With the introduction of the Internet of Things (IoT) and cyberphysical system (CPS) concepts in industrial application scenarios, industrial automation is undergoing a tremendous change. This is made possible in part by recent advances in technology that allow interconnection on a wider and more fine-grained scale. The purpose of this article is to review technological trends and the impact they may have on industrial communication. We will review the impact of IoT and CPSs on industrial automation from an industry 4.0 perspective, give a survey of the current state of work on Ethernet time-sensitive networking (TSN), and shed light on the role of fifth-generation (5G) telecom networks in automation. Moreover, we will point out the need for harmonization beyond networking.

The Future of Industrial Communication: Automation Networks in the Era of the Internet of Things and Industry 4.0

With the developments and applications of the new information technologies, such as cloud computing, Internet of Things, big data, and artificial intelligence, a smart manufacturing era is coming. At the same time, various national manufacturing development strategies have been put forward, such as Industry 4.0 , Industrial Internet , manufacturing based on Cyber-Physical System , and Made in China 2025 . However, one of specific challenges to achieve smart manufacturing with these strategies is how to converge the manufacturing physical world and the virtual world, so as to realize a series of smart operations in the manufacturing process, including smart interconnection, smart interaction, smart control and management, etc. In this context, as a basic unit of manufacturing, shop-floor is required to reach the interaction and convergence between physical and virtual spaces, which is not only the imperative demand of smart manufacturing, but also the evolving trend of itself. Accordingly, a novel concept of digital twin shop-floor (DTS) based on digital twin is explored and its four key components are discussed, including physical shop-floor, virtual shop-floor, shop-floor service system, and shop-floor digital twin data. What is more, the operation mechanisms and implementing methods for DTS are studied and key technologies as well as challenges ahead are investigated, respectively.

Digital Twin Shop-Floor: A New Shop-Floor Paradigm Towards Smart Manufacturing

We propose a novel regularized adaptation technique for context dependent deep neural network hidden Markov models (CD-DNN-HMMs). The CD-DNN-HMM has a large output layer and many large hidden layers, each with thousands of neurons. The huge number of parameters in the CD-DNN-HMM makes adaptation a challenging task, esp. when the adaptation set is small. The technique developed in this paper adapts the model conservatively by forcing the senone distribution estimated from the adapted model to be close to that from the unadapted model. This constraint is realized by adding Kullback-Leibler divergence (KLD) regularization to the adaptation criterion. We show that applying this regularization is equivalent to changing the target distribution in the conventional backpropagation algorithm. Experiments on Xbox voice search, short message dictation, and Switchboard and lecture speech transcription tasks demonstrate that the proposed adaptation technique can provide 2%-30% relative error reduction against the already very strong speaker independent CD-DNN-HMM systems using different adaptation sets under both supervised and unsupervised adaptation setups.

/pdf/kl-divergence-regularized-deep-neural-network-adaptation-for-41ifebp06f.pdf

KL-divergence regularized deep neural network adaptation for improved large vocabulary speech recognition

In this paper, we propose a new fast speaker adaptation method for the hybrid NN-HMM speech recognition model. The adaptation method depends on a joint learning of a large generic adaptation neural network for all speakers as well as multiple small speaker codes (one per speaker). The joint training method uses all training data along with speaker labels to update adaptation NN weights and speaker codes based on the standard back-propagation algorithm. In this way, the learned adaptation NN is capable of transforming each speaker features into a generic speaker-independent feature space when a small speaker code is given. Adaptation to a new speaker can be simply done by learning a new speaker code using the same back-propagation algorithm without changing any NN weights. In this method, a separate speaker code is learned for each speaker while the large adaptation NN is learned from the whole training set. The main advantage of this method is that the size of speaker codes is very small. As a result, it is possible to conduct a very fast adaptation of the hybrid NN/HMM model for each speaker based on only a small amount of adaptation data (i.e., just a few utterances). Experimental results on TIMIT have shown that it can achieve over 10% relative reduction in phone error rate by using only seven utterances for adaptation.

