS-MPC) method and keep the balance between the steady-state performance and switching frequency of the system, this article proposes an improved MPC method based on the conventional MPC method. First, the reference voltage vector of the next control instant is predicted to determine the candidate voltage vectors of the next two control instants. Then, an optimal vector selection way is proposed, in which candidate voltage vectors selected by the reference voltage vector and designed cost function structure in series. Moreover, the selection process of optimal voltage vector from the candidate voltage vectors is analyzed. Finally, the comparison experiments between the conventional FCS-MPC method and the proposed MPC method are conducted. And experimental results show the validity of the proposed control strategy.

Emergency steering collision avoidance control based on distributed driving intelligent vehicles

SummaryBased on the distributed drive intelligent electric vehicle, considering the collision avoidance effect of the planned path, the path tracking accuracy, and the body stability requirements in the process of steering collision avoidance, this article proposes a control strategy for emergency steering collision avoidance. First, the fifth‐order polynomial method is used to generate collision avoidance paths, and then a designed cost function is used to evaluate and select the optimal collision avoidance path. Second, a decision‐making strategy that integrates multiple risk judgment models is built. Based on the model prediction method, a path tracking controller is built, and an optimization objective function is established to improve the tracking accuracy. In order to ensure the stability of the vehicle body, a vehicle stability controller is designed based on fuzzy theory. Finally, a vehicle‐in‐the‐loop intelligent driving algorithm test platform is built, and the control algorithm is tested through this platform. The test results show that compared with the pure path tracking control method, the control strategy is reasonable and effective and can better reduce the path tracking error and improve the vehicle's driving stability in the collision avoidance process.

Game-based coordination control of multi-agent systems

In this paper, we are interested in how agents interact with each other in multi-agent systems. First, we model the interactions among agents in multi-agent systems as a multi-player game. The topology of the interactions among agents is a directed graph. We design cost functions for the game and assume that each agent in the systems tends to minimize its own cost. Then, the unique Nash equilibrium solution to the proposed multi-player game is obtained as the next state of the agent. A necessary and sufficient condition for achieving multi-agent consensus is established using the system transformation method and graph theory. Furthermore, we extend the result to the problem of containment control in the presence of leaders. The criterion for solving the containment control is also given in this paper. Finally, several simulation examples are given to verify the effectiveness of the theoretical results.

Design of Model Predictive Control Weighting Factors for PMSM Using Gaussian Distribution-Based Particle Swarm Optimization

An improved particle swarm optimization (PSO) algorithm based on Gaussian distribution model is proposed to realize the autotuning of weighting factors for the cost function design in the model predictive control method. First, the design principle of the weighting factors in model predictive torque control for permanent magnet synchronous motor system is analyzed. Then, using the root mean square of the current error in the two-phase rotating coordinate system and the system switching frequency as references, the objective function of the particles in the PSO is designed by considering the main control goals of reducing the torque ripple, the current total harmonic distortion, and the switching frequency. The Gaussian individual optimal distribution model is used to update the particle position on the structure of the conventional PSO algorithm. The experimental results show that the proposed method can solve the problem of weighting factors design as it reduces the switching frequency of the system while achieving excellent steady-state performance.

Infrared and visible image fusion based on contrast enhancement guided filter and infrared feature decomposition

Infrared (IR) and visible (VIS) images represent the features of the scene at different wavelengths, and the features they contain have different properties. Therefore, the traditional weighted fusion strategy is challenging to preserve the different types of feature information. In addition, VIS images are highly susceptible to bad weather, which also seriously affects the quality of fused images in complex environments. To solve the above problems, we propose a feature enhancement fusion method. First, a fusion model called contrast enhancement guided filter (CEGF) is proposed. The new model enables the texture information of VIS images to be presented with the intensity of infrared pixels, which solves the problem of combining different attribute features and removes the influence of harsh lighting conditions. To improve the visibility of texture details under different lighting conditions, a contrast modulation factor is added to the cost function design of the filter to enhance the contrast of visible details. Second, we use a dual-scale decomposition strategy to enhance the infrared feature information of the fusion results. Finally, we apply the method of this paper with ten classical image fusion algorithms in two types of datasets. The visual effect and objective evaluation of the fusion results verify that the proposed method preserves the characteristics of the high contrast of IR images and improves the visibility of infrared scenes for subsequent target identification and detection.

UAV formation control based on distributed Kalman model predictive control algorithm

To address the perturbation of formation of multiple unmanned aerial vehicles (UAVs) subject to external disturbances, an algorithm of distributed Kalman model predictive control is proposed in this paper to improve the accuracy of maintaining a formation in flight. A UAV two-order discrete-time system model was built before devising a Kalman prediction model based on the standard prediction model. The desired formation configuration and neighbor Kalman optimal state estimation were conducted to determine the reference state of UAVs. While taking into account the formation tracking error and input stability, a logarithmic barrier function was introduced in the design of the overall cost function to ensure flight safety. Meanwhile, information was exchanged with neighbors with the directed and time-invariant communication topological structure. With the Lyapunov stability theorem, sufficient conditions were defined for the asymptotic stability of the formation system. Simulation results revealed that the algorithm could effectively suppress the perturbation in the formation of UAVs arising from external disturbances, allowing the formation to cope with the conflicts between individual UAVs.

Rate-Distortion-Based Stego: A Large-Capacity Secure Steganography Scheme for Hiding Digital Images.

