Multi-step prediction of photovoltaic power based on two-stage decomposition and BILSTM

The isothermal compressive tests are performed to analyze the hot deformation characteristics of Inconel 625 superalloy over a wide temperature range of 1000–1150 °C and the strain rate scope of 0.01–10 s−1. The hot deformation mechanisms and the impacts of deformation parameters on the thermal deformation features, are analyzed. It reveals that the true stress first rises to a maximum stress level and then progressively drops to a comparatively stable level, demonstrating apparent work-hardening (WH) and dynamic softening (DS) features. Meanwhile, the true stress swiftly rises with the reduced deformation temperature or the increased strain rate. Moreover, the almost fine and uniform grains surround the whole microstructure under relatively large true strain, elevated temperature, and depressed strain rate. Besides, three kinds of constitutive models, including the Arrhenius-type (AT) model, physical-based (PB) model, and artificial-neural-network (ANN) model, complied with the backpropagation learning (BP) and new binary particle swarm optimization (NBPSO) algorithms are developed, respectively. The correlation coefficient (R) and absolute relative error (ARE) distribution are estimated to assess the predictive abilities of the established models. It can be concluded that the maximum absolute relative errors are near 12%, 15%, and 7.2% for AT, PB, and NBPSO-BP ANN models, respectively. Meanwhile, the correlation coefficient of the NBPSO-BP ANN model is as high as 0.9983, while the corresponding ones for the AT and PB models are 0.9951 and 0.9903, respectively. Thus, the established NBPSO-BP ANN model can depict the hot deformation features of the studied superalloy precisely. 

http://manuscript.elsevier.com/S2352492822007127/pdf/S2352492822007127.pdf

High-temperature deformation characteristics and constitutive models of Inconel 625 superalloy

DAFA-BiLSTM: Deep Autoregression Feature Augmented Bidirectional LSTM network for time series prediction

Discriminative and regularized echo state network for time series classification

Time-series data is an appealing study topic in data mining and has a broad range of applications. Many approaches have been employed to handle time series classification (TSC) challenges with promising results, among which deep neural network methods have become mainstream. Echo State Networks (ESN) and Convolutional Neural Networks (CNN) are commonly utilized as deep neural network methods in TSC research. However, ESN and CNN can only extract local dependencies relations of time series, resulting in long-term temporal data dependence needing to be more challenging to capture. As a result, an encoder and decoder architecture named LA-ESN is proposed for TSC tasks. In LA-ESN, the encoder is composed of ESN, which is utilized to obtain the time series matrix representation. Meanwhile, the decoder consists of a one-dimensional CNN (1D CNN), a Long Short-Term Memory network (LSTM) and an Attention Mechanism (AM), which can extract local information and global dependencies from the representation. Finally, many comparative experimental studies were conducted on 128 univariate datasets from different domains, and three evaluation metrics including classification accuracy, mean error and mean rank were exploited to evaluate the performance. In comparison to other approaches, LA-ESN produced good results.

/pdf/la-esn-a-novel-method-for-time-series-classification-2ux95cih.pdf

LA-ESN: A Novel Method for Time Series Classification

Echo state network with logistic mapping and bias dropout for time series prediction

How to improve the behavioral decision-making ability and adaptability to unknown environments is of great importance for an agent. The traditional decision-making methods usually suffer from long training time, due to the large amount of training samples, or low adaptability to the unknown environments, or lack of the continuous learning capacities, etc. In response to these problems, this work proposes a novel motivated developmental network (MDN) to improve the decision-making ability of the agent. During the environment exploration, if the agent encounters an unknown environment, new layers and neurons are dynamically inserted to the MDN, according to the task requirements. Through the interaction between internal neurons and the inserted new neurons, the agent can autonomously develop and learn in the unknown environments without training data, but the behavioral decision-making at this stage is random. To further improve the agent’s decision-making ability, in the off-task process, through the gated self-organization mechanism, the agent can selectively recall the specific knowledge in its “brain”, and the MDN will transfer and generate a large amount of new data, according to the recalled knowledge, then the new layers and neurons will be inserted to memorize the new knowledge. Hence the knowledge base of the MDN becomes more and more complete, thereby improving its decision-making ability and adaptability to the new environment. To demonstrate the performance of the MDN model, a mobile robot navigation in different environments are executed. The experimental results illustrate that the agent can not only autonomously learn in static environments, but also has better decision-making ability in unknown dynamic environment, i.e., better adaptability to the new environment. Comparison with other algorithms further demonstrate the potential of the proposed MDN model.

Behavioral Decision-Making of Mobile Robot in Unknown Environment with the Cognitive Transfer

Reservoir Computing (RC) defines a Recurrent Neural Network (RNN) method that optimally processes transient information that generated based on recurrent tasks utilizing sequential datasets. As a typical actualization of the RC classic, Echo State Network (ESN) can train the transient data by linear optimization using very small training datasets and make the network operations fairly fast. However, classical ESNs use a completely randomized reservoir matrix to delineate the fundamental recursive operations and are accompanied by a large number of hyper-parameters to be optimized, such as input weights, reservoir weights, warm-up time steps, and activation functions. The recently proposed Nonlinear Vector Autoregressive (NVAR) algorithm which does not require random inputs and reservoir weights is equated with the traditional ESN reservoir. As a future direction of RC development, NVAR requires few hyper-parameters to be optimized. In this study, a multi-layer ESN model based on the NVAR (NVAR-MLESN) is constructed and applied to time series prediction. Three sequential benchmarks which are synthetic Mackey-Glass chaotic sequence, Sunspot sequence, and real-world Santa Fe Laser sequential dataset are implemented to show the advantage of NVAR-MLESN. Exhaustive simulated outcomes exhibit that our NVAR-MLESN is efficient to develop the performance of classical MLESN and single layer NVAR-ESN. Meanwhile, NVAR-MLESN has fewer parameters to be optimized, faster computing speed, and better performance comparing with the traditional deep ESNs or MLESNs. 

Multi-layer Echo State Network with Nonlinear Vector Autoregression Reservoir for Time Series Prediction

Substation equipment identification is a key step in the process of intelligent substation designing. The target recognition of 3D point cloud of substation equipment firstly uses a 3D laser scanner to obtain the 3D point cloud data that can express the surface shape and spatial position of a substation device, then identify the device information by matching it with template point cloud data. Because the point cloud data obtained by 3D laser scanner is affected by equipment attitude and substation environment, there is redundant, missing, and shadow phenomena in the data, resulting in a low algorithm recognition rate. This paper proposes a method combining plane detection and point cloud registration algorithm, achieving substation equipment identification better result. The main steps of the method include: point cloud preprocessing, plane detecting, preliminary filtering based on Umeyama registration method, and Iterative Closest Point (ICP) algorithm identifying. Plane detection is used for the pre-processed point cloud to obtain planar features, then samples and the planar feature templates in template library are preliminarily selected by Umeyama registration method, and substation equipment is finally identified by Iterative Closest Point algorithm. Experiments show that the recognition rate of devices point cloud reaches 92.2% and the recognition time is shorter, which proves that this recognition algorithm has a good recognition effect. 

Heshan Wang

Papers

Echo state network with logistic mapping and bias dropout for time series prediction

Discriminative and regularized echo state network for time series classification

Behavioral Decision-Making of Mobile Robot in Unknown Environment with the Cognitive Transfer

Multi-layer Echo State Network with Nonlinear Vector Autoregression Reservoir for Time Series Prediction

3D point cloud recognition of substation equipment based on plane detection