Gene Regulatory Network Inference (GRNI) aims to identify causal relationships among genes using gene expression data, providing insights into regulatory mechanisms. A significant yet often overlooked challenge is selection bias, a process where only cells meeting specific criteria, such as thresholds, survive or are observed, distorting the true joint distribution of and thus biasing GRNI results. Furthermore, influenced by latent confounders, non-coding RNAs, which add complexity GRNI. To...

Recent advances have shown that statistical tests for the rank of cross-covariance matrices play an important role in causal discovery. These include partial correlation as special cases and provide further graphical information about latent variables. Existing typically assume all continuous variables can be perfectly measured, yet, practice many only measured after discretization. For example, psychometric studies, level certain personality dimensions a person being discretized into...

Visual surveillance for autonomous power facility inspection is considered to bethe prominent field of study in the industry. This research completely focuses on either object detection or image fusion which lacks overall consideration. By considering this, a single end-to-end method by incorporating named Fast Detection Fusion Network (FDFNet) proposed this paper output qualitative fused images with results. The parameters FDFnet are greatly reduced sharing feature extraction network...

Multi-source domain adaptation (MSDA) methods aim to transfer knowledge from multiple labeled source domains an unlabeled target domain. Although current achieve joint distribution identifiability by enforcing minimal changes across domains, they often necessitate stringent conditions, such as adequate number of monotonic transformation latent variables, and invariant label distributions. These requirements are challenging satisfy in real-world applications. To mitigate the need for these...

Autonomous robots deal with unexpected scenarios in real environments. Given input images, various visual perception tasks can be performed, e.g., semantic segmentation, depth estimation and normal estimation. These different provide rich information for the whole robotic system. All have their own characteristics while sharing some latent correlations. However, of task predictions may suffer from unreliability dealing complex scenes anomalies. We propose an attention-based failure detection...

Energy storage technology has always been an important lubricant for power systems, especially after wind photovoltaics have connected to the grid on a large scale. equipment played active role in system peaking, frequency regulation, voltage regulation and accident backup. The article analyzes development of different types energy technologies at home abroad, compares several common performance indicators, establishes optimization model that maximizes benefits participating peaking...

Fish behavior analysis for recognizing stress is very important fish welfare and production management in aquaculture. Recent advances have been made based on deep learning. However, most existing methods with top performance rely considerable memory computational resources, which impractical the real-world scenario. In order to overcome limitations of these methods, a new method knowledge distillation proposed identify states schools. The architecture transfers additional inter-class...

Abstract With the rapid development of ranging technology, people’s requirements for technology are more and high. In some special occasions, such as live, high temperature, explosive other inconvenient scenes, importance non-contact is highlighted. As a measurement laser has faster speed, higher accuracy, longer distance stronger anti-interference ability compared with methods. However, at present, accuracy in China not very high, design relatively complex. Aiming above problems, this paper...

At present, the on-site safety monitoring of power is mainly monitored by personnel through whole process surveillance video, but use manual detection method not only a waste time, also prone to missing situation, so that personal staff cannot be guaranteed. In order realize intelligent recognition workers' behavior on job site, dangerous technology based OpenPose was proposed. The extracts key bone information electric workers from video stream images, uses deep neural network human and...

Testing conditional independence has many applications, such as in Bayesian network learning and causal discovery. Different test methods have been proposed. However, existing generally can not work when only discretized observations are available. Specifically, consider $X_1$, $\tilde{X}_2$ $X_3$ observed variables, where is a discretization of latent variables $X_2$. Applying to the lead false conclusion about underlying $X_2$ $X_3$. Motivated by this, we propose specifically designed...

Identifying the causal relations between interested variables plays a pivotal role in representation learning as it provides deep insights into dataset. Identifiability, central theme of this approach, normally hinges on leveraging data from multiple distributions (intervention, distribution shift, time series, etc.). Despite exciting development field, practical but often overlooked problem is: what if those shifts happen sequentially? In contrast, any intelligence possesses capacity to...

Localization of people and machine is important for underground substations. However, conventional global navigation satellite system (GNSS) cannot work in such situation. Other localization technologies as WiFi RFID perform poorly substations due to the influence environment, e.g. blockage walls, congestion space, complexity environment. Moreover, an additional communication also required collect information from deployed anchors, which leads more cost. To solve problem substations,...

A log-ratio signal processing technique in photon beam position monitors (PBPMs) was presented this paper. The main performances (e.g. sensitivity, offset and linearity range) of split PBPM two-wire were analyzed. An inexpensive logarithmic amplifier chip (LOG112) that can measure currents from 0.1nA to 3.5mA used circuits.

Ultra Wideband (UWB) technique is a popular choice for indoor localization due to its immunity interference and high range-estimation accuracy. However, the range-range based requires multiple anchors, e.g., 2D 3 which not efficient limits application in complicated or underground environment. Moreover, if line-of-sight (LoS) path blocked, estimated range will suffer large errors reliable. To overcome these drawbacks, an UWB-based reliable scheme proposed this paper. We first develop basic...

At present, target detection based on deep learning has become a trend. The large model in high accuracy, but with the huge network depth and width, making them difficult to deploy embedded systems limited hardware resources. To address this limitation, Firstly, we build feature extraction yolov5, CBAM (Convolutional Block Attention Module) attention structure are used improve accuracy. Finally, force iterative channel-level pruning guide Sparse training of BatchNormalization (BN) layers....

Abstract To improve the safety and convenience of traditional measurement instruments (such as Total station, Laser tracker), while reducing system cost, a contactless instrument consists single-point laser range finder (LRF) tracking module (a VIO providing 6DoF poses) is proposed in this paper. Its measuring procedure includes three steps: 1) Obtain calibrated relative position parameters from 3D LRF. 2) Determine remote target point positions by fusing value 3) In order to accuracy,...

Abstract At present, the transmission line in field has a large potential safety hazard, and professionals often encounter difficulties patrol inspection distance measurement. In view of current situation safe between equipment power grid construction, this paper proposes long-span measurement method based on aircraft vision, which can automatically achieve high-precision equipment, ensure personal safety, avoid failure other problems. Then ranging technology vision is analyzed, research...

Multi-station integration is an important foundation for the construction of power Internet Things and support achieving low-carbon development. The variety modes discussed, in-depth analysis whole-site monitoring technology under multi-station scenarios, including data characteristic analysis, system architecture design, information model, edge computing processing, main plant service application. pilot project presented, it can provide guidance reference unified whole station station,...

Overhead lines are an important task of power engineering construction, and they often face the problems platform tilt passing through pulley grooves. We have proposed a set intelligent line monitoring system, which is mainly used for detection trolley threading balance state board during high-voltage line. propose to use improved RCF (Richer Convolutional Features Edge Detection) algorithm identify wire on groove determine whether it has out-of-bounds fault. Our practical engineering. The...

In the current overhead line threading process, status monitoring of board and pulley are generally performed by arranging construction personnel on towers section to observe. Usually, when an abnormal state occurs during then contact ground pay-off operator side tensioner. This process will delay a lot time, failure notify in time cause even greater accidents, which seriously affect efficiency. Based lines, this paper proposes set video pose systems for intelligent pulleys boards. The...