Individual Objective Functions Research Articles

Well Trajectory Optimization under Geological Uncertainties Assisted by a New Deep Learning Technique

Summary The large number of geological realizations and well trajectory parameters make field development optimization under geological uncertainty a time-consuming task. A novel deep learning-based surrogate model with a novel well trajectory parametrization technique is proposed in this study to optimize the trajectory of wells under geological uncertainty. The proposed model is a deep neural network with ConvLSTM layers to extract the most salient features from highly channelized and layered reservoirs efficiently. ConvLSTM layers are used because they can extract spatiotemporal features simultaneously since layered reservoirs can be regarded as a time series of spatially distributed reservoir properties. The proposed surrogate model could predict the individual objective function with a coefficient of determination of 0.96. After verifying the validity of the surrogate model, four approaches were used to optimize well trajectories. Two of the approaches consumed all available realizations (surrogate model-based and simulation-based approaches), while the remaining two used a subset of realizations. The selection of the subset was based on the cumulative oil production (COP) and the diffusive time of flight (DTOF). Results showed that although the surrogate model used all realizations, it could provide similar results to the simulation-based optimization with only a 5% computational cost of the simulation-based approach. The novelty of this work lies in its proposal of an innovative surrogate model to improve the analysis of channelized and layered reservoirs and its introduction of a novel well trajectory optimization framework that effectively addresses the challenge of optimizing well trajectories in complex three-dimensional spaces, a problem not adequately tackled in previous works.

Distributed time-varying optimization with coupled constraints: Application in UAV swarm predefined-time cooperative consensus

This paper addresses the global optimal consensus problem with time-varying objective functions and state-coupled inequality constraints. By integrating the predefined-time consensus and prediction-correction schemes, a distributed optimization protocol is proposed for the cooperative interception scenarios involving the unmanned aerial vehicle (UAV) swarm. Specifically, the δ-exact penalty method is utilized to eliminate the coupled constraints. A prediction-correction term with the penalty function is then established to track the trajectory of the time-varying optimal solution. Additionally, an edge-based predefined-time consensus term is introduced to facilitate the state consensus among UAVs. Under the undirected and connected topology, the proposed protocol ensures the predefined-time state consensus and asymptotically minimizes the sum of individual convex objective functions. Furthermore, the protocol is extended to other scenarios involving switching topology, directed detail-balanced topology, and external disturbances. Numerical simulations illustrate that the UAV swarm can form the predesigned interception formation with a predefined-time frame and effectively track the target swarm.

QoS driven cost-efficient resource allocation in edge computing: A distributed game theoretic approach

In edge computing, app vendors provide task-offloading services to small intelligent devices with limited resources via edge servers. Users of these devices aim to utilize a better quality of service at a lower cost, and the app vendor tries to maximize their user base. In this paper, we propose an edge computing resource allocation technique to address the critical issue of maintaining a trade-off between user cost and QoS while increasing the number of users served by the app vendor. In the proposed technique, we formulate the problem as an Edge Server Selection Game (ESSGame), where users can optimize individual objective functions. An Edge Server Selection Algorithm (ESSA) is presented by which the system quickly converges to a pure Nash equilibrium (PNE). It is proved that ESSAGame is a potential game with at least one PNE. The quality of the PNE in ESSGame is also determined in terms of the Price of Stability (PoS). Numerical results are also presented for the performance of the proposed technique.

PINN training using biobjective optimization: The trade-off between data loss and residual loss

Physics informed neural networks (PINNs) have proven to be an efficient tool to represent problems for which measured data are available and for which the dynamics in the data are expected to follow some physical laws. In this paper, we suggest a multiobjective perspective on the training of PINNs by treating the data loss and the residual loss as two individual objective functions in a truly biobjective optimization approach.As a showcase example, we consider COVID-19 predictions in Germany and built an extended susceptibles-infected-recovered (SIR) model with additionally considered leaky-vaccinated and hospitalized populations (SVIHR model) to model the transition rates and to predict future infections. SIR-type models are expressed by systems of ordinary differential equations (ODEs). We investigate the suitability of the generated PINN for COVID-19 predictions and compare the resulting predicted curves with those obtained by applying the method of non-standard finite differences to the system of ODEs and initial data.The approach is applicable to various systems of ODEs that define dynamical regimes. Those regimes do not need to be SIR-type models, and the corresponding underlying data sets do not have to be associated with COVID-19.

