Weijuan Cao, T. Robinson, Yang Hua, F. Boussuge, Andrew R. Colligan, Wanbin Pan
In this paper, the application of deep learning methods to the task of machining feature recognition in CAD models is studied. Four contributions are made: 1. An automatic method to generate large datasets of 3D CAD models is proposed, where each model contains multiple machining features with face labels. 2. A concise and informative graph representation for 3D CAD models is presented. This is shown to be applicable to graph neural networks. 3. The graph representation is compared with voxels on their performance of training deep neural networks to segment 3D CAD models. 4. Experiments are also conducted to evaluate the effectiveness of graph-based deep learning for interacting feature recognition. Results show that the proposed graph representation is a more efficient representation of 3D CAD models than voxels for deep learning. It is also shown that graph neural networks can be used to recognize individual features on the model and also identify complex interacting features.
{"title":"Graph Representation of 3D CAD Models for Machining Feature Recognition With Deep Learning","authors":"Weijuan Cao, T. Robinson, Yang Hua, F. Boussuge, Andrew R. Colligan, Wanbin Pan","doi":"10.1115/detc2020-22355","DOIUrl":"https://doi.org/10.1115/detc2020-22355","url":null,"abstract":"\u0000 In this paper, the application of deep learning methods to the task of machining feature recognition in CAD models is studied. Four contributions are made:\u0000 1. An automatic method to generate large datasets of 3D CAD models is proposed, where each model contains multiple machining features with face labels.\u0000 2. A concise and informative graph representation for 3D CAD models is presented. This is shown to be applicable to graph neural networks.\u0000 3. The graph representation is compared with voxels on their performance of training deep neural networks to segment 3D CAD models.\u0000 4. Experiments are also conducted to evaluate the effectiveness of graph-based deep learning for interacting feature recognition.\u0000 Results show that the proposed graph representation is a more efficient representation of 3D CAD models than voxels for deep learning. It is also shown that graph neural networks can be used to recognize individual features on the model and also identify complex interacting features.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124152694","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Topology optimization is a powerful tool to generate mechanical designs that use minimal mass to achieve their function. However, the designs obtained using topology optimization are often not manufacturable using a given manufacturing process. There exist some modifications to the traditional topology optimization algorithm that are able to impose manufacturing constraints for a limited set of manufacturing methods. These approaches have the drawback that they are often based on heuristics to obtain the manufacturability constraint and thus cannot be applied generally to multiple manufacturing methods. In order to create a general approach to imposing manufacturing constraints on topology optimization, generative adversarial networks (GANs) are used. GANs have the capability to produce samples from a distribution defined by training data. In this work, the GAN is trained by generating synthetic 3D voxel training data that represent the distribution of designs that can be created by a particular manufacturing method. Once trained, the GAN forms a mapping from a latent vector space to the space of manufacturable designs. The topology optimization is then performed on the latent vector space ensuring that the design obtained is manufacturable. The effectiveness of this approach is demonstrated by training a GAN on designs intended to be manufacturable on a 3-axis computer numerically controlled (CNC) milling machine.
