Optimization Time Research Articles

Dynamic Machine Learning Global Optimization Algorithm and Its Application to Aerodynamics

A dynamic extreme gradient boosting (XGBoost) and MaxLIPO trust region parallel global optimization algorithm is proposed in this paper, and it is applied to the turbomachinery blade aerodynamic optimization coupled with an in-house graphics processing unit (GPU) heterogeneous accelerated compressible flow solver, AeroWhale. The algorithm combines an accurate machine learning regression model, an efficient nongradient optimization method with no hyperparameters, a dynamic update regression strategy, and double convergence criteria to achieve high optimization accuracy and efficiency. The optimization results on the test function indicate that the number of objective function calls is less than 2% of that required by a traditional genetic algorithm, which greatly reduces the optimization time. The dynamic XGBoost model ensures that the regression model accuracy near the optimum is relatively high, which is attributed to the update strategy. The error between the optimal value identified by the proposed algorithm and the theoretical value is only 0.52% after several objective function calls. Finally, the aerodynamic optimization algorithm is applied to the LS89 high-pressure turbine, and the total pressure loss is reduced by 13.16%. The sensitivity of each optimization feature to the objective function is determined, showing that the blade suction surface control point near the trailing edge has the greatest impact on aerodynamic performance.

The Joint Decision-Making Support through Piecewise Objective Optimization Model for Integrated Supplier Selection, Inventory Management, and Production Planning Involving Discounts

The decision-makers in manufacturing industries continuously optimize every supply-chain part to achieve optimal profit. In this paper, three crucial activities in the supply chain are observed as profit contributors: supplier selection, inventory management, and production planning. Decision-making support is needed to optimize those activities, especially when prices/costs involve discounts. Therefore, this study aims to develop integrated decision-making support for supplier selection, inventory management, and production planning involving discounted prices. The problem was considered with multi-supplier, multi-raw material, multi-product, and multi-observation time instant. The objective was based on maximizing the profit for the entire activity, i.e., from the raw material procurement and storage to the production. This supply chain was modeled as mixed-integer linear programming with a piecewise objective function representing the profit, which was maximized. It was also modeled with a bunch of constraint functions, including product demand satisfaction. The proposed model was tested with computational simulations using randomly generated supply chain data. The primal simplex algorithm was also employed to calculate the real value of the optimal decision, which was combined with the Branch-and-Bound approach to calculate the appropriate integer solution. The results showed that the optimal decision was achieved, namely (1) The optimal quantity of raw materials ordered to each supplier, (2) The optimal production quantity, and (3) The optimal inventory level, which provided the maximal profit for the whole optimization time horizon. This indicated that the proposed decision-making support model is implementable for industrial decision-makers.

Do additional target points speed up evolutionary algorithms?

Most runtime analyses of randomised search heuristics focus on the expected number of function evaluations to find a unique global optimum. We ask a fundamental question: if additional search points are declared optimal, or declared as desirable target points, do these additional optima speed up evolutionary algorithms? More formally, we analyse the expected hitting time of a target set OPT∪S where S is a set of non-optimal search points and OPT is the set of optima and compare it to the expected hitting time of OPT.We show that the answer to our question depends on the number and placement of search points in S. For all black-box algorithms and all fitness functions with polynomial expected optimisation times we show that, if additional optima are placed randomly, even an exponential number of optima has a negligible effect on the expected optimisation time. Considering Hamming balls around all global optima gives an easier target for some algorithms and functions and can shift the phase transition with respect to offspring population sizes in the (1,λ) EA on OneMax. However, for the one-dimensional Ising model the time to reach Hamming balls of radius (1/2−ε)n around optima does not reduce the asymptotic expected optimisation time in the worst case. Finally, on functions where search trajectories typically join in a single search point, turning one search point into an optimum drastically reduces the expected optimisation time.