Optimal fog node selection based on hybrid particle swarm optimization and firefly algorithm in dynamic fog computing services

Fog computing is a qualifiedly roseate technology introduced to support latency-sensitive and mission-critical applications by bringing resources closer to the end users. To exploit the full potentials of this auspicious technology, it is important to select optimal fog nodes for secured service provisioning. Therefore, in this work, an optimal fog node selection is formulated as a multi-objective optimization problem. To solve this problem, an efficient selection strategy based on improved particle swarm optimization (PSO) and modified firefly algorithm (FA) is developed. This study considers two important objectives that have not been optimized in the previous works: trust and rate of sojourn, in addition to remaining node capacity and energy consumption. The linear weighted-sum approach is used to aggregate the individual objective functions, after which the Best–worst method (BWM) is used to determine the weight vector of the aggregated function. The impact of the variation in the weight values of the most sensitive objective indicate that in the worst-case scenario, the proposed model is robust and insensitive with an improvement of about 42.16%–50.02% over the related methods. Moreover, we conduct an experimental simulation of the proposed selection strategy as well as comparative analysis with the state-of-the-art algorithms using six performance metrics. The results show that the proposed hybrid PSO-FA achieves a higher accuracy and faster convergence. The proposed solution also records an improvement of about 18.64%–69.45% in resource utilization, 10.94%–40.45% in energy consumption, 12.50%–75% in trust violation, 4.94%–31.12% in computation delay, and 24.71%–37.7% in makespan over the related advanced algorithms.

Religiosity and Charitable Giving on Investors’ Trading Behaviour in the Indonesian Islamic Stock Market: Islamic vs Market Logic

This study examines retail investors’ trading behaviour and its determinants in the Indonesian Shari’ah stock market by mainly focusing on the religious practice-related factors in the form of sadaqah or charitable giving on individual investors’ trading behaviour. Contextually, the Islamic moral economy (IME) assumes a direct relationship between religiosity and sadaqah giving due to the falah (salvation) oriented individual objective function, which can be reached through doing ihsan (beneficence for equilibrium). The findings based on a questionnaire survey distributed to individual investors on Shari’ah Online Trading System (SOTS) delineate that religiosity, accounting information, neutral information, personal financial needs, and the sadaqah feature have significantly affected investors’ trading behaviour in which the sadaqah feature is positively correlated, while religiosity factors are negatively correlated. Thus, despite the theoretical expectation through IME, this study evidences that Islamic logic is not the main determining factor, as market logic related factors seem to be more dominant in the behaviour of investors in the Indonesian capital market.

Personalized control system via reinforcement learning: maximizing utility based on user ratings

In this paper, we address the design of personalized control systems, which pursue individual objectives defined for each user. To this end, a problem of reinforcement learning is formulated where an individual objective function is estimated based on the user rating on his/her current control system and its corresponding optimal controller is updated. The novelty of the problem setting is in the modelling of the user rating. The rating is modelled by a quantization of the user utility gained from his/her control system, defined by the value of the objective function at his/her control experience. We propose an algorithm of the estimation to update the control law. Through a numerical experiment, we find out that the proposed algorithm realizes the personalized control system.

Personalization of Control Systems by Policy Update With Improving User Trust

In this letter, we address the design of personalized control systems, which pursue an individual and private objective defined for each user. To this end, a problem of policy update is formulated where an individual objective function is estimated and the corresponding optimal control law is updated. The novelty of the problem setting is in the presence of a system-user and the policy update driven by his/her rating. The system-user rates on the control system to be updated and the rating is used for estimating his/her objective function. It is assumed that the rating depends not only on his/her private objective but also on his/her trust on the control system. Then, we address the problem of the policy update improving the rating while not impairing the trust. Algorithms of the policy update, which is essentially the objective function estimation, are developed and their convergence analysis is presented. Finally, through a numerical experiment, the effectiveness of the algorithms is shown.

A Totally Asynchronous Algorithm for Time-Varying Convex Optimization Problems

This paper presents a decentralized algorithm for a team of agents to track time-varying fixed points that are the solutions to time-varying convex optimization problems. The algorithm is first-order, and it allows for total asynchrony, i.e., all communications and computations can occur with arbitrary timing and arbitrary (finite) delays. Convergence rates are computed in terms of the communications and computations that agents execute, without specifying when they must occur. These rates apply to convergence to the minimum of each individual objective function, as well as agents’ long-run behavior as their objective functions change. Then, to improve the usage of limited communication and computation resources, we optimize the timing of agents’ operations relative to changes in their objective functions to minimize fixed point tracking error over time. Simulation results illustrate these developments.

