Is random splitting of a time series suitable for neural networks training when temporal aspects are important? Is random splitting of a time series, e.g., into 80/10/10,  suitable for neural networks (NARX) training when temporal aspects are important? 
To the best of my knowledge ANNs can generalize that means that after learning data from a sample, ANNs can correctly infer the unseen part of a population. But, on the other hand, "a variation caused by different resamplings is higher than a variation caused by network conditions." 
 A: Roberts, David R., et al. "Cross‐validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure." Ecography 40.8 (2017): 913-929.

Ecological data often show temporal, spatial, hierarchical (random effects), or phylogenetic structure. Modern statistical approaches are increasingly accounting for such dependencies. However, when performing cross‐validation, these structures are regularly ignored, resulting in serious underestimation of predictive error. One cause for the poor performance of uncorrected (random) cross‐validation, noted often by modellers, are dependence structures in the data that persist as dependence structures in model residuals, violating the assumption of independence. Even more concerning, because often overlooked, is that structured data also provides ample opportunity for overfitting with non‐causal predictors. This problem can persist even if remedies such as autoregressive models, generalized least squares, or mixed models are used. Block cross‐validation, where data are split strategically rather than randomly, can address these issues. However, the blocking strategy must be carefully considered. Blocking in space, time, random effects or phylogenetic distance, while accounting for dependencies in the data, may also unwittingly induce extrapolations by restricting the ranges or combinations of predictor variables available for model training, thus overestimating interpolation errors. On the other hand, deliberate blocking in predictor space may also improve error estimates when extrapolation is the modelling goal. Here, we review the ecological literature on non‐random and blocked cross‐validation approaches. We also provide a series of simulations and case studies, in which we show that, for all instances tested, block cross‐validation is nearly universally more appropriate than random cross‐validation if the goal is predicting to new data or predictor space, or for selecting causal predictors. We recommend that block cross‐validation be used wherever dependence structures exist in a dataset, even if no correlation structure is visible in the fitted model residuals, or if the fitted models account for such correlations.