Ensemble-Learning-Based Multiobjective Optimization for Antenna Design

An ensemble-learning-based multiobjective optimization is proposed for antenna design. By integrating the local search into multiobjective evolutionary algorithm based on decomposition (MOEA/D) and selecting appropriate solutions of the local search acquired with different offspring reproduction (OR) operators, the MOEA/D combined with ensemble OR (MOEA/D-EOR) is presented. Parallel local OR operators based on samples are also merged, for the first time, by exhaustively mining the evolution data of the optimization searching. The diversity and convergence of MOEA/D-EOR are verified by several widely used benchmark problems. The efficiency of MOEA/D-EOR is demonstrated by designing a high-performance bow-tie multiple-input and multiple-output (MIMO) antenna, which saves at least 25% of the optimization time. The overall performance of MOEA/D-EOR is further demonstrated by designing a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2\times $ </tex-math></inline-formula> 2 MIMO patch antenna in a compact size of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.667\lambda _{0} \times 0.667\lambda _{0}$ </tex-math></inline-formula> , which achieves a high isolation of 24 dB in an operation band of 4.95–5.07 GHz.

Development of Cloud Autonomous System for Enhancing the Performance of Robots’ Path

With the development of computer technology and artificial intelligence (AI), service robots are widely used in our daily life. At the same time, the manufacturing cost of the robots is too expensive for almost all small companies. The greatest technical limitations are the design of the service robot and the resource sharing of the robot groups. Path planning for robots is one of the issues playing an important role in every application of service robots. Path optimization, fast computation, and minimum computation time are required in all applications. This paper aims to propose the Google Cloud Computing Platform and Amazon Web Service (AWS) platforms for robot path planning. The aim is to identify the effect and impact of using a cloud computing platform for service robots. The cloud approach shifts the computation load from robots to the cloud server. Three different path-planning algorithms were considered to find the path for robots using the Google Cloud Computing Platform, while with AWS, three different types of environments, namely dense, moderate, and sparse, were selected to run the path-planning algorithms for robots. The paper presents the comparison and analysis of the results carried out for robot path planning using cloud services with that of the traditional approach. The proposed approach of using a cloud platform performs better in this case. The time factor is crucially diagnosed and presented in the paper. The major advantage derived from this experiment is that as the size of the environment increases, the respective relative delay decreases. This proves that increasing the scale of work can be beneficial by using cloud platforms. The result obtained using the proposed methodology proves that using cloud platforms improves the efficiency of path planning. The result reveals that using the cloud computing platform for service robots can change the entire perspective of using service robots in the future. The main advantage is that with the increase in the scale of services, the system remains stable, while the traditional system starts deteriorating in terms of performance.

Focused jump-and-repair constraint handling for fixed-parameter tractable graph problems closed under induced subgraphs

Repair operators are often used for constraint handling in constrained combinatorial optimization. We investigate the (1+1) EA equipped with a tailored jump-and-repair operation that can be used to probabilistically repair infeasible offspring in graph problems. Instead of evolving candidate solutions to the entire graph, we expand the genotype to allow the (1+1) EA to develop in parallel a feasible solution together with a growing subset of the instance (an induced subgraph). With this approach, we prove that the EA is able to probabilistically simulate an iterative compression process used in classical fixed-parameter algorithmics to obtain a randomized FPT performance guarantee on three NP-hard graph problems. For k-VertexCover, we prove that the (1+1) EA using focused jump-and-repair can find a k-vertex cover (if one exists) in O(2kn2log⁡n) iterations in expectation. This leads to an exponential (in k) improvement over the best-known parameterized bound for evolutionary algorithms on VertexCover. For the k-FeedbackVertexSet problem in tournaments, we prove that the EA finds a feasible feedback set in O(2kk!n2log⁡n) iterations in expectation, and for OddCycleTransversal, we prove the optimization time for the EA is O(3kkmn2log⁡n). For the latter two problems, this constitutes the first parameterized result for any evolutionary algorithm. We discuss how to generalize the framework to other parameterized graph problems closed under induced subgraphs and report experimental results that illustrate the behavior of the algorithm on a concrete instance class.