Steganography is one of the most crucial methods for information hiding, which embeds secret data on an ordinary file or a cover message for avoiding detection. We designed a novel rate-distortion-based large-capacity secure steganographic system, called rate-distortion-based Stego (RD-Stego), to effectively solve the above requirement. The considered effectiveness of our system design includes embedding capacity, adaptability to chosen cover attacks, and the stability of the trained model. The proposed stego scheme can hide multiple three-channel color images and QR codes within another three-channel color image with low visual distortion. Empirically, with a certain degree of robustness against the chosen cover attack, we state that the system offers up to 192+ bits-per-pixel (bpp) embedding of a payload and leaks no secret-related information. Moreover, to provide theoretical foundations for our cost function design, a mutual information-based explanation of the choices of regulation processes is herein included. Finally, we justify our system’s claimed advantages through a series of experiments with publicly available benchmark datasets.

Q-Soft: QoS-Aware Traffic Forwarding in Software-Defined Cyber–Physical Systems

The next-generation cyber–physical systems (CPSs) with heterogeneous applications have diverse Quality-of-Service (QoS) requirements in terms of throughput, end-to-end latency, and packet drop reliability. To meet such diverse QoS requirements, in this article, we propose a QoS-aware traffic forwarding scheme in software-defined CPS. The proposed scheme is presented as a two-stage optimization framework to minimize the associated costs in traffic forwarding. In the first stage, we aim to minimize the required number of “candidate” switches for a given network to minimize network deployment costs. In the second stage, we design a comprehensive cost function considering end-to-end delay, flow-rule utilization, and link utilization in the network. Based on the designed cost function, we formulate another optimization problem for optimal traffic forwarding (OTF). As solving OTF is NP-hard, we propose an efficient greedy-heuristic approach to solve the problem while considering application-specific QoS requirements. Further, we propose a packet-tagging method to assist the controller in mitigating rule congestion at the software-defined networking devices, and hence improve the overall network performance. Extensive results show that the proposed scheme minimizes the network delay and QoS-violated flows by up to 50% and 90%, respectively, compared to the state-of-the-art schemes.

Energy Function Based Finite Control Set Predictive Control Strategy for Single-Phase Split Source Inverters

This article presents energy-function based model predictive control (EF-MPC) for single-phase split source inverters. Unlike conventional MPC methods, the proposed EF-MPC is designed by considering the stability of inverter. According to the Lyapunov's direct method, the stability of inverter can be assured if the derivative of energy-function is always negative. Motivated from this fact, the design of cost function in EF-MPC is based on the energy function which involves dc- and ac-side inductor current errors. It is shown that the coefficient used in the formulation of energy function is redundant which eliminates the weighting factor required in the multiobjective cost function. Elimination of the weighting factor not only simplifies the controller design in the practical implementation, but also removes the weighting factor tuning process needed in the conventional predictive methods. Another advantage of the proposed EF-MPC is that the amplitude of low-frequency component in the dc-side inductor current is suppressed. The steady-state and dynamic performances of the proposed EF-MPC are investigated experimentally under linear and nonlinear loads. In both cases, the dc-side variables are regulated at the desired values while the sinusoidal output voltage with low total harmonic distortion is obtained in the ac-side.

Data-driven adaptive tuning of iterative learning control

In this paper, we propose two data-driven adaptive tuning (DDAT) approaches of iterative learning control (ILC) for nonlinear non-affine systems. First, a compact-form iterative dynamic linearization (CFIDL) method is introduced to transfer the original nonlinear system into a linear data model. Then, we design an objective function for the tuning of the learning gains of a PD-type ILC law. By optimizing the designed cost function subjected to the linear data model, a CFIDL-based DDAT method is proposed, where only the real I/O data are used without requiring any mechanistic model information. Furthermore, the results are extended by introducing a partial-form iterative dynamic linearization (PFIDL) method for the purpose of utilizing more additional control information. Following the similar steps, a PFIDL-based DDAT method is developed for learning gain tuning of the PD-type ILC scheme. Both the proposed DDAT methods can help the PD-type ILC have a better robustness against to the uncertainties since they can use the real I/O data to iteratively tune the learning gains. The convergence of the DDAT-based PD-type ILC methods has been proved rigorously. The effectiveness of the two proposed DDAT-based ILC methods are further verified through simulations.

Generalized link cost function and network design for dedicated truck platoon lanes to improve energy, pavement sustainability and traffic efficiency

Recent development on autonomous and connected trucks (ACT) has provided the freight industry with the new option of using truck platooning to improve their fuel efficiency, traffic throughput, and safety. However, closely spaced and longitudinally aligned trucks impose frequent and concentrated loading on pavements, which often accelerates pavement deterioration and increase the life-cycle costs for the highway agency. Also, effectiveness of truck platooning can only be maximized in dedicated lanes, and its benefits and costs need to be properly balanced between stakeholders. This paper proposes a network design model to optimize (i) placement of dedicated truck platoon lanes and toll price in a highway network, (ii) pooling and routing of ACT traffic from multiple origins and destinations to utilize these lanes, and (iii) configuration of truck platoons within these lanes (e.g., lateral displacements and vehicle separations). The problem is formulated as an integrated bi-level optimization model. The upper level makes decisions on converting existing highway lanes into dedicated platoon lanes, as well as setting user fees. The lower level decisions are made by independent shippers regarding the choice of routes, use of platoon lanes vs. regular lanes, and they collectively determine truck traffic in all lanes. Link cost functions for platoon lanes are obtained by simultaneously optimizing, through dynamic programming, pavement rehabilitation activities and platoon configuration in the pavement’s life cycle. A numerical case study is used to demonstrate the applicability and performance of the proposed model framework over the Illinois freeway system. It is shown that the freight traffic is effectively channelized on a few corridors of platoon lanes and, by setting proper user fees to cover pavement rehabilitation costs, system-wide improvements for both freight shippers and highway agencies can be achieved.