In complex scene synthesis, the effective representation of layouts is paramount. This paper introduces LayoutEnc, an advanced approach specifically designed to enhance layout by improving interpretability, robustness, and expressiveness, thereby facilitating more efficient image transformation. Distinct from conventional approaches that homogenize data, LayoutEnc distinctively processes various data modalities, enhancing fidelity interpretability representation. We apply stochastic noise...

To provide efficient pigging operation, a stable and moderate speed of the pig is preferred. This paper describes brake device for controlling in gas pipeline. A model with unit pipeline presented. In model, relevant equation combined flow equations. It assumed that velocity at tail nose equal to pig. Moreover, inlet rate outlet pressure are kept constant during simulation. The method characteristics (MOC) applied solve transient flow, Runge–Kutta used steady equations obtain values initial...

Aiming at the problem that traditional trend prediction method has a low accuracy in predicting state change of pump source system due to high working intensity and bad environment aircraft system, this paper designs GRU network model studies system. Firstly, acquired data are preprocessed, processed reconstructed phase space construct training sample set parameters. The is trained with samples, optimal created by constantly adjusting number hidden layer neurons. used obtain results. In...

Gas system is the key of aero-engine. Because its harsh working environment, it difficult to diagnose effectively. For this reason, paper researches and proposes a method aero-engine gas fault diagnosis based on WOA-SVM. The problems long optimization time genetic algorithm early convergence particle swarm are overcome. Six parameters, T2, T25, T6, N2c25, Alfa2, N1 pneumatic selected. First, WOA used optimize relevant parameters SVM obtain best optimized parameters. Secondly, obtained...

The health of the aircraft's anti-skid braking system is critical to flight aircraft. failure brake will affect efficiency and safety. Therefore, this paper proposes an aircraft fault diagnosis method based on Grey Wolf Optimizer (GWO) Probabilistic Neural Networks (PNN). First, preprocess acquired data five parameters a certain type system: wheel speed, time, pressure, servo valve control current create PNN model. Aiming at shortcomings network smoothing factor selection experience, GWO...

The hydraulic system is one of the key systems on aircraft, with rapid development aviation technology, structure aircraft becoming more and complex, there are many types faults, which make it difficult to perform effective fault diagnosis. Therefore, in order improve accuracy diagnosis system, this paper proposes a method based SSA-DBN. Firstly, adopts status monitoring data certain type Deep Belief Network (DBN) network established. On basis, number nodes hidden layer DBN optimized by...

The prediction of the remaining useful life (RUL) aeroengines is very important in practical engineering applications. Aiming at problem that it difficult to accurately predict aeroengine RUL applications, this paper proposes a method for predicting based on Sequeeze-and-Excitation (SE) and bidirectional long short-term memory network (BiLSTM). First, use SE improve BiLSTM structure create SE-BiLSTM model. Second, engine performance degradation data set provided by NASA train model determine...

Aiming at the problem that accuracy of remaining useful life (RUL) aviation lithium-ion batteries is not high, this paper proposes a method for predicting based on combination Support Vector Regression (SVR) and Markov Chain (MC). First, use NASA battery aging data set, design SVR network model to predict RUL battery. Second, MC correct prediction results model. Finally, proposed SVR-MC tested verified. The show in can achieve performance battery, has good accuracy, application prospects.

Because human visual perception is highly adapted for extracting structural information from a scene, but the existing SSIM index full reference method which needs entire of images. In this paper, we develop reduced and evaluate its performance through set assessment criteria, as well comparison to both EPSNR methods on database images compressed with JPEG JPEG2000.

Generative adversarial networks have been widely used in image synthesis recent years and the quality of generated has greatly improved. However, flexibility to control decouple facial attributes (e.g., eyes, nose, mouth) is still limited. In this paper, we propose a novel approach, called ChildGAN, generate child's according images parents with heredity prior. The main idea disentangle latent space pre-trained generation model precisely face child clear semantics. We use distances between...

Aiming at the problem of difficult fault diagnosis caused by serious noise pollution and weak characteristic information in rolling bearing vibration signal, a method based on combination variational mode decomposition (VMD) deep learning is proposed. First, VMD performed original signal to obtain several Intrinsic Mode Functions (IMF). Then, envelope spectral entropy each IMF can be obtained calculating. The with smallest selected as main analysis IMF. Secondly, stacked auto encoder (SAE)...

For the life of aircraft air refrigerator trend analysis problems, this paper proposed research method that based on a combination SVM and neighborhood rough set attribute reduction. First, to find out core factors which are most significant ones affecting refrigerator, original information data is mined by reduction algorithm. Then place as input vectors for analysis. In certain type application shows that, compared traditional model according experience select vectors, new greatly improved...

Aiming at the problem of complex working mechanism aeroengine gas path system and difficulty in effective fault diagnosis actual work, a new method based on Gray Wolf Optimization Deep Extreme Learning Machine (GWO-DELM) is proposed. Firstly, analyze large amount monitoring data certain type components, sort out health sample sets. Secondly, create DELM model by set circuit system. To reduce influence artificially setting network parameters results, algorithm (GWO) used to optimize...

Aiming at the problems of excessive reliance on manual experience and processing complex signals in traditional gear fault prediction methods, a method based DSAE GRU network is proposed. Firstly, time domain features vibration signal are extracted divided into training samples test samples. Secondly, trained to adaptively extract health index using Then, with The optimal model created by constantly adjusting number hidden layer neurons. Finally, from used verify model. In order...