Search

Abstract Data-driven methods are an innovative model-free approach for engineering and sciences, still in process of maturation. The idea behind is the combination of data analytics techniques, to handle the huge amount of data derived from continuous monitoring or experimental measurements, and of the constraints imposed by universal physical laws, particular to the field in hands. A well-known problem in the former corresponds to the quality and completeness of the available data that, sometimes, are so poor that make the predictions useless. In data-driven simulation-based engineering and sciences (DDSBES), the intrinsic physical constraints may help in completing the missing data in a more precise manner, by forcing them to remain in the manifold defined by the physical laws. In this work, a suitable imputation method to complete incomplete data that preserves the data context-dependent structure is presented. This is accomplished by enforcing the data to fulfill the set of physical constraints, specific to the problem. For this purpose, a generalization of the weighted mean concept is proposed, where the distance to the admissible points (in a physical sense) is used as a weighting function to get the optimal candidate. The method is evaluated in a classical regression problem, where it is compared with other standard methods, showing better results. Then, its application is illustrated in two data-driven problems, where no filling data procedure has been yet proposed, showing good predictive capability, provided that the data are close enough to the actual system state. [ABSTRACT FROM AUTHOR]

Substitution of well-grounded theoretical models by data-driven predictions is not as simple in engineering and sciences as it is in social and economic fields. Scientific problems suffer many times from paucity of data, while they may involve a large number of variables and parameters that interact in complex and non-stationary ways, obeying certain physical laws. Moreover, a physically-based model is not only useful for making predictions, but to gain knowledge by the interpretation of its structure, parameters, and mathematical properties. The solution to these shortcomings seems to be the seamless blending of the tremendous predictive power of the data-driven approach with the scientific consistency and interpretability of physically-based models. We use here the concept of Physically-Guided Neural Networks (PGNN) to predict the input–output relation in a physical system, while, at the same time, fulfilling the physical constraints. With this goal, the internal hidden state variables of the system are associated with a set of internal neuron layers, whose values are constrained by known physical relations, as well as any additional knowledge on the system. Furthermore, when having enough data, it is possible to infer knowledge about the internal structure of the system and, if parameterized, to predict the state parameters for a particular input–output relation. We show that this approach, besides getting physically-based predictions, accelerates the training process, reduces the amount of data required to get similar accuracy, partly filters the intrinsic noise in the experimental data and improves its extrapolation capacity. • Physically guided Neural Networks are used to get physical predictions. • Only measurable values are used to train the network. • The NN topology permits to identify hidden variables with internal neuron values. • Both prediction and characterization problems are solved. • This method accelerates training, and reduces the data required for similar accuracy. • It filters partly the noise in the data and provides improved extrapolation capacity. [ABSTRACT FROM AUTHOR]

Data Science has burst into simulation-based engineering sciences with an impressive impulse. However, data are never uncertainty-free and a suitable approach is needed to face data measurement errors and their intrinsic randomness in problems with well-established physical constraints. As in previous works, this problem is here faced by hybridizing a standard mathematical modeling approach with a new data-driven solver accounting for the phenomenological part of the problem, with the aim of finding a solution point, satisfying some constraints, that minimizes a distance to a given data-set. However, unlike such works that are established in a deterministic framework, we use the Mahalanobis distance in order to incorporate statistical second order uncertainty of data in computations, i.e. variance and correlation. We develop the underlying stochastic theoretical framework and establish the fundamental mathematical and statistical properties. The performance of the resulting reliability-based data-driven procedure is evaluated in a simple but illustrative unidimensional problem as well as in a more realistic solution of a 3D structural problem with a material with intrinsically random constitutive behavior as concrete. The results show, in comparison with other data-driven solvers, better convergence, higher accuracy, clearer interpretation, and major flexibility besides the relevance of allowing uncertainty management with low computational demand. [ABSTRACT FROM AUTHOR]