{"title":"Generative Adversarial Networks With Synthetic Training Data for Enforcing Manufacturing Constraints on Topology Optimization","authors":"M. Greminger","doi":"10.1115/detc2020-22399","DOIUrl":"https://doi.org/10.1115/detc2020-22399","url":null,"abstract":"\u0000 Topology optimization is a powerful tool to generate mechanical designs that use minimal mass to achieve their function. However, the designs obtained using topology optimization are often not manufacturable using a given manufacturing process. There exist some modifications to the traditional topology optimization algorithm that are able to impose manufacturing constraints for a limited set of manufacturing methods. These approaches have the drawback that they are often based on heuristics to obtain the manufacturability constraint and thus cannot be applied generally to multiple manufacturing methods. In order to create a general approach to imposing manufacturing constraints on topology optimization, generative adversarial networks (GANs) are used. GANs have the capability to produce samples from a distribution defined by training data. In this work, the GAN is trained by generating synthetic 3D voxel training data that represent the distribution of designs that can be created by a particular manufacturing method. Once trained, the GAN forms a mapping from a latent vector space to the space of manufacturable designs. The topology optimization is then performed on the latent vector space ensuring that the design obtained is manufacturable. The effectiveness of this approach is demonstrated by training a GAN on designs intended to be manufacturable on a 3-axis computer numerically controlled (CNC) milling machine.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"26 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126995302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Network-like engineering systems, such as transport networks and lattice metamaterials, are featured by high dimensional, complex topological characteristics, which pose a great challenge for uncertainty quantification (UQ). Existing UQ approaches are only applicable to parametric uncertainties, or high dimensional random quantities distributed in a simply connected space (e.g., line section, rectangular area, etc.). The topological characteristics of the input space cannot be captured by existing UQ models. To resolve this issue, a network-based Gaussian random process UQ approach is proposed in this work. By representing the topological input space as a node-edge network, network distance is employed to replace the Euclidean distance in characterizing the spatial correlations. Furthermore, a conditional simulation-based approach is proposed for sampling. Realizations of random quantities on each edge of the network is sampled conditionally on the node values, which are modeled by a multivariable Gaussian distribution. The effectiveness of the proposed approach is demonstrated with two engineering case studies: stochastic thermal conduction analysis of a 3D lattice structure, and characterization of the distortion pattern of an additively manufactured cellular structure.
{"title":"Quantification of Uncertainties Distributed in Network-Like Systems","authors":"Zihan Wang, Hongyi Xu","doi":"10.1115/detc2020-22082","DOIUrl":"https://doi.org/10.1115/detc2020-22082","url":null,"abstract":"\u0000 Network-like engineering systems, such as transport networks and lattice metamaterials, are featured by high dimensional, complex topological characteristics, which pose a great challenge for uncertainty quantification (UQ). Existing UQ approaches are only applicable to parametric uncertainties, or high dimensional random quantities distributed in a simply connected space (e.g., line section, rectangular area, etc.). The topological characteristics of the input space cannot be captured by existing UQ models. To resolve this issue, a network-based Gaussian random process UQ approach is proposed in this work. By representing the topological input space as a node-edge network, network distance is employed to replace the Euclidean distance in characterizing the spatial correlations. Furthermore, a conditional simulation-based approach is proposed for sampling. Realizations of random quantities on each edge of the network is sampled conditionally on the node values, which are modeled by a multivariable Gaussian distribution. The effectiveness of the proposed approach is demonstrated with two engineering case studies: stochastic thermal conduction analysis of a 3D lattice structure, and characterization of the distortion pattern of an additively manufactured cellular structure.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128250134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
While current neural networks (NNs) are becoming good at deriving single types of abstractions for a small set of phenomena, for example, using a single NN to predict a flow velocity field, NNs are not good at composing large systems as compositions of small phenomena and reasoning about their interactions. We want to study how NNs build both the abstraction and composition of phenomena when a single NN model cannot suffice. Rather than a single NN that learns one physical or social phenomenon, we want a group of NNs that learn to abstract, compose, reason, and correct the behaviors of different parts in a system. In this paper, we investigate the joint use of Physics-Informed (Navier-Stokes equations) Deep Neural Networks (i.e., Deconvolutional Neural Networks) as well as Geometric Deep Learning (i.e., Graph Neural Networks) to learn and compose fluid component behavior. Our models successfully predict the fluid flows and their composition behaviors (i.e., velocity fields) with an accuracy of about 99%.
