Abstract:The distributed control system is designed to match the advanced variable cycle engine (VCE) with significant advantages, such as the reduced weight and the low cost by introducing the network. In addition to these advantages, the packet dropout follows and may destroy the performances of the controlled systems. Therefore, except for the engine's strong nonlinearity and uncertainty, the effective control design should also possess the ability to handle the packet dropout. The contribution of this study is not only improving the control quality by creatively designing a large-scale intelligent dynamic estimator but also reinforcing the robustness and extending the application of engine control. A novel intelligent estimation-based large-scale envelope (IEBLE) control design for the VCE with packet dropout is proposed. The new framework consists of two parts: high-accuracy estimation and robust control. The intelligent dynamic estimator is developed by deep learning architecture to compensate for the missing information and supply the essential inputs for the control. With the estimated inputs, a model-based large-scale envelope control design is proposed. This separation between information compensation and control design distinguishes our approach from the centralized control of packet dropout. We formulate the engine's nonlinear dynamic in the full envelope as an optimal Takagi-Sugeno system. The engine dynamic and steady parameters are extracted as features. By the intelligent clustering of the features, the rule number and consequent models of each rule are optimized, which renders the resulting optimal system with the high accuracy of the engine's dynamic and steady performances as well as the minimum rules. Finally, the control gain is scheduled based on the fuzzy inference of the optimized system. Afterwards, this design is applied to a VCE as a demonstration. The comparisons between the proposed control and the H∞ and LQR controllers are conducted in different scenarios to evaluate the control quality and robustness. The experimental results show the superior performances, the effectiveness and robustness of the new control, regardless of the packet dropout, large-scale nonlinearity and uncertainty.

Analytical Redundancy of Variable Cycle Engine Based on