The forecast of electricity consumption plays an essential role in marketing management. In this study, a random forest (RF) model coupled with ensemble empirical mode decomposition (EEMD) named ...EEMD-RF is presented for forecasting the daily electricity consumption of general enterprises. The candidate data is first decomposed into several intrinsic mode functions (IMFs) by the EEMD. Through fast Fourier transformation, the features in each IMF are extracted in the time-frequency domain, then simulated and predicted by the RF model. Finally, the results of each IMF are integrated into the overall trend of the daily electricity consumption for those enterprises. The proposed method was applied to two enterprises located in the Jiangsu High-Tech Zone, and the period of collected data was from January 1, 2015 to May 3, 2016. To show the applicability and superiority of the EEMD-RF approach, two basic models (a back-propagation neural network (BPNN) and least squares support vector regression (LSSVM) and five model experiments (EEMD-BPNN, EEMD-LSSVM, RF, BPNN and LSSVM) were selected for comparison. Among these approaches, the proposed model exhibited the best forecast performance in terms of mean absolute error, mean absolute percentage error, and root-mean-square error.
•A random forest model coupled with ensemble empirical mode decomposition (EEMD-RF) is proposed.•The EEMD is applied for extracting complex features of different modes.•The RF is applied for modeling the changes of different modes.•The EEMD-RF has high accuracy in enterprise electricity consumption forecasting.
In recent years, convolutional neural networks (CNNs) have become widely adopted for computer vision tasks. Field-programmable gate arrays (FPGAs) have been adequately explored as a promising ...hardware accelerator for CNNs due to its high performance, energy efficiency, and reconfigurability. However, prior FPGA solutions based on the conventional convolutional algorithm is often bounded by the computational capability of FPGAs (e.g., the number of DSPs). To address this problem, the feature maps are transformed to a special domain using fast algorithms to reduce the arithmetic complexity. Winograd and fast Fourier transformation (FFT), as fast algorithm representatives, first transform input data and filter to Winograd or frequency domain, then perform element-wise multiplication, and apply inverse transformation to get the final output. In this paper, we propose a novel architecture for implementing fast algorithms on FPGAs. Our design employs line buffer structure to effectively reuse the feature map data among different tiles. We also effectively pipeline the Winograd/FFT processing element (PE) engine and initiate multiple PEs through parallelization. Meanwhile, there exists a complex design space to explore. We propose an analytical model to predict the resource usage and the performance. Then, we use the model to guide a fast design space exploration. Experiments using the state-of-the-art CNNs demonstrate the best performance and energy efficiency on FPGAs. We achieve 854.6 and 2479.6 GOP/s for AlexNet and VGG16 on Xilinx ZCU102 platform using Winograd. We achieve 130.4 GOP/s for Resnet using Winograd and 201.1 GOP/s for YOLO using FFT on Xilinx ZC706 platform.
This paper works on the detection of physical random access channel (NPRACH) in Narrowband Internet of Things (NB-IoT) system. The frequency hopping preamble design and increasing number of IoT ...terminals lead to inter-cell interference among different cells, resulting in inevitable increase of false alarm rate. Due to the ambiguity between preamble and interference, it is a great challenge for NPRACH detection methods to achieve low false alarm rate when having strong interference. In this paper, we analyze the difference between preamble and interference in the propagation environments of NPRACH signals in the 2-dimensional Fast Fourier Transformation (2-D FFT) domain. Then we propose a deep learning-based NPRACH detection method, dubbed Mask Assisted Anti-Interference Universal Detection Scheme (MIUS), in the 2-D FFT domain for preamble detection with inter-cell interference in different repetition cases. In the proposed MIUS, the Mask-ResNet Block is designed as a building block to extract features distinguishing the preamble and interference based on masking operations. Our proposed MIUS utilizes the Mask-ResNet Block in a separate manner to detect the preambles in sequential repetitions across different repetition cases. Simulation results show that MIUS can simultaneously maintain the low false alarm rate and achieve high detection accuracy in low Signal to Interference and Noise Ratio (SINR) regime in all repetition cases.
We have investigated the sensing characteristics of ethanol, methanol, acetone, and 2-propanol using wet chemical synthesized copper oxide (CuO) thin films deposited on fused quartz substrates. As ...compared to the volatile organic component (VOC) sensing characteristics of CuO films reported in recent literatures, our thin film sensors offer relatively higher response (%), lower optimized temperature (corresponds to highest response%), better stability, and faster response time. In order to address the cross-sensitivity towards these VOC sensing we had performed fast Fourier transformation (FFT) analyses of the resistance transients. The resultant data matrices extracted from these FFT analyses were used as input parameter in a linear unsupervised principal component analysis (PCA) pattern recognition technique. We have demonstrated that FFT combined with PCA is an excellent tool for differentiating these reducing gases.