Solving Specified-Time Distributed Optimization Problem via Sampled-Data-Based Algorithm

Despite significant advances on distributed continuous-time optimization of multi-agent networks, there is still lack of an efficient algorithm to achieve the goal of distributed optimization at a pre-specified time, especially for the case with unbalanced directed topologies. Herein, a new out-degree based design structure is proposed for connected agents with directed topologies to collectively minimize the sum of individual objective functions and keep satisfying an equality constraint. With the designed algorithm, the settling time of distributed optimization can be exactly predefined. The specified selection of such a settling time is independent of not only the initial conditions of agents, but also the algorithm parameters and the communication topologies. Furthermore, the proposed algorithm can realize specified-time optimization by exchanging information among neighbors only at discrete sampling instants and thus reduces the communication burden. In addition, the equality constraint is always satisfied during the whole process, which makes the proposed algorithm applicable to online solving distributed optimization problems such as energy resource allocation. For the special case of undirected communication topologies, a reduced-order algorithm is also designed. Finally, the effectiveness of the theoretical analysis is justified by numerical simulations.

Hierarchical design of material microstructures with thermal insulation properties

Composite materials with multiple properties are important for a range of engineering applications. Hence, this study focuses on topological design of hierarchical materials with multiple performance in both thermal insulation and mechanics. First, a novel multi-objective optimization function is defined to find a solution from the Pareto frontier, where the weight coefficients can be adjusted adaptively, to keep all the individual objective functions and their sensitivities stabilized at the same level during the optimization. Second, a new design strategy is proposed to achieve the hierarchical designs of biphasic material microstructures, they are periodically arranged by the porous base materials that are known in advance and independent of topology optimization. Third, sensitivity information and algorithm implementation are given in detail, and the bi-directional evolutionary structural optimization method is adopted to iteratively update the micro-structural topologies, by combining with the homogenization method. Last, numerical examples are provided to illustrate the benefits of the proposed design method, such as high efficiency, implementation easiness, good connectivity and clear interface between adjacent phases, etc.

Efficient and sparse neural networks by pruning weights in a multiobjective learning approach

Overparameterization and overfitting are common concerns when designing and training deep neural networks, that are often counteracted by pruning and regularization strategies. However, these strategies remain secondary to most learning approaches and suffer from time and computational intensive procedures. We suggest a multiobjective perspective on the training of neural networks by treating its prediction accuracy and the network complexity as two individual objective functions in a biobjective optimization problem. As a showcase example, we use the cross entropy as a measure of the prediction accuracy while adopting an l1-penalty function to assess the total cost (or complexity) of the network parameters. The latter is combined with an intra-training pruning approach that reinforces complexity reduction and requires only marginal extra computational cost. From the perspective of multiobjective optimization, this is a truly large-scale optimization problem. We compare two different optimization paradigms: On the one hand, we adopt a scalarization-based approach that transforms the biobjective problem into a series of weighted-sum scalarizations. On the other hand we implement stochastic multi-gradient descent algorithms that generate a single Pareto optimal solution without requiring or using preference information. In the first case, favorable knee solutions are identified by repeated training runs with adaptively selected scalarization parameters. Numerical results on exemplary convolutional neural networks confirm that large reductions in the complexity of neural networks with negligible loss of accuracy are possible.

Analytical Investigation and Heuristic Optimization of Surface Mounted Permanent Magnet Machines With Hybrid Magnetic Structure

Cogging torque causes major operational setbacks for Permanent Magnet (PM) machine operation, particularly in applications where a quiet performance is desired. This paper presents a heuristic optimization framework to optimize the cogging torque in Surface Mounted Permanent Magnet (SPM) machines consisting of a hybrid magnetic structure (i.e., rare-earth and ferrite magnets). To avoid excessive computational time and volume associated with Finite Element (FE)-based optimization solutions, analytical approach is paired up with the optimization algorithm to determine the optimal design while FE is utilized for verification and validation purposes. First, analytical expressions are established for individual objective functions (i.e., airgap PM flux distribution, and cogging torque), and their corresponding spatial harmonics are identified using the air-gap field modulation theory. Next, the presented analytical model is utilized to optimize the system (i.e., minimize the cogging torque) to the desired target level via two different solutions (i.e., Genetic Algorithm (GA) and Particle Swarm Optimization (PSO)), and their respective performance are compared. To determine the efficacy of the presented solutions, the optimal hybrid machine response is compared against the baseline structure.