Optimizing Hand Gesture Recognition Using CNN Model Supported by Raspberry pi for Self-Service Technology

This study describes the optimization of hand gesture recognition on Raspberry Pi 4 technology has advanced over the past years, some computers are now able to compute much more complex problems like real-time object detection. But for small devices, optimization is required to run in real-time with acceptable performance in terms of latency and low-cost effect on accuracy. Low latency is a requirement for most technology, especially when integrating real-time object detection as input into Self-Service Technology on Raspberry Pi for the store. This research was conducted on 288 pictures with six types of chosen hand gestures for command inputs that have been configured in the Self-Service Technology as a training dataset. In the experiment carried out with 5 CNN object detection models were used, namely YOLOv3-Tiny-PRN, YOLOv4-Tiny, MobileNetV2-Yolov3-NANO, YOLO-Fastest-1.1, and YOLO-Fastest-1.1-XL. Based on the experiment after optimization, the FPS and inference time metrics have improved performance. The performance improves due to a gained average value of FPS by 3 FPS and a reduced average value of inference time by 119,260 ms. But such an improvement also comes with a reduction in overall accuracy. The rest of the parameters have a reduced score on Precision, Recall, F1-Score, and some for IoU. Only YOLO-Fastest-1.1-XL have an improved value of IoU by about 0.58%. Some improvements in the CNN and dataset might improve the performance even more without sacrificing too much on the accuracy, but it's most likely suitable for another research as a continuation of this topic.

Bayesian Optimization Framework for HVAC System Control

The use of machine-learning algorithms in optimizing the energy efficiency of HVAC systems has been widely studied in recent years. Previous research has focused mainly on data-driven model predictive controls and reinforcement learning. Both approaches require a large amount of online interactive data; therefore, they are not efficient and stable enough for large-scale practical applications. In this paper, a Bayesian optimization framework for HVAC control has been proposed to achieve near-optimal control performance while also maintaining high efficiency and stability, which would allow it to be implemented in a large number of projects to obtain large-scale benefits. The proposed framework includes the following: (1) a method for modeling HVAC control problems as contexture Bayesian optimization problems and a technology for automatically constructing Bayesian optimization samples, which are based on time series raw trending data; (2) a Gaussian process regression surrogate model for the objective function of optimization; (3) a Bayesian optimization control loop, optimized for the characteristics of HVAC system controls, including an additional exploration trick based on noise estimation and a mechanism to ensure constraint satisfaction. The performance of the proposed framework was evaluated by using a simulation system, which was calibrated by using trending data from a real data center. The results of our study showed that the proposed approach achieved more than a 10% increase in energy-efficiency savings within a few weeks of optimization time compared with the original building automation control.

AROPS: A Framework of Automated Reaction Optimization with Parallelized Scheduling.

With the development of automated experimental platforms and optimization algorithms, chemists can easily optimize chemical reactions in an automated and high-throughput fashion. However, the modules in existing automated experimental platforms are operated in a linear fashion without orchestrating with the optimization algorithm, thus leaving room for further efficiency improvement. Here, we introduced a framework of automated reaction optimization with parallelized scheduling (AROPS) to realize the integration of the optimization algorithm and module scheduling. AROPS relies on a customized Bayesian optimizer to solve multi-reactor/analyzer reaction optimization problems with three different scheduling modes to arrange tasks for various experimental modules. In addition, a mechanism based on probability of improvement (PI) for discarding unpromising ongoing experiments was developed to facilitate freeing up valuable experimental resources in parallelized optimization. We tested the performance of AROPS using a hardware emulator on three representative benchmark reactions encountered in organic synthesis, illustrating that AROPS can trade off optimization time and cost according to the chemists' preference.

Finding an energy efficient path for plug-in electric vehicles with speed optimization and travel time restrictions

Transportation is one of the main factors when global total energy consumption is considered and is a significant contributor to emissions of harmful gases including carbon dioxide (CO2). Due to their lower tailpipe CO2 emissions compared to the vehicles with internal combustion engines, electric vehicles provide an opportunity to reduce environmental impacts of transportation. In this direction, a problem for plug-in electric vehicles (PEVs) is studied where the aim is to find an energy efficient path. Given an origin–destination pair over a directed network, this problem involves determining a path joining origin and destination, the speed of the PEV on each road segment, i.e., arc, along the path, the charging stations the PEV will stop by, and how much to recharge at each stop so as to minimize the total amount energy consumption. There are speed limits on each road segment, and PEV has to arrive at the destination on or before a given total time limit. For this problem, firstly, a mixed-integer second order cone programming formulation (MISOCP) is proposed. Secondly, to be able to solve larger size instances, a matheuristic is developed. Lastly, an iterated local search (ILS) algorithm is designed for this problem. Solution quality and computation times of the heuristics and the exact algorithm are compared on different instances. Differently from the literature, the speed values of the PEV on the arcs are considered as continuous decision variables in all proposed solution approaches. Moreover, consideration of the speed limits which can be legal limits or limits imposed by congestion makes our problem more realistic. The analysis of the results of the computational experiments gives the user an insight to select the proper solution approach based on the instance settings. MISOCP formulation becomes inadequate for larger instances. On the other hand, the heuristic solution approaches can solve such instances within reasonable computational times and therefore they have the potential to be integrated in some software to dynamically find energy efficient paths.

