Wyniki wyszukiwania - BazTech

1

Shallot Price Forecasting Models: Comparison among Various Techniques

Kasemset Chompoonoot, Phuruan Kanokrot, Opassuwan Takron

Production Engineering Archives

|

2023

|

Vol. 29, Iss. 4

348-355

EN

Shallot is one of several horticultural products exported from Thailand to various countries. Despite an increase in shallot prices over the years, farmers face challenges in price forecasting due to fluctuations and other relevant factors. While different forecasting techniques exist in the literature, there is no universal approach due to varying problems and datasets. This study focuses on predicting shallot prices in Northern Thailand from January 2014 to December 2020. Traditional and machine learning models, including ARIMA, Holt-Winters, LSTM, and ARIMA-LSTM hybrids, are proposed. The LSTM model considers temperature and rainfall as influencing factors. Evaluation metrics include RMSE, MAE, and MAPE. Results indicate that the ARIMA-LSTM hybrid model performs best, with RMSE, MAE, and MAPE values of 10.275 Baht, 8.512 Baht, and 13.618%, respectively. Implementing this hybrid model can provide shallot farmers with advanced price information for informed decision-making regarding cultivation expansion and production management.

2

Analysis of the stator current prediction capabilities in induction motor drive using the LSTM network

Teler Krystian, Orlowska-Kowalska Teresa

Power Electronics and Drives

|

2023

|

Vol. 8 (43)

31-52

EN

In modern areas of knowledge related to electric drive automation, there is often a need to predict the state variables of the drive system state variables, such as phase current and voltage, electromagnetic torque, stator and rotor flux, and others. This need arises mainly from the use of predictive control algorithms but also from the need to monitor the state of the drive to diagnose possible faults that have not yet occurred but may occur in the future. This paper presents a method for predicting stator phase current signals using a network composed of long-short-term memory units, allowing the simultaneous prediction of two signals. The developed network was trained on a set of current signals generated by software. Its operation was verified by simulation tests in a direct rotor flux-oriented control (DRFOC) structure for an induction motor drive in the Matlab/Simulink environment. An important property of this method is the possibility of obtaining a filtering action on the output of the network, whose intensity can be controlled by varying the sampling frequency of the training signals.

3

Using an LSTM network to monitor industrial reactors using electrical capacitance and impedance tomography - a hybrid approach

Kłosowski Grzegorz, Rymarczyk Tomasz, Niderla Konrad, Kulisz Monika, Skowron Łukasz, Soleimani Manuchehr

Eksploatacja i Niezawodność

|

2023

|

Vol. 25, no. 1

art. no. 11

EN

The article presents a new concept for monitoring industrial tank reactors. The presented concept allows for faster and more reliable monitoring of industrial processes, which increases their reliability and reduces operating costs. The innovative method is based on electrical tomography. At the same time, it is non-invasive and enables the imaging of phase changes inside tanks filled with liquid. In particular, the hybrid tomograph can detect gas bubbles and crystals formed during industrial processes. The main novelty of the described solution is the simultaneous use of two types of electrical tomography: impedance and capacitance. Another novelty is the use of the LSTM network to solve the tomographic inverse problem. It was made possible by taking the measurement vector as a data sequence. Research has shown that the proposed hybrid solution and the LSTM algorithm work better than separate systems based on impedance or capacitance tomography.

4

Wykrywanie anomalii w zachowaniu robota usługowego z wykorzystaniem sieci LSTM

Lis Kamila, Szynkiewicz Wojciech, Niewiadomska-Szynkiewicz Ewa, Ciecierski Konrad

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2022

|

z. 197, t. 1

211-222

PL

Praca dotyczy zagadnień cyberbezpieczeństwa mobilnych robotów usługowych z rozproszoną architekturą sterowania. Prezentowane jest autorskie podejście do detekcji anomalii w działaniu robota na podstawie odczytów z czujników, przy założeniu, że ataki skutkują niezgodnym z zakładanym zachowaniem robota. Opracowany system wykrywania włamań RIDS wykorzystuje głębokie, rekurencyjne sieci neuronowe. W pracy przedstawiona jest architektura sieci, jej parametry oraz atrybuty, na podstawie których identyfikowane są potencjalne ataki. Rozwiązanie zostało zweryfikowane w środowisku laboratoryjnym z wykorzystaniem robota TIAGo firmy PAL Robotics. Wyniki badań potwierdzają, że proponowany system może skutecznie wspierać proces wykrywania zagrożeń komputerowych wpływających negatywnie na funkcjonowanie systemów robotycznych.

EN

The paper addresses cybersecurity issues of mobile service robots with distributed control architecture. A novel robot intrusion detection system (RIDS) that employs deep recurrent neural networks to detect anomalies in robot performance based on sensor readings, under the assumption that attacks result in incompatible robot behavior is described. The performance of the RIDS system was evaluated in a laboratory environment using the TIAGo robot from PAL Robotics. The presented experimental results confirm that RIDS can effectively support the process of detecting computer threats that negatively affect the service robot operating.

5

Speech emotion recognition using wavelet packet reconstruction with attention-based deep recurrent neutral networks

Meng Hao, Yan Tianhao, Wei Hongwei, Ji Xun

Bulletin of the Polish Academy of Sciences. Technical Sciences

|

2021

|

Vol. 69, nr 1

art. no. e136300

EN

Speech emotion recognition (SER) is a complicated and challenging task in the human-computer interaction because it is difficult to find the best feature set to discriminate the emotional state entirely. We always used the FFT to handle the raw signal in the process of extracting the low-level description features, such as short-time energy, fundamental frequency, formant, MFCC (mel frequency cepstral coefficient) and so on. However, these features are built on the domain of frequency and ignore the information from temporal domain. In this paper, we propose a novel framework that utilizes multi-layers wavelet sequence set from wavelet packet reconstruction (WPR) and conventional feature set to constitute mixed feature set for achieving the emotional recognition with recurrent neural networks (RNN) based on the attention mechanism. In addition, the silent frames have a disadvantageous effect on SER, so we adopt voice activity detection of autocorrelation function to eliminate the emotional irrelevant frames. We show that the application of proposed algorithm significantly outperforms traditional features set in the prediction of spontaneous emotional states on the IEMOCAP corpus and EMODB database respectively, and we achieve better classification for both speaker-independent and speaker-dependent experiment. It is noteworthy that we acquire 62.52% and 77.57% accuracy results with speaker-independent (SI) performance, 66.90% and 82.26% accuracy results with speaker-dependent (SD) experiment in final.

6

Application of Recurrent Neural Networks for User Verification based on Keystroke Dynamics

Kobojek P., Saeed K.

Journal of Telecommunications and Information Technology

|

2016

|

nr 3

80--90

EN

Keystroke dynamics is one of the biometrics techniques that can be used for the verification of a human being. This work briefly introduces the history of biometrics and the state of the art in keystroke dynamics. Moreover, it presents an algorithm for human verification based on these data. In order to achieve that, authors’ training and test sets were prepared and a reference dataset was used. The described algorithm is a classifier based on recurrent neural networks (LSTMand GRU). High accuracy without false positive errors as well as high scalability in terms of user count were chosen as goals. Some attempts were made to mitigate natural problems of the algorithm (e.g. generating artificial data). Experiments were performed with different network architectures. Authors assumed that keystroke dynamics data have sequence nature, which influenced their choice of classifier. They have achieved satisfying results, especially when it comes to false positive free setting.