/pdf/fast-speaker-adaptation-of-hybrid-nn-hmm-model-for-speech-3zz9lbo1ua.pdf

Fast speaker adaptation of hybrid NN/HMM model for speech recognition based on discriminative learning of speaker code

This paper proposes a simple yet effective model-based neural network speaker adaptation technique that learns speaker-specific hidden unit contributions given adaptation data, without requiring any form of speaker-adaptive training, or labelled adaptation data. An additional amplitude parameter is defined for each hidden unit; the amplitude parameters are tied for each speaker, and are learned using unsupervised adaptation. We conducted experiments on the TED talks data, as used in the International Workshop on Spoken Language Translation (IWSLT) evaluations. Our results indicate that the approach can reduce word error rates on standard IWSLT test sets by about 8–15% relative compared to unadapted systems, with a further reduction of 4–6% relative when combined with feature-space maximum likelihood linear regression (fMLLR). The approach can be employed in most existing feed-forward neural network architectures, and we report results using various hidden unit activation functions: sigmoid, maxout, and rectifying linear units (ReLU).

/pdf/learning-hidden-unit-contributions-for-unsupervised-speaker-5b2x3q6lij.pdf

Learning hidden unit contributions for unsupervised speaker adaptation of neural network acoustic models

Linear hidden transformations for adaptation of hybrid ANN/HMM models

EtherCAT is a real-time Ethernet protocol conceived explicitly for industrial applications. It is characterized by high communication efficiency, which permits control loops to be closed with short cycle times, and is provided with a suitable mechanism, known as distributed clock (DC), that enables synchronized operations to take place across the controlled system. These features can be profitably adopted, for instance, to support motion control applications. In this paper, the performance of the DC mechanism is evaluated by means of a thorough campaign of experimental measurements carried out on a real network setup. A number of factors have been taken into account that can affect accuracy and precision, and their effects studied in depth.

Evaluation of EtherCAT Distributed Clock Performance

This paper presents a front-end consisting of an Artificial Neural Network (ANN) architecture trained with multilingual corpora. The idea is to train an ANN front-end able to integrate the acoustic variations included in databases collected for different languages, through different channels, or even for specific tasks. This ANN front-end produces discriminant features that can be used as observation vectors for language or task dependent recognizers. The approach has been evaluated on three difficult tasks: recognition of non-native speaker sentences, training of a new language with a limited amount of speech data, and training of a model for car environment using a clean microphone corpus of the target language and data collected in car environment in another language.

/pdf/on-the-use-of-a-multilingual-neural-network-front-end-navrnna1gi.pdf

On the Use of a Multilingual Neural Network Front-End

This paper describes reference broadcast infrastructure synchronization (RBIS), a clock synchronization protocol for IEEE 802.11 infrastructure wireless networks. The protocol is especially tailored for industrial and home automation networks, and in many application contexts, it offers several advantages compared with other solutions targeted at similar purposes. RBIS has been conceived to rely on conventional Wi-Fi equipment and, in particular, on unmodified access points. It is based on the master/slave approach and follows the receiver/receiver paradigm. An implementation of RBIS—carried out completely in software and based on timestamps taken at the interrupt handler level—has been developed, which achieves a synchronization error below $3\,{\upmu }{\rm s}$ . Then, a simple distributed hard real-time control application has been set up, which consists in two PCs running real-time application interface for Linux (RTAI) and connected through Wi-Fi. The actuation error, measured on the generation of synchronous pulses, is strictly below $13\,{\upmu}{\rm s}$ .

Implementation and Evaluation of the Reference Broadcast Infrastructure Synchronization Protocol

A technique is proposed for the adaptation of automatic speech recognition systems using Hybrid models combining Artificial Neural Networks with Hidden Markov Models. The application of linear transformations not only to the input features, but also to the outputs of the internal layers is investigated. The motivation is that the outputs of an internal layer represent a projection of the input pattern into a space where it should be easier to learn the classification or transformation expected at the output of the network. A new solution, called Conservative Training, is proposed that compensates for the lack of adaptation samples in certain classes. Supervised adaptation experiments with different corpora and for different adaptation types are described. The results show that the proposed approach always outperforms the use of transformations in the feature space and yields even better results when combined with linear input transformations.

/pdf/adaptation-of-hybrid-ann-hmm-models-using-linear-hidden-580jjwmoh6.pdf

Stefano Scanzio

Papers

Linear hidden transformations for adaptation of hybrid ANN/HMM models

Evaluation of EtherCAT Distributed Clock Performance

On the Use of a Multilingual Neural Network Front-End

Implementation and Evaluation of the Reference Broadcast Infrastructure Synchronization Protocol

Adaptation of Hybrid ANN/HMM Models Using Linear Hidden Transformations and Conservative Training