The marriage of cloud and software defined network (SDN) can work out the challenge which exist in typical platform such as private isolation user, flow control. But SDN based cloud, controller manages whole system is vulnerable to distributed-denial-of-service (DDoS) attack, causing paralysis entire network. It critical for be quick-speed, low false positive, high precise against attack detection. In this paper, we use extreme gradient boosting (XGBoost), detection method cloud. addition,...

Driving style recognition plays a key role in ensuring driving safety and improving vehicle traffic efficiency. With the development of sensing technology, data-driven methods are more widely uesd to recognize style. However, adequately labeling data is difficult for supervised learning methods, while classification accuracy not sufficiently approved unsupervised methods. This paper proposes new method based on Tri-CatBoost, which takes CatBoost as base classifier effectively utilizes...

The turbofan engine is a crucial component of the aircraft. In order to provide an appropriate maintenance for improve reliability system, it necessary estimate remaining useful life (RUL) engine. this paper, data-based RUL prediction method proposed using Light Gradient Boosting Machine (LightGBM). To capture more degradation information, time window row data and runtime are used as inputs after normalization. LightGBM works very well with these high-dimensional model easy interpret....

It is crucial to predict the remaining useful life (RUL) of aircraft engines accurately and timely for operation safety appropriate maintenance decision. The key issue how efficiently mine internal relation hiding in historical time series monitoring data with high dimension features. In this paper, a data-driven prediction method proposed by combining window (TW) extreme learning machine (ELM). First, based on specific properties data, sliding introduced sample obtain input vector. Then,...

The remaining useful life estimation has been widely studied for engineering systems. A system commonly works under varying operating conditions, which may affect the degradation trajectory differently and consequently reduce accuracy of estimation. In this paper, we propose CNN-XGB with extended time window to tackle issue. Firstly, is created by feature extension processing in data preprocessing. extension, multiple features are extracted an improved differential method, these appended raw...

For periodical beacon broadcasting in cellular vehicle-to-everything (C-V2X) networks, a distributed reservation media access control (MAC) protocol, the sensing-based semi-persistent scheduling (SPS), is adopted. However, how to quantify communication reliability and latency an open issue, which critical for low-latency high-reliability services. In this paper, analytical model SPS presented, based on impacts of rate, range settings system configuration collision probability delay outage...

Remaining useful life (RUL) estimation is expected to provide appropriate maintenance for components or systems in industry improve the reliability of systems. Most data-based methods are limited a single model, which susceptible various factors like environmental variability and diversity operating conditions. In this paper, we propose an optimal stacking ensemble method combining different learning algorithms as meta-learners mitigate impact multi-operating The selection follows...

The health stage division problem has been attracting interest for its potential role in Prognostics and Health Management (PHM). In traditional division, multiple indicators (HIs) are extracted from original data, then one suitable HI is constructed. According to the trend of HI, whole degradation process divided into different stages based on change points. points transition between stages. But difficult construct just special applications. Therefore how consider HIs simultaneously a big...

Supplier selection is a critical multi-criteria decision making problem for supply chain management. With the emergence of big data, there an urgent need data-driven methods. A hybrid DEA-Adaboost model proposed to meet challenge. The split into DEA and learner. fuzzy multi-objective used build expert database, which contains appropriate inappropriate suppliers. learner trained by Adaboost from database. Thus, derived are combined as reduce time consumption computational complexities...

We consider the problem of efficiently using smartphone users to augment stationary infrastructure sensors for better situation awareness in smart cities. envision a dynamic sensing platform that intelligently assigns tasks volunteered users, order answer queries by performing at specific locations may not be covered in-situ sensors. mathematically formulate into an integer programming minimize overall energy consumption while satisfying required query accuracy. present optimal algorithm...

Recent years has witnessed a boom in fog-assisted crowdsensing, which exploits powerful sensing capabilities of various mobile devices or vehicles distributed large-scale areas to efficiently gather information and make better decisions. However, the crowdsensing system is totally open, provides opportunity for malicious individuals organizations launch different attacks. In order cope with security threats from participants, blockchain-based framework proposed, helps check authentication...

Nowadays, spam is pervasive in the mailbox, and not only caused a waste of network resources, but also brings lot trouble to people's daily life. How filter quickly accurately challenge we are facing. For handling this challenge, paper proposes fast content-based filtering algorithm with fuzzy-SVM k-means. First, K-means clustering used compress data retain most effective information. Then, fuzzy support vector machine train classification model, order deal uncertain factors better. The...

An accurate state-of-charge (SOC) estimation for a lithium-ion battery is highly dependent on the knowledge of aging, which usually costly or not available through online measurements. In this paper, novel aging-aware features can simultaneously characterize aging and SOC are extracted from discharging process. Then, extreme gradient boosting (XGBoost) algorithm combined stage division applied to acquire nonlinear relationship model between proposed offline training. The method does require...

Abstract To effectively tackle the intricate and dynamic challenges encountered in proton exchange membrane fuel cells (PEMFCs), this paper introduces a model-free reinforcement learning approach to address its water management issue. Recognizing limitations of conventional methods such as Q-learning handling continuous actions nonlinearity inherent PEMFCs management, we propose prioritized deep deterministic policy gradient (DDPG) method. This method, rooted Actor-Critic framework,...

In transportation fields, safety, efficiency, and reliability of engines are primary concerns. Remaining useful life (RUL) estimation technology is used to assess the current health status make effective maintenance plans for engines. How estimate RUL accurately increasing safety systems a challenge issue. However, existing methods focus on single size sequence. To address it, this paper proposes remaining prediction model based hybrid long-short sequences For long sequence, short-term...

Home energy management plays a key role in demand response for residential customers to reduce the total cost via scheduling household loads consumption. However, excessive consumption by will bring great challenge stability of grid. To address challenge, day-ahead multi-agent reinforcement learning method is proposed home under peak power-limiting. We first formulate minimization problem as Markov game, and then novel algorithm based on Mutil-agent Deep Deterministic Policy Gradient...

Imbalanced data is ubiquitous and brings much difficulty for classification. In this paper, we propose an ensemble strategy to address binary classification imbalanced problem by treatment of difficult samples or hard classifying samples. The within a two layers framework, which combines the advantages resampling method classifier. first layer, novel proposed increase weight in dataset. Then, efficient utilized second applies extreme gradient boosting classifier combined with train model....

Electric vehicle charging stations (EVCS) play a vital role in providing support to EV users. In order facilitate users terms of speed, two different modes (L2 and L3) are currently available at public stations. L3 mode provides quick with higher power, whereas L2 offers moderate speed low power. The integration an EVCS into the power grid requires coordinated strategies reduce electricity bill for profitable operation. However, effective utilization multi-mode strategy serve maximum number...