{"title":"Learning to Abstract and Compose Mechanical Device Function and Behavior","authors":"Jun Wang, Kevin N. Chiu, M. Fuge","doi":"10.1115/detc2020-22714","DOIUrl":"https://doi.org/10.1115/detc2020-22714","url":null,"abstract":"\u0000 While current neural networks (NNs) are becoming good at deriving single types of abstractions for a small set of phenomena, for example, using a single NN to predict a flow velocity field, NNs are not good at composing large systems as compositions of small phenomena and reasoning about their interactions. We want to study how NNs build both the abstraction and composition of phenomena when a single NN model cannot suffice. Rather than a single NN that learns one physical or social phenomenon, we want a group of NNs that learn to abstract, compose, reason, and correct the behaviors of different parts in a system. In this paper, we investigate the joint use of Physics-Informed (Navier-Stokes equations) Deep Neural Networks (i.e., Deconvolutional Neural Networks) as well as Geometric Deep Learning (i.e., Graph Neural Networks) to learn and compose fluid component behavior. Our models successfully predict the fluid flows and their composition behaviors (i.e., velocity fields) with an accuracy of about 99%.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123655195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Katherine Edmonds, Alex Mikes, Bryony DuPont, R. Stone
Expanding on previous work of automating functional modeling, we have developed a more informed automation approach by assigning a weighted confidence metric to the wide variety of data in a design repository. Our work focuses on automating what we call linear functional chains, which are a component-based section of a full functional model. We mine the Design Repository to find correlations between component and function and flow. The automation algorithm we developed organizes these connections by component-function-flow frequency (CFF frequency), thus allowing the creation of linear functional chains. In previous work, we found that CFF frequency is the best metric in formulating the linear functional chain for an individual component; however, we found that this metric did not account for prevalence and consistency in the Design Repository data. To better understand our data, we developed a new metric, which we refer to as weighted confidence, to provide insight on the fidelity of the data, calculated by taking the harmonic mean of two metrics we extracted from our data, prevalence, and consistency. This method could be applied to any dataset with a wide range of individual occurrences. The contribution of this research is not to replace CFF frequency as a method of finding the most likely component-function-flow correlations but to improve the reliability of the automation results by providing additional information from the weighted confidence metric. Improving these automation results, allows us to further our ultimate objective of this research, which is to enable designers to automatically generate functional models for a product given constituent components.
{"title":"A Weighted Confidence Metric to Improve Automated Functional Modeling","authors":"Katherine Edmonds, Alex Mikes, Bryony DuPont, R. Stone","doi":"10.1115/detc2020-22495","DOIUrl":"https://doi.org/10.1115/detc2020-22495","url":null,"abstract":"\u0000 Expanding on previous work of automating functional modeling, we have developed a more informed automation approach by assigning a weighted confidence metric to the wide variety of data in a design repository. Our work focuses on automating what we call linear functional chains, which are a component-based section of a full functional model. We mine the Design Repository to find correlations between component and function and flow. The automation algorithm we developed organizes these connections by component-function-flow frequency (CFF frequency), thus allowing the creation of linear functional chains. In previous work, we found that CFF frequency is the best metric in formulating the linear functional chain for an individual component; however, we found that this metric did not account for prevalence and consistency in the Design Repository data. To better understand our data, we developed a new metric, which we refer to as weighted confidence, to provide insight on the fidelity of the data, calculated by taking the harmonic mean of two metrics we extracted from our data, prevalence, and consistency. This method could be applied to any dataset with a wide range of individual occurrences. The contribution of this research is not to replace CFF frequency as a method of finding the most likely component-function-flow correlations but to improve the reliability of the automation results by providing additional information from the weighted confidence metric. Improving these automation results, allows us to further our ultimate objective of this research, which is to enable designers to automatically generate functional models for a product given constituent components.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126216017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Emerging 3D printing technologies are enabling the design and fabrication of novel architected structures with advantageous mechanical responses. Designing complex structures, such as lattices, with a targeted response is challenging because build materials, fabrication process, and topological design have unique influences on the structure’s mechanical response. Changing any factor may have unanticipated consequences, even for simpler lattice structures. Here, we conduct mechanical compression experiments to investigate varied lattice design, fabrication, and material combinations using stereolithography printing with a biocompatible polymer. Mechanical testing demonstrates that a higher ultraviolet curing time increases elastic modulus. Material testing demonstrated that anisotropy does not strongly influence lattice mechanics. Designs were altered by comparing homogenous lattices of single unit cell types and heterogeneous lattices that combine two types of unit cells. Unit cells for heterogeneous structures include a Cube design for a high elastic modulus and Cross design for improved shear response. Mechanical testing of three heterogeneous layouts demonstrated how unit cell organization influences mechanical outcomes, therefore enabling the tuning of an elastic modulus that surpasses the law of averages designed for application-dependent mechanical needs. These findings provide a foundation for linking design, process, and material for engineering 3D printed structures with preferred properties, while also facilitating new directions in design automation and optimization.