Loss-Voltage Sensitivity Analysis Based Battery Energy Storage Systems Allocation and Distributed Generation Capacity Upgrade

Distributed Generation (DG) has become a key component in modern power industry due its significant advantages over the traditional power generation methods. Nevertheless, best expected outcomes can only be achieved with optimal allocation of DG resources where inappropriate allocation may impose problems in power system stability, protection and quality. This paper presents analytical approaches for optimizing the DG size, BESS (Battery Energy Storage Systems) capacities and power dispatch in Medium Voltage (MV) networks. Since most of the existing analytical approaches related to optimizing DG sizes for minimizing network losses and voltage deviations have considered individual objective function separately, both parameters may not be minimized simultaneously. Thus, in this paper, an analytical methodology was formulated based on an objective function built on new parameter Loss-Voltage Sensitivity Index (LVSI) that evaluates both minimum impact of network loss and voltage variations for optimizing the DG size. The results obtained from this approach were compared with a conventional Genetic Algorithm (GA) formulated by the authors. The BESS capacities are determined considering the effects of Load Proportionality Factor (LPF), State of Charge limits (SOC) of battery storages, number of load areas and the portion of daily off-peak solar generation period energy consumption expected to be served by each BESS unit. The significance of this BESS capacity determination methodology is highlighted as the BESS capacities can be numerically calculated whereas in existing work, they are heavily relying on conventional optimization techniques which do not give an idea about the internal behavior of parameters which determine the capacities of BESS units. Moreover, an optimal BESS dispatch algorithm is also presented in this paper for minimizing the energy losses and voltage deviations. The applicability of the proposed methodologies are verified using the standard IEEE-33 and IEEE-69 test bus systems. Simulations carried out in MATLAB is used to illustrate the accuracy and the appropriateness of the proposed approaches.

Multidimensional Aspects of Sustainable Biofuel Feedstock Production

Bioenergy is becoming increasingly relevant as an alternative to fossil fuels. Various bioenergy feedstocks are suggested as environmentally friendly solutions due to their positive impact on stream health and ability to sequester carbon, but most evaluations for bioenergy feedstocks have not evaluated the implications of bioenergy crop production holistically to date. Through the application of multi-objective optimization on 10 bioenergy feedstock rotations in a Michigan watershed, a Pareto front is searched to identify optimal trade-off solutions for three objective functions representing stream health, environmental emissions/carbon footprint, and economic feasibility. Various multi-criteria decision-making techniques are then applied to the resulting Pareto front to select a set of most-preferred trade-off solutions, which are compared to optimal solutions from each individual objective function. The most-preferred trade-off solutions indicate that a diverse mix of rotations are necessary to optimize all three objectives, whereas the individually optimal solutions do not consider a diverse range of feedstocks, thereby making the proposed multi-objective treatment an important and pragmatic strategy.

Suboptimal output consensus for a group of weakly nonminimum phase linear systems

This paper studies the suboptimal output consensus problem for a multi-agent system, where each agent is described by a weakly nonminimum phase linear system and endowed with an objective function. For each agent, a low gain feedback based protocol consisting of a local term and a distributed term and parameterized with a low gain parameter is constructed. Under the assumption that the communication topology is strongly connected and weight-balanced, these protocols achieve suboptimal output consensus for the multi-agent system, that is, for a small enough value of the low gain parameter, the outputs of all agents reaching consensus at a value in an a priori specified arbitrarily small neighborhood of the optimal value of the overall objective function, which is the sum of the individual objective functions of all agents. The theoretical results are verified by numerical examples.

Multi-objective optimization of a biodiesel production process using process simulation

Economic and environmental aspects have been contributing to the gradual reduction of fossil fuels and to the growth of research on renewable energy, such as biodiesel, that has been gaining representation in the national and world market. In this scenario, the optimal operation of continuous biodiesel processes may be an attractive tool to optimize some economic/environmental indicators, e.g., biodiesel production and purity, residues, and energy consumption. To develop a multicriteria optimization, a multi-objective methodology must be performed in which two or more functions are used as performance measurements. In this sense, the present work aims to propose the multi-objective optimization (MOO) of a continuous biodiesel plant. The purpose of the optimization was to maximize the biodiesel flow rate and minimize the energy rate consumption subject to a minimum biodiesel purity. An additional study was performed to quantify a set of potential environmental impacts (PEIs) of the optimized scenarios. The results of MOO showed that there is a set of optimal solutions depending on the weights assigned to each individual objective function. It was verified that the reactor temperature has a significant influence on biodiesel production, but less influence on the energy rate consumption. The suggested methodology could serve as a basis for future work addressing multi-criteria optimization, where the optimal operating point solution depends on the decision-maker. These results suggest that the PEIs can serve as environmental constraints for the optimization problem, allowing the decision maker to evaluate these criteria for choosing the optimum operating point.