Echocardiography- versus intracardiac electrocardiogram-based optimization of cardiac resynchronization therapy: A systematic review.

This systematic review aimed to evaluate the performance of echocardiography-based programming in comparison with the intracardiac electrocardiogram (IEGM)-based method for the optimization of cardiac resynchronization therapy (CRT). A literature review was conducted using digital databases to systematically identify the studies reporting CRT optimization through echocardiography compared with IEGM. Detailed patient-level study characteristics including the type of study, sample size, therapy, the New York Heart Classification (NYHA) status, lead placement, and other parameters were abstracted. Finally, postprogramming outcomes were extracted for each article. In a total of 11 studies, 919 patients were recruited for the final analysis. Overall, 692 (75.29%) were males. The mean duration of the QRS complex in our study population ranged from 145.2± 21.8ms to 183 ± 19.9ms. There was an equal improvement in the NYHA class between the two methods while the left ventricular ejection fraction (LVEF) demonstrated an improvement by IEGM. Many studies supported IEGM to increase the 6-minute walk test and left ventricular outflow tract velocity time interval (LVOT VTI) when compared to echocardiography. The mean time for echocardiography-based optimization was 60.15 min while that of IEGM-based optimization was 6.65 min. IEGM is an alternative method for CRT optimization in improving the NYHA class, LVEF, and LVOT VTI, and is less time-consuming when compared to the echocardiography-based methods.

Transfer learning accelerating complex parameters optimizations based on quantum-inspired parallel multi-layer Monte Carlo algorithm: Theory, application, implementation

In doubly-fed induction generator-based wind turbines (DFIG-WTs), the rotor-side controller (RSC) with optimized parameters improves wind energy utilization efficiency. With long optimization times and inadequate exploration and development capabilities, conventional intelligent optimization algorithms are hard to find the controller parameters quickly in the complexity and nonlinearity of DFIG-WTs. A quantum-inspired parallel multi-layer Monte Carlo algorithm accelerated by transfer learning (QPMMCOA-TL) is proposed to shorten the optimization time of parameters and obtain the controller parameters more satisfactorily simultaneously. The QPMMCOA-TL possesses strong optimization capabilities through an accelerated search method based on transfer learning, a diversified population coding way, a parallel multi-layer structure, way of searching in the narrowing feasible region. In the optimization process, the fitness function replaced by trained deep neural networks is transferred to the search process of the QPMMCOA-TL for shorting the optimization time. The QPMMCOA-TL is applied to test two benchmark functions and compared with seven metaheuristic algorithms for completing the validity verification. The optimization time of the QPMMCOA-TL when searching the parameters of the RSC is 1188 s, which is one-tenth or less than other algorithms. Furthermore, the reliability and stability of the optimized controller are comprehensively enhanced.

Enumeration optimization of open pit production scheduling based on mobile capacity search domain

The optimization of open pit mine production scheduling is not only a multistage decision-making problem but also involves space–time dynamic action among multiple factors, which makes it difficult to optimize production capacity, mining sequence, mining life, and other factors simultaneously in optimizing design. In addition, the production capacity is disorderly expanded, the calculation scale is large, and the optimization time is long. Therefore, this article designs a mobile capacity search domain method to improve computing efficiency without omitting the optimal production capacity. At the same time, taking the maximum net present value as the objective function, an enumeration method is used to optimize the possible paths in different capacity domains and calculate the infrastructure investment and facility idle cost required to meet the maximum production capacity on each possible path to control the disorderly expansion and violent fluctuation of production capacity. The research shows that the open pit mine production scheduling optimization algorithm proposed in this article can not only realize the simultaneous optimization of the three elements of production capacity, mining sequence, and mining life but also improve the computing efficiency by 200 times. Furthermore, the production capacity fluctuation is less than 1.4%. The mining life of the mine is extended by 13 years, and the overall economic benefit is increased by 18%.