{"title":"Structure, Process, and Material Influences for 3D Printed Lattices Designed With Mixed Unit Cells","authors":"Gabriel Briguiet, P. Egan","doi":"10.1115/detc2020-22575","DOIUrl":"https://doi.org/10.1115/detc2020-22575","url":null,"abstract":"\u0000 Emerging 3D printing technologies are enabling the design and fabrication of novel architected structures with advantageous mechanical responses. Designing complex structures, such as lattices, with a targeted response is challenging because build materials, fabrication process, and topological design have unique influences on the structure’s mechanical response. Changing any factor may have unanticipated consequences, even for simpler lattice structures. Here, we conduct mechanical compression experiments to investigate varied lattice design, fabrication, and material combinations using stereolithography printing with a biocompatible polymer. Mechanical testing demonstrates that a higher ultraviolet curing time increases elastic modulus. Material testing demonstrated that anisotropy does not strongly influence lattice mechanics. Designs were altered by comparing homogenous lattices of single unit cell types and heterogeneous lattices that combine two types of unit cells. Unit cells for heterogeneous structures include a Cube design for a high elastic modulus and Cross design for improved shear response. Mechanical testing of three heterogeneous layouts demonstrated how unit cell organization influences mechanical outcomes, therefore enabling the tuning of an elastic modulus that surpasses the law of averages designed for application-dependent mechanical needs. These findings provide a foundation for linking design, process, and material for engineering 3D printed structures with preferred properties, while also facilitating new directions in design automation and optimization.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131027455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bayesian optimization is a versatile numerical method to solve global optimization problems of high complexity at a reduced computational cost. The efficiency of Bayesian optimization relies on two key elements: a surrogate model and an acquisition function. The surrogate model is generated on a Gaussian process statistical framework and provides probabilistic information of the prediction. The acquisition function, which guides the optimization, uses the surrogate probabilistic information to balance the exploration and the exploitation of the design space. In the case of multi-objective problems, current implementations use acquisition functions such as the multi-objective expected improvement (MEI). The evaluation of MEI requires a surrogate model for each objective function. In order to expand the Pareto front, such implementations perform a multi-variate integral over an intricate hypervolume, which require high computational cost. The objective of this work is to introduce an efficient multi-objective Bayesian optimization method that avoids the need for multi-variate integration. The proposed approach employs the working principle of multi-objective traditional methods, e.g., weighted sum and min-max methods, which transform the multi-objective problem into a single-objective problem through a functional mapping of the objective functions. Since only one surrogate is trained, this approach has a low computational cost. The effectiveness of the proposed approach is demonstrated with the solution of four problems: (1) an unconstrained version of the Binh and Korn test problem (convex Pareto front), (2) the Fonseca and Fleming test problem (non-convex Pareto front), (3) a three-objective test problem and (4) the design optimization of a sandwich composite armor for blast mitigation. The optimization algorithm is implemented in MATLAB and the finite element simulations are performed in the explicit, nonlinear finite element analysis code LS-DYNA. The results are comparable (or superior) to the results of the MEI acquisition function.
