Convolutional neural networks (CNNs) with log-mel audio representation and CNN-based end-to-end learning have both been used for environmental event sound recognition (ESC). However, features can be complemented by learned from the raw waveform an effective fusion method. In this paper, we first propose a novel stacked CNN model multiple convolutional layers of decreasing filter sizes to improve performance models either feature input or input. These two are then combined using...

Currently gear fault diagnosis is mainly based on vibration signals with a few studies acoustic signal analysis. However, acquisition limited by its contact measuring while traditional acoustic-based relies heavily prior knowledge of processing techniques and diagnostic expertise. In this paper, novel deep learning-based method proposed sound By establishing an end-to-end convolutional neural network (CNN), the time frequency domain can be fed into model as raw without feature engineering....

Traffic congestion prediction is critical for implementing intelligent transportation systems improving the efficiency and capacity of networks. However, despite its importance, traffic severely less investigated compared to flow prediction, which partially due severe lack large-scale high-quality data advanced algorithms. This paper proposes an accessible general workflow acquire create datasets based on image analysis. With this we a dataset named Seattle Area Congestion Status (SATCS) map...

Many text mining tasks such as retrieval, summarization, and comparisons depend on the extraction of representative keywords from main text. Most existing keyword algorithms are based discrete bag-of-words type word representation In this paper, we propose a patent algorithm (PKEA) distributed Skip-gram model for classification. We also develop set quantitative performance measures evaluation information gain cross-validation, Support Vector Machine (SVM) classification, which valuable when...

The gear fault signal under different working conditions is non-linear and non-stationary, which makes it difficult to distinguish faulty signals from normal signals. Currently, diagnosis mainly based on vibration However, acquisition limited by its requirement for contact measurement, while analysis methods relies heavily diagnostic expertise prior knowledge of processing technology. To solve this problem, a novel acoustic-based (ABD) method multi-scale convolutional learning structure...

With the digital transformation of manufacturing industry, data monitoring and collecting in process become essential. Pointer meter reading recognition (PMRR) is a key element throughout process. However, existing PMRR methods have low accuracy insufficient robustness due to issues such as blur, uneven illumination, tilt, complex backgrounds images. To address these challenges, we propose an end-to-end method based on decoupled circle head detection algorithm (YOLOX-DC) Unet-like pure...

Instrument reading detection in industrial scenarios poses significant challenges due to contour distortion caused by perspective transformation the instrument images. However, existing methods fail accurately read display automatically incorrect labeling of target box vertices, which arises from vertex entanglement problem. To address these challenges, a novel Quadrilateral Contour Disentangled Detection Network (QCDNet) is proposed this paper, utilizes quadrilateral disentanglement idea....

Objective: To investigate the interaction of workplace noise, body mass index (BMI) and systemic inflammatory response on hypertension. Methods: In January 2019, 1124 male workers from an automobile factory in Wuhan were selected by cluster random sampling method. The study population was divided into normal weight group (BMI<24 kg/m(2)) overweight (BMI≥24 according to BMI, followed up for 3 years. occupational health examination carried out every year, blood routine biochemical indexes...

Acoustic-based diagnosis (ABD) is a promising method for machinery fault detection due to its ability overcome the limitation of vibration measurement through non-contact by air-couple. However, most ABD approaches are not widely used in real-industrial scenario strong and highly non-stationary background noise interference. To address shortcoming, novel based on recursive denoising learning (RDL) proposed this article. In method, new multi-stage attention mechanism designed as fundament RDL...

Abstract We have designed, built and tested a switched‐capacitor circuit for generating pseudo‐random signals. the is an electronic implementation of one‐dimensional discrete map, which has been fully understood mathematically. time series extremely rich dynamics, e.g. aperiodicity, non‐asymptoticity, ergodicity fractal dimension. That is, indeed chaotic. Furthermore, general method designing chaotic circuits with prescribed random properties proposed. Numerous examples are given to...

With the increasing concern on energy crisis, coordination of multiple sources and low-carbon economic operation integrated system (IES) have drawn more attention in recent years. In IES, accurate effective multi-energy load forecasting becomes a research hotspot, especially using high-performance data mining machine learning algorithms. However, due to huge difference utilization between IES traditional systems, is difficult complex. fact, not only related external factors such as...

In order to improve the efficiency of transportation networks, it is critical forecast traffic congestion. Large-scale congestion data have become available and accessible, yet they need be properly represented in avoid overfitting, reduce requirements computational resources, utilized effectively by various methodologies models. Inspired pooling operations deep learning, we propose a representation framework for urban road networks. This consists grid-based partition networks operation...

The health condition of gears has been a topic study in the past few decades due to importance for transmission system. In recent years, some studies have used acoustic signals gear diagnosis, which can overcome limitation vibration through noncontact measurement by air-couple. Although many acoustic-based diagnosis (ABD) methods achieved good performance stable working conditions, these suffer from effectiveness loss as change load actual industry causes domain shift problem. To above...

In cognitive radio networks, unlicensed users need to learn from environmental changes. This is a process that can be done in cooperative or non-cooperative manner. Due the competition for channel utilization among users, approach may lead overcrowding available channels. paper about fuzzy logic based decision making algorithm competition-based selection. The underlying criterion integrates both statistics of licensed users' occupancy and level users. By using such an algorithm, user...

This work motivates and details the concept of QoE-aware sustainable throughput in area video streaming. Sustainable serves as a mean to compare streaming solutions terms Quality Experience (QoE) energy efficiency (EE). It builds upon QoE Provisioning-Delivery Hysteresis (PDH) denotes maximal at which deteriorations can be kept below quantifiable level, turn allows EE different on QoE-fair grounds. In this work, we particularly focus delivery problems stemming from outage-prone links, they...

Opportunistic spectrum access (OSA) is a technology that allows unlicensed users to holes and provide so efficient use of radio resources. Most studies done on OSA focus the situation when user performs handoff only within single cognitive network (so-called intra-handoff). In this paper, we consider (licensed or unlicensed) be able do inter-handoff among different cells as well. The priority users. By considering multiple being in steady-state showing identical statistics, arrival rates are...

As a practical application of Optical Character Recognition (OCR) for the digital situation, instrument recognition is significant to achieve automatic information management in real-industrial scenarios. However, different from normal task such as license plate recognition, CAPTCHA and handwritten digit multi-type instruments faces greater challenges due reading strings are variable-length with fonts, spacing aspect ratios. In order overcome this shortcoming, we propose novel short-memory...

The induction of classification decision tree is an important algorithm for data mining now. support vector machine technology and the have combined into one multi-class classifier so as to solve problems. In this paper, SVM extended non-linear by using kernel functions a new method NSVM proposed based on traditional tree. Classification experiments prove effective.