DAACO: adaptive dynamic quantity of ant ACO algorithm to solve the traveling salesman problem

AbstractThe traveling salesman problem (TSP) is an NP-hard problem. Thus far, a large number of researchers have proposed different ant colony optimization (ACO) algorithms to solve the TSP. These algorithms inevitably encounter problems such as long convergence time and the tendency to easily fall into local optima. On the basis of the ACO algorithm, this study proposes a dynamic adaptive ACO algorithm (DAACO). DAACO realizes the diversity of initialization of the ACO algorithm by dynamically determining the number of ants to be prevented from falling into local optimization. DAACO also adopts a hybrid local selection strategy to increase the quality of ant optimization and reduce the optimization time. Among the 20 instances of the TSPLIB dataset, the DAACO algorithm obtains 19 optimal values, and the solutions of 10 instances are better than those of other algorithms. The experimental results on the TSPLIB dataset show that the DAACO algorithm has obvious advantages in terms of convergence time, solution quality, and average value relative to existing state-of-the-art ACO algorithms.

Deep-learning-based aerodynamic shape optimization of rotor airfoils to suppress dynamic stall

The use of computational fluid dynamics (CFD) to optimize the aerodynamic shape of rotor airfoils with the aim of suppressing dynamic stall is computationally expensive and inefficient. To address this issue, a surrogate model based on deep learning (DL) is employed to replace a CFD module of optimization process in this study. The optimization framework is demonstrated by optimizing a SC1095 rotor airfoil in subsonic flow. The airfoil shapes are parameterized using the class function/shape function transformation method, and an airfoil dataset for a deep neural network (DNN) is generated by Latin hypercube sampling. The surrogate model for predicting the aerodynamic coefficients of the airfoils is established by training the DNN, which can get the predicted results within a second. This surrogate model is then combined with the multi-island genetic algorithm for rotor airfoil optimization. Finally, the aerodynamic performance of a rotor with the optimized airfoil is investigated to verify the dynamic stall suppression effect. The results demonstrate that the peak drag and moment coefficients of the optimized airfoil can be reduced by 82.5% and 88.6%, respectively, compared with the baseline airfoil, while the lift coefficients increase during almost all of the pitching period. This means that the optimized airfoil has significantly improved dynamic stall characteristics. Moreover, by suppressing the dynamic stall, the rotor with the optimized airfoil achieves better aerodynamic performance than the baseline rotor. Statistical data show that our use of a DL-based surrogate model instead of the CFD module will reduce the optimization time by at least one order of magnitude.

Adaptive Resource Planning for AI Workloads with Variable Real-Time Tasks

AI (Artificial Intelligence) workloads are proliferating in modern real-time systems. As the tasks of AI workloads fluctuate over time, resource planning policies used for traditional fixed real-time tasks should be re-examined. In particular, it is difficult to immediately handle changes in real-time tasks without violating the deadline constraints. To cope with this situation, this paper analyzes the task situations of AI workloads and finds the following two observations. First, resource planning for AI workloads is a complicated search problem that requires much time for optimization. Second, although the task set of an AI workload may change over time, the possible combinations of the task sets are known in advance. Based on these observations, this paper proposes a new resource planning scheme for AI workloads that supports the re-planning of resources. Instead of generating resource plans on the fly, the proposed scheme pre-determines resource plans for various combinations of tasks. Thus, in any case, the workload is immediately executed according to the resource plan maintained. Specifically, the proposed scheme maintains an optimized CPU (Central Processing Unit) and memory resource plan using genetic algorithms and applies it as soon as the workload changes. The proposed scheme is implemented in the open-source simulator SimRTS for the validation of its effectiveness. Simulation experiments show that the proposed scheme reduces the energy consumption of CPU and memory by 45.5% on average without deadline misses.