{"title":"A Simple and Effective Methodology to Perform Multi-Objective Bayesian Optimization: An Application in the Design of Sandwich Composite Armors for Blast Mitigation","authors":"H. Valladares, A. Tovar","doi":"10.1115/detc2020-22564","DOIUrl":"https://doi.org/10.1115/detc2020-22564","url":null,"abstract":"\u0000 Bayesian optimization is a versatile numerical method to solve global optimization problems of high complexity at a reduced computational cost. The efficiency of Bayesian optimization relies on two key elements: a surrogate model and an acquisition function. The surrogate model is generated on a Gaussian process statistical framework and provides probabilistic information of the prediction. The acquisition function, which guides the optimization, uses the surrogate probabilistic information to balance the exploration and the exploitation of the design space. In the case of multi-objective problems, current implementations use acquisition functions such as the multi-objective expected improvement (MEI). The evaluation of MEI requires a surrogate model for each objective function. In order to expand the Pareto front, such implementations perform a multi-variate integral over an intricate hypervolume, which require high computational cost. The objective of this work is to introduce an efficient multi-objective Bayesian optimization method that avoids the need for multi-variate integration. The proposed approach employs the working principle of multi-objective traditional methods, e.g., weighted sum and min-max methods, which transform the multi-objective problem into a single-objective problem through a functional mapping of the objective functions. Since only one surrogate is trained, this approach has a low computational cost. The effectiveness of the proposed approach is demonstrated with the solution of four problems: (1) an unconstrained version of the Binh and Korn test problem (convex Pareto front), (2) the Fonseca and Fleming test problem (non-convex Pareto front), (3) a three-objective test problem and (4) the design optimization of a sandwich composite armor for blast mitigation. The optimization algorithm is implemented in MATLAB and the finite element simulations are performed in the explicit, nonlinear finite element analysis code LS-DYNA. The results are comparable (or superior) to the results of the MEI acquisition function.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125450982","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In product design, it is essential to understand customers’ preferences for product features. Traditional methods including the survey and interview are time-consuming and costly. As an alternative, research on utilizing online data for user analysis has been actively conducted. Although various methods have been proposed in this domain, most of them focus on the main features or usages of the product. However, from the manufacturer’s perspective, sub-features are as crucial as main features or usages, because the preference for sub-features is necessary for component configuration in actual product development. As the first step to solve this problem, this paper proposes a methodology to extract and cluster sub-features by incorporating phrase embedding into the previous word embedding. Also, the presented methodology increases the accuracy and diversity of the clustering result by using X-means clustering as a noise filter and adopting spectral clustering.
{"title":"Improving the Accuracy and Diversity of Feature Extraction From Online Reviews Using Keyword Embedding and Two Clustering Methods","authors":"Seyoung Park, Harrison M. Kim","doi":"10.1115/detc2020-22642","DOIUrl":"https://doi.org/10.1115/detc2020-22642","url":null,"abstract":"\u0000 In product design, it is essential to understand customers’ preferences for product features. Traditional methods including the survey and interview are time-consuming and costly. As an alternative, research on utilizing online data for user analysis has been actively conducted. Although various methods have been proposed in this domain, most of them focus on the main features or usages of the product. However, from the manufacturer’s perspective, sub-features are as crucial as main features or usages, because the preference for sub-features is necessary for component configuration in actual product development. As the first step to solve this problem, this paper proposes a methodology to extract and cluster sub-features by incorporating phrase embedding into the previous word embedding. Also, the presented methodology increases the accuracy and diversity of the clustering result by using X-means clustering as a noise filter and adopting spectral clustering.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125101560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The energy management strategy plays a critical role in scheduling the operations and enhancing the overall efficiency for electric vehicles. This paper proposes an effective model predictive control-based (MPC) energy management strategy to simultaneously control the battery thermal management system (BTMS) and the cabin air conditioning (AC) system for electric vehicles (EVs). We aim to improve the overall energy efficiency, while retaining soft constraints from both BTMS and AC systems. It is implemented by optimizing the operation and discharging schedule to avoid peak load and by directly utilizing the regenerative power instead of recharging. Compared to the systematic performance without any control coordination between BTMS and AC, results reveal that there are a 4.3% reduction for the recharging energy, and a 6.5% improvement for the overall energy consumption that gained from the MPC-based energy management strategy. Overall the MPC-based energy management is a promising solution to enhance the efficiency for electric vehicles.
