Model Structure Adaptation: A Gradient-Based Approach

2015 ◽  
Author(s):  
William G. La Cava ◽  
Kourosh Danai
Keyword(s):  
First Principles ◽  
Dynamic Models ◽  
Model Structure ◽  
Symbolic Form ◽  
Proposed Model ◽  
Simulation Based ◽  
Gradient Based ◽  
Structure Adaptation ◽  
Continuous Dynamic ◽  
Adaptation Method

A gradient-based method of symbolic adaptation is introduced for a class of continuous dynamic models. The proposed Model Structure Adaptation Method (MSAM) starts with the first-principles model of the system and adapts its structure after adjusting its individual components in symbolic form. A key contribution of this work is its introduction of the model’s parameter sensitivity as the measure of symbolic changes to the model. This measure, which is essential to defining the structural sensitivity of the model, not only accommodates algebraic evaluation of candidate models in lieu of more computationally expensive simulation-based evaluation, but also makes possible the implementation of gradient-based optimization in symbolic adaptation. The applicability of the proposed method is evaluated in application to several models which demonstrate its potential utility.

scholarly journals Predicting the Energy Consumption of a Robot in an Exploration Task Using Optimized Neural Networks

Electronics ◽  
2021 ◽  
Vol 10 (8) ◽  
pp. 920
Author(s):  
Liesle Caballero ◽  
Álvaro Perafan ◽  
Martha Rinaldy ◽  
Winston Percybrooks
Keyword(s):  
Neural Network ◽  
Energy Consumption ◽  
Mobile Robot ◽  
Energy Budget ◽  
Dynamic Models ◽  
Pearson Correlation ◽  
Experimental Conditions ◽  
Grid Map ◽  
Proposed Model ◽  
Exploration Task

This paper deals with the problem of determining a useful energy budget for a mobile robot in a given environment without having to carry out experimental measures for every possible exploration task. The proposed solution uses machine learning models trained on a subset of possible exploration tasks but able to make predictions on untested scenarios. Additionally, the proposed model does not use any kinematic or dynamic models of the robot, which are not always available. The method is based on a neural network with hyperparameter optimization to improve performance. Tabu List optimization strategy is used to determine the hyperparameter values (number of layers and number of neurons per layer) that minimize the percentage relative absolute error (%RAE) while maximize the Pearson correlation coefficient (R) between predicted data and actual data measured under a number of experimental conditions. Once the optimized artificial neural network is trained, it can be used to predict the performance of an exploration algorithm on arbitrary variations of a grid map scenario. Based on such prediction, it is possible to know the energy needed for the robot to complete the exploration task. A total of 128 tests were carried out using a robot executing two exploration algorithms in a grid map with the objective of locating a target whose location is not known a priori by the robot. The experimental energy consumption was measured and compared with the prediction of our model. A success rate of 96.093% was obtained, measured as the percentage of tests where the energy budget suggested by the model was enough to actually carry out the task when compared to the actual energy consumed in the test, suggesting that the proposed model could be useful for energy budgeting in actual mobile robot applications.

scholarly journals Efficient gradient-based parameter estimation for dynamic models using qualitative data

2021 ◽  
Author(s):  
Leonard Schmiester ◽  
Daniel Weindl ◽  
Jan Hasenauer
Keyword(s):  
Parameter Estimation ◽  
Qualitative Data ◽  
Dynamic Models ◽  
Unknown Parameters ◽  
Differential Equation Models ◽  
Gradient Calculation ◽  
Order Of Magnitude ◽  
Gradient Based ◽  
Objective Function Values