Research on multi-objective optimization of construction engineering based on improved genetic algorithm

The role of cost, time and labor intensity in the construction process is very important, but there is a problem of poor optimization effect. The standard genetic algorithm cannot solve the problem of index optimization in the construction process, and the optimization rationality is low. Therefore, this paper proposes an improved genetic algorithm to study the objectives of construction projects. Firstly, the k-value clustering theory is used to analyze the characteristic values of the index and divide them according to the changing trend to reduce the disturbance factors in optimization. Then, the k-value clustering theory is used to search for the child-parent node of the indicator to form the final optimization result and complete the multi-objective optimization. MATLAB simulation shows that under certain optimization criteria, improving the genetic algorithm can improve the rationality of multi-objective optimization, and the optimization time is shorter than that standard improves genetic algorithms.

APSO: An A*-PSO Hybrid Algorithm for Mobile Robot Path Planning

Aiming at the problems of the A* algorithm in mobile robot path planning, such as multiple nodes, low path accuracy, long running time and difficult path initialization of particle swarm optimization, an APSO algorithm combining A* and PSO was proposed to calculate the optimal path. First, a redundant point removal strategy is adopted to preliminarily optimize the path planned by the A* algorithm and obtain the set of key nodes. Second, a stochastic inertia weight is proposed to improve the search ability of PSO. Third, a stochastic opposition-based learning strategy is proposed to further improve the search ability of PSO. Fourth, the global path is obtained by using the improved PSO to optimize the set of key nodes. Fifth, a motion time objective function that is more in line with the actual motion requirements of the mobile robot is used to evaluate the algorithm. The simulation results of path planning show that the path planned by APSO not only reduces the running time of the mobile robot by 17.35%, 14.84%, 15.31%, 15.21%, 18.97%, 15.70% compared with the A* algorithm in the six environment maps but also outperforms other path planning algorithms to varying degrees. Therefore, the proposed APSO is more in line with the actual movement of the mobile robot.

A Continuous Actor–Critic Deep Q-Learning-Enabled Deployment of UAV Base Stations: Toward 6G Small Cells in the Skies of Smart Cities

Uncrewed aerial vehicle-mounted base stations (UAV-BSs), also know as drone base stations, are considered to have promising potential to tackle the limitations of ground base stations. They can provide cost-effective Internet connection to es that are out of infrastructure. They can also take over quickly as service providers when ground base stations fail in an unanticipated manner. UAV-BSs benefit from their mobile nature that enables them to change their 3D locations if the demand profile changes rapidly. In order to effectively leverage the mobility of UAV-BSs so as to maximize the performance of the network, 3D location of UAV-BSs requires continuous optimization. However, solving the optimization problem of UAVBSs is NP-hard with no deterministic solution in polynomial time. In this paper, we propose a continuous actor-critic deep reinforcement learning solution in order to solve the location optimization problem of UAV-BSs in the presence of mobile endpoints. The simulation results show that the proposed model significantly improves the network performance compared to Qlearning, deep Q-learning and conventional algorithms. While the Q-learning and deep Q-learning-based baselines reach the sum data rate of 35 Mbps and 42 Mbps respectively, our proposed ACDQL-based strategy maximizes the sum data rate of endpoints to 45 Mbps. Furthermore, the proposed ACDQLbased methodology reduces the convergence time of the UAV-BS placement optimization by 85 percent compared to the Q-learning and deep Q-learning baselines.

Transport digitalization: Business analytics results from Shift2Rail CONNECTIVE Project

Shift2Rail programme is the first European rail initiative and seeks focused Research and Innovation (R&I) and market-driven solutions by accelerating the integration of new and advanced technologies into innovative rail products and, in a more general way, Mobility as a Service. CONNECTIVE project, one of the project of this programme, works towards the digital transformation of transport services and proposes Business Analytics tools to improve the service provided. The project focused on: 1) the definition of a large set of use cases with the support of different European stakeholders to better understand the scope of their needs; 2) the development of a large set of components to build processing chains to perform Predictive and Prescriptive Analytics. Results are illustrated with one use case concerning train timetable optimisation and dwelling time adaptation in station, regarding the predicted demand.