{"title":"A Model Predictive Control-Based Energy Management Strategy Considering Electric Vehicle Battery Thermal and Cabin Climate Control","authors":"Yuan Liu, Jie Zhang","doi":"10.1115/detc2020-22318","DOIUrl":"https://doi.org/10.1115/detc2020-22318","url":null,"abstract":"\u0000 The energy management strategy plays a critical role in scheduling the operations and enhancing the overall efficiency for electric vehicles. This paper proposes an effective model predictive control-based (MPC) energy management strategy to simultaneously control the battery thermal management system (BTMS) and the cabin air conditioning (AC) system for electric vehicles (EVs). We aim to improve the overall energy efficiency, while retaining soft constraints from both BTMS and AC systems. It is implemented by optimizing the operation and discharging schedule to avoid peak load and by directly utilizing the regenerative power instead of recharging. Compared to the systematic performance without any control coordination between BTMS and AC, results reveal that there are a 4.3% reduction for the recharging energy, and a 6.5% improvement for the overall energy consumption that gained from the MPC-based energy management strategy. Overall the MPC-based energy management is a promising solution to enhance the efficiency for electric vehicles.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121083832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mojtaba Arezoomand, Elliott J. Rouse, J. Austin-Breneman
A key assumption of new product development is that user requirements and related preferences do not vary on time scales of the process length. However, prior work has identified cases in which user preferences for product attributes can vary with time. This study proposes a method, Design for Dynamic User Preferences, which adapts reinforcement learning (RL) algorithms for designing physical systems whose functionality changes with user feedback. An illustrative case comprised of the design of a variable stiffness prosthetic ankle is presented to evaluate the potential usefulness of the framework. Lifetime user satisfaction for static and dynamic design strategies are compared over simulated user preferences under a number of conditions. Results suggest RL-based strategies outperform static strategies for cases with dynamic user preferences despite significantly less initial information. Within RL methods, upper-confidence bound policies led to higher user satisfaction on average. This study suggests that further investigation into RL-based design strategies is warranted for situations with possibly dynamic preferences.
{"title":"Theoretical Framework for Design for Dynamic User Preferences","authors":"Mojtaba Arezoomand, Elliott J. Rouse, J. Austin-Breneman","doi":"10.1115/detc2020-22460","DOIUrl":"https://doi.org/10.1115/detc2020-22460","url":null,"abstract":"\u0000 A key assumption of new product development is that user requirements and related preferences do not vary on time scales of the process length. However, prior work has identified cases in which user preferences for product attributes can vary with time. This study proposes a method, Design for Dynamic User Preferences, which adapts reinforcement learning (RL) algorithms for designing physical systems whose functionality changes with user feedback. An illustrative case comprised of the design of a variable stiffness prosthetic ankle is presented to evaluate the potential usefulness of the framework. Lifetime user satisfaction for static and dynamic design strategies are compared over simulated user preferences under a number of conditions. Results suggest RL-based strategies outperform static strategies for cases with dynamic user preferences despite significantly less initial information. Within RL methods, upper-confidence bound policies led to higher user satisfaction on average. This study suggests that further investigation into RL-based design strategies is warranted for situations with possibly dynamic preferences.","PeriodicalId":415040,"journal":{"name":"Volume 11A: 46th Design Automation Conference (DAC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129667716","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：