AbstractMotivationUnknown parameters of dynamical models are commonly estimated from experimental data. However, while various efficient optimization and uncertainty analysis methods have been proposed for quantitative data, methods for qualitative data are rare and suffer from bad scaling and convergence.ResultsHere, we propose an efficient and reliable framework for estimating the parameters of ordinary differential equation models from qualitative data. In this framework, we derive a semi-analytical algorithm for gradient calculation of the optimal scaling method developed for qualitative data. This enables the use of efficient gradient-based optimization algorithms. We demonstrate that the use of gradient information improves performance of optimization and uncertainty quantification on several application examples. On average, we achieve a speedup of more than one order of magnitude compared to gradient-free optimization. Additionally, in some examples, the gradient-based approach yields substantially improved objective function values and quality of the fits. Accordingly, the proposed framework substantially improves the parameterization of models from qualitative data.AvailabilityThe proposed approach is implemented in the open-source Python Parameter EStimation TOolbox (pyPESTO). All application examples and code to reproduce this study are available at https://doi.org/10.5281/zenodo.4507613.

The Offshore Wave Simulation Based on the Improved P-M Spectrum and Multiple Fractal Interpolation

2014 ◽  
Vol 1030-1032 ◽  
pp. 1832-1836
Author(s):  
Ying Li ◽  
Rui Zhou ◽  
Hao Kuan Li ◽  
Ming Wang
Keyword(s):  
Shallow Water ◽  
Interpolation Method ◽  
Fractal Interpolation ◽  
Fractal Method ◽  
Growth State ◽  
Water Factor ◽  
Proposed Model ◽  
Simulation Based ◽  
Offshore Wave ◽  
The Waves

The Pierson - Moskowitz model is only applicable to full growth state of the waves, and it has low authenticity and hopping phenomenon under the condition of offshore shallow water. This paper proposes a simulation model of offshore wave based on the improved P-M spectrum and multiple fractal interpolation methods. In order to calculate the sea wave with shallow water, a spectrum peak regulation factor and a depth of the water factor are introduced to the P - M spectrum model. Based on this model, the wavelength and wave speed are used as the initial values of wave height. Then, the amplitude and the number of iterations in diamond square fractal method are controlled to obtain the fractal static sea. In order to reduce the influence of the hopping phenomenon to the simulation authenticity, meanwhile, a multiple dynamic non-uniform interpolation method is proposed. The experimental results show that the proposed model can simulate offshore wave with better effect and in real time.

scholarly journals Optimising Window Design on Residential Building Facades by Considering Heat Transfer and Natural Lighting in Nontropical Regions of Australia

Buildings ◽  
2020 ◽  
Vol 10 (11) ◽  
pp. 206
Author(s):  
Zixuan Chen ◽  
Ahmed W. A. Hammad ◽  
Imriyas Kamardeen ◽  
Assed Haddad
Keyword(s):  
Heat Transfer ◽  
Energy Savings ◽  
Significant Proportion ◽  
Residential Building ◽  
Pareto Optimum ◽  
Natural Lighting ◽  
Proposed Model ◽  
Simulation Based ◽  
Building Information ◽  
Window Design

Windows account for a significant proportion of the total energy lost in buildings. The interaction of window type, Window-to-Wall Ratio (WWR) scheduled and window placement height influence natural lighting and heat transfer through windows. This is a pressing issue for nontropical regions considering their high emissions and distinct climatic characteristics. A limitation exists in the adoption of common simulation-based optimisation approaches in the literature, which are hardly accessible to practitioners. This article develops a numerical-based window design optimisation model using a common Building Information Modelling (BIM) platform adopted throughout the industry, focusing on nontropical regions of Australia. Three objective functions are proposed; the first objective is to maximise the available daylight, and the other two emphasize undesirable heat transfer through windows in summer and winter. The developed model is tested on a case study located in Sydney, Australia, and a set of Pareto-optimum solutions is obtained. Through the use of the proposed model, energy savings of up to 8.57% are achieved.