Joint modeling of Rayleigh wave dispersion and H/V spectral ratio using Pareto-based multiobjective particle swarm optimization

Estimating the shear wave velocity and thickness of geologic units in a sedimentary structure is important for quantifying the local site effect caused by an earthquake. Inversion of the Rayleigh wave dispersion alone is sensitive to the absolute average shear wave velocity, while the H/V spectral ratio is sensitive to velocity contrasts. The solution of these models in a joint system using conventional inversion techniques suffers from difficulties while evaluating partial derivatives, dependencies to the initial model that is sometimes difficult to estimate, and trapping at a local minimum. Herein, a joint model using the Pareto optimality technique with multiobjective particle swarm optimization was constructed as an effective global optimization method to decrease weakness and increase performance while estimating the depths and velocities. The presented approach is a multiobjective optimization method that does not require weighting, and allows evaluations of the individual objective function separately in a joint system. This study presents 2 synthetic joint system examples to account for a gradient-type and sharp velocity contrast cases, and an application from field data obtained from the Bursa Basin in Turkey. With these examples, an automated parameter search space was demonstrated to account for the abrupt velocity changes and a number of additional key points that are essential for a reasonable optimal solution.

Самообучение ранжированию предприятия железнодорожного транспорта по энергоэффективности

Adaptive methods of energy efficiency control are looked upon under ambiguity conditions, based on self-instruction and human factor. The objective of dichotomous ranking of energy efficiency stochastic potential was set and solved in case when its probabilistic characteristics are known. In a reverse situation, it is suggested to use self-instruction for determination of the most suitable parameter of ranking procedure. A case of asymmetrical awareness of a decision-making person about ranking and a personnel responsible for energy efficiency was studied. Herewith, the forward-looking personnel chooses its indicators in such a way that the individual objective function which depends on ranking results , can be maximized. Such activity of the personnel may bring to under-estimation of the energy efficiency indicator as compared to the potential. As a consequence, application of self-instruction may pervert ranking parameter assessment. In order to solve this problem there were suggested self-instruction mechanisms in ranking which simulate personnel management under conditions of ambiguity. The results of these mechanisms functioning are estimates of ranking parameters, norms and ranks, determining personnel stimuli. There were found out appropriate conditions of mechanism synthesis of self-instruction in ranking, when a decision-making person identifies the most suitable parameter of the ranking procedure. These conditions are illustrated with the example of energy efficiency ranking self-instruction of the railway enterprise in the process of realization of «RZD» holding energy efficiency improvement program.

Stochastic parallel extreme artificial hydrocarbon networks: An implementation for fast and robust supervised machine learning in high-dimensional data

Artificial hydrocarbon networks (AHN) – a supervised learning method inspired on organic chemical structures and mechanisms – have shown improvements in predictive power and interpretability in comparison with other well-known machine learning models. However, AHN are very time-consuming that are not able to deal with large data until now. In this paper, we introduce the stochastic parallel extreme artificial hydrocarbon networks (SPE-AHN), an algorithm for fast and robust training of supervised AHN models in high-dimensional data. This training method comprises a population-based meta-heuristic optimization with defined individual encoding and objective function related to the AHN-model, an implementation in parallel-computing, and a stochastic learning approach for consuming large data. We conducted three experiments with synthetic and real data sets to validate the training execution time and performance of the proposed algorithm. Experimental results demonstrated that the proposed SPE-AHN outperforms the original-AHN method, increasing the speed of training more than 10,000x times in the worst case scenario. Additionally, we present two case studies in real data sets for solar-panel deployment prediction (regression problem), and human falls and daily activities classification in healthcare monitoring systems (classification problem). These case studies showed that SPE-AHN improves the state-of-the-art machine learning models in both engineering problems. We anticipate our new training algorithm to be useful in many applications of AHN like robotics, finance, medical engineering, aerospace, and others, in which large amounts of data (e.g. big data) is essential.