Display omitted
•CuO thin film was grown on quartz substrate using sol gel route.•Gas sensing properties of CuO thin film was performed using a quasi-static system.•Excellent response is obtained towards ethanol, methanol, acetone and 2-propanol.•High response (%) and low response time(s) are achieved at optimized temperature.•FFT and PCA were studied to address the cross sensitivity.
•A novel string stability criterion by dividing a mixed vehicular platoon into multiple interconnected sub-systems.•A practical string stability criterion for a mixed vehicular platoon over ...predominant acceleration frequency boundaries.•A new CAV control strategy incorporating human-driven vehicles’ car following characteristics.•A new CAV control strategy to stabilize mixed vehicular platoons with guaranteed feasibility.
This paper presents a car-following control strategy of connected automated vehicles (CAVs) to stabilize a mixed vehicular platoon consisting of CAVs and human-driven vehicles. This study first establishes a string stability criterion for a mixed vehicular platoon. Specifically, a mixed vehicular platoon is decomposed into “subsystems” that are all possible sequential subsets of the platoon. String stability is then defined as the “head-to-tail” string stability for all subsystems: the magnitude of a disturbance is not amplified from the first vehicle to the last vehicle of each subsystem. Based on this definition, distributed frequency-domain-based CAV control is proposed to increase the number of head-to-tail string stable subsystems and consequently dampen stop-and-go disturbances drastically. Specifically, an H-infinity control problem is formulated, where the maximum disturbance “damping ratios” in each subsystem is minimized within the predominant acceleration frequency boundaries of human-driven vehicles. Simulation experiments, embedded with real human-driven vehicle trajectories, were conducted, and results show that the proposed control can effectively dampen stop-and-go disturbances.
As a fundamental concept in information theory, mutual information (MI) has been commonly applied to quantify association between random vectors. Most existing nonparametric estimators of MI have ...unstable statistical performance since they involve parameter tuning. We develop a consistent and powerful estimator, called fastMI, that does not incur any parameter tuning. Based on a copula formulation, fastMI estimates MI by leveraging Fast Fourier transform-based estimation of the underlying density. Extensive simulation studies reveal that fastMI outperforms state-of-the-art estimators with improved estimation accuracy and reduced run time for large data sets. fastMI provides a powerful test for independence that exhibits satisfactory type I error control. Anticipating that it will be a powerful tool in estimating mutual information in a broad range of data, we develop an R package fastMI for broader dissemination.
With the development of the brain-computer interface (BCI) community, motor imagery-based BCI system using electroencephalogram (EEG) has attracted increasing attention because of its portability and ...low cost. Concerning the multi-channel EEG, the frequency component is one of the most critical features. However, insufficient extraction hinders the development and application of MI-BCIs. To deeply mine the frequency information, we proposed a method called tensor-based frequency feature combination (TFFC). It combined tensor-to-vector projection (TVP), fast fourier transform (FFT), common spatial pattern (CSP) and feature fusion to construct a new feature set. With two datasets, we used different classifiers to compare TFFC with the state-of-the-art feature extraction methods. The experimental results showed that our proposed TFFC could robustly improve the classification accuracy of about 5% (<inline-formula> <tex-math notation="LaTeX">{p} < 0.01 </tex-math></inline-formula>). Moreover, visualization analysis implied that the TFFC was a generalization of CSP and Filter Bank CSP (FBCSP). Also, a complementarity between weighted narrowband features (wNBFs) and broadband features (BBFs) was observed from the averaged fusion ratio. This article certificates the importance of frequency information in the MI-BCI system and provides a new direction for designing a feature set of MI-EEG.
Discrete Fourier transform (DFT) is an important transformation technique in signal processing tasks. Due to its ultrahigh computing complexity as O(N 2 ), N-point DFT is usually implemented in the ...format of fast Fourier transformation (FFT) with the complexity of O(N log N). Despite this significant reduction in complexity, the hardware cost of the multiplication-intensive N-point FFT is still very prohibitive, particularly for many large-scale applications that require large N. This brief, for the first time, proposes high-accuracy low-complexity scalingfree stochastic DFT/FFT designs. With the use of the stochastic computing technique, the hardware complexity of the DFT/FFT designs is significantly reduced. More importantly, this brief presents the scaling-free stochastic adder and the random number generator sharing scheme, which enable a significant reduction in accuracy loss and hardware cost. Analysis results show that the proposed stochastic DFT/FFT designs achieve much better hardware performance and accuracy performance than state-of-the-art stochastic design.