System RBDO With Correlated Variables Using Probabilistic Re-Analysis and Local Metamodels

2010 ◽  
Author(s):  
Ramon C. Kuczera ◽  
Zissimos P. Mourelatos ◽  
Efstratios Nikolaidis
Keyword(s):  
System Reliability ◽  
Random Variables ◽  
Trust Region ◽  
Optimization Approach ◽  
Reliability Based Design Optimization ◽  
Reliability Based Design ◽  
Sensitivity Derivatives ◽  
Simulation Based ◽  
Gradient Based ◽  
Mc Simulation

A simulation-based, system reliability-based design optimization (RBDO) method is presented that can handle problems with multiple failure regions and correlated random variables. Copulas are used to represent dependence between random variables. The method uses a Probabilistic Re-Analysis (PRRA) approach in conjunction with a sequential trust-region optimization approach and local metamodels covering each trust region. PRRA calculates very efficiently the system reliability of a design by performing a single Monte Carlo (MC) simulation per trust region. Although PRRA is based on MC simulation, it calculates “smooth” sensitivity derivatives, allowing the use of a gradient-based optimizer. The PRRA method is based on importance sampling. One requirement for providing accurate results is that the support of the sampling PDF must contain the support of the joint PDF of the input random variables. The trust-region optimization approach satisfies this requirement. Local metamodels are constructed sequentially for each trust region taking advantage of the potential overlap of the trust regions. The metamodels are used to determine the value of the indicator function in MC simulation. An example with correlated input random variables demonstrates the accuracy and efficiency of the proposed RBDO method.

scholarly journals A Simulation-Based Tool to Support Decision-Making in Logistics Design of a Can Packaging Line

2019 ◽  
Vol 0 (0) ◽  
Author(s):  
Victoria G. Achkar ◽  
Valentina Bär ◽  
Franco Cornú ◽  
Carlos A. Méndez
Keyword(s):  
Decision Making ◽  
Material Handling ◽  
Discrete Event ◽  
Inventory Systems ◽  
Dynamic Interactions ◽  
Event Simulation ◽  
Proposed Model ◽  
Simulation Based ◽  
And Performance ◽  
Support Decision Making

AbstractThis study proposes an advanced discrete-event simulation-based tool to support decision-making in the internal logistic design of a packaging line of a multinational brewery company. The selected software, Simio, allows emulating, advising and predicting the behavior of complex real-world systems. The simulation model provides a 3D interface that facilitates verification and validation. In this work, the designed model is used to understand the dynamic interactions between multiple factors and performance measures including both material-handling and inventory systems and to define necessary quantities and/or capacities of resources for a future can packaging line. Based on the proposed model, a what-if analysis is performed to determine inventory threshold values and other critical variables in order to optimize the configuration of internal logistics in potential scenarios.

A Simulation-Based RBDO Method Using Probabilistic Re-Analysis and a Trust Region Approach

2009 ◽  
Author(s):  
Ramon C. Kuczera ◽  
Zissimos P. Mourelatos ◽  
Efstratios Nikolaidis ◽  
Jing Li
Keyword(s):  
System Reliability ◽  
Combustion Engine ◽  
Trust Region ◽  
Optimization Approach ◽  
Reliability Based Design ◽  
Support Set ◽  
Sensitivity Derivatives ◽  
Simulation Based ◽  
Gradient Based ◽  
Mc Simulation

A simulation-based, system reliability-based design optimization (RBDO) method is presented which can handle problems with multiple failure regions. The method uses a Probabilistic Re-Analysis (PRRA) approach in conjunction with a trust-region optimization approach. PRRA calculates very efficiently the system reliability of a design by performing a single Monte Carlo (MC) simulation. Although PRRA is based on MC simulation, it calculates “smooth” sensitivity derivatives, allowing therefore, the use of a gradient-based optimizer. The PRRA method is based on importance sampling. It provides accurate results, if the support (set of all values for which a function is non zero) of the sampling PDF contains the support of the joint PDF of the input random variables and, if the mass of the input joint PDF is not concentrated in a region where the sampling PDF is almost zero. A sequential, trust-region optimization approach satisfies these two requirements. The potential of the proposed method is demonstrated using the design of a vibration absorber, and the system RBDO of an internal combustion engine.

Sign in / Sign up

Export Citation Format

Share Document