Engineering data are often highly nonlinear and contain high-frequency noise, so the Levenberg–Marquardt (LM) algorithm may not converge when a neural network optimized by is trained with engineering data. In this work, we analyzed reasons for LM network’s poor convergence commonly associated algorithm. Specifically, effects of different activation functions such as Sigmoid, Tanh, Rectified Linear Unit (RELU) Parametric (PRLU) were evaluated on general performance networks, special values...

Monitoring gas turbines' health, in particular, detecting abnormal behaviors time, is critical ensuring turbine operating safety and preventing costly unplanned maintenance. One most popular anomaly detection method to obtain a classification-prediction model by training classifier using the real-life data of turbine. The excellent ability this attributed enough annotated samples, especially samples. Nevertheless, monitoring data, normal far more than even no data. Advanced technologies that...

Abstract Traditionally, most simulation approaches for pistons are conducted based on thermal-structure coupling, where the cooling boundary conditions determined by empirical formulas. However, these often suffer some decreases in accuracy of results. Aiming at problem, this paper conducts a thermal-fluid-structure coupling method piston large marine diesel engine, which employs computational fluid dynamics to identify conditions. Firstly, solid and models established. Then, process chamber...

At present, the aero engines fault diagnosis method based on convolutional neural network can only deal with features of local region and it was difficult to effectively capture long-range dependencies which leads problem low accuracy. Attention mechanism solve this some extent now. However, typical attention blocks don't have sufficient interaction for input in a certain dimension. To it, we propose multiple attentional mechanism, is implemented form block. In block, introduce channel that...

The prediction of Exhaust Gas Temperature Margin (EGTM) after washing aero-engines can provide a theoretical basis for airlines not only to evaluate the energy-saving effect and emission reduction, but also formulate reasonable maintenance plans. However, EGTM encounters step changes aeroengines, while, in traditional models, persistence tendency exists between results previous data, resulting low accuracy prediction. In order solve problem, this paper develops parameters model based on...

The performance prediction of gas turbines is one the key factors to improve flight safety and reduce unnecessary maintenance costs. Although many methods have already been developed, accuracy fails meet needs practical application owing several seasonal effects change points in time series. To address these issues, this paper proposes a method for using Prophet model. On hand, model introduces varying degradation rate into trend item, which can better fit points. other Fourier Moreover,...

The aging of aero-engine gas path components will cause the increase exhaust temperature, lower thermal efficiency. directly measured temperature was height or not can been determined as external environment very different on aircraft cruise conditions. must be convert into a stable environment, it called correction. Analysis deviation results engine manufacturers performance monitoring software carried out, according to similarity in fluid mechanics theory, new correction model proposed by...

Predicting the remaining useful life of turbofan engines can provide early warning potential failures and improve maintenance schedules. When using data-driven methods to estimate systems, extremely large amounts run-to-failure labels are needed. In this paper, we propose a novel label correction method based on unsupervised reconstruction. Aiming modify piece-wise linear degradation model, reconstruction error was used distinguish different engine operating phases, initial failure time...

In the industrial knowledge graph, relations between entities has important physical significance. some cases, determines product quality or fault prediction accuracy. Therefore, relation is particularly important. The existing translation models are poor in prediction, because current method of generating incorrect triples cannot generate with relations. this paper develops an triple generation by combining uniform distribution and Bernoulli to improve Then UBTransH model developed TransH...

The phased array radar radiation unit is a precision product in the military industry. In recent years, with advancement of automation technology, manufacturing technology has been greatly developed, but design relatively lagging. After investigation, it found that lot data information generated during production radar, which can be used to guide radar. At present, this not effectively used. For reason, paper proposes secondary optimization algorithm based on hierarchical Gaussian mixture...

Aeroengine configuration is complex, and there are many failure reasons mechanisms of components, which bring difficulty to aeroengine fault diagnosis. Using machine learning method analyze historical operation maintenance data mining association rules can provide effective support for In this paper, an correlation analysis based on extension FP-Growth algorithm proposed by combining superiority evaluation with algorithm. Aiming at the problems low value weak relevance data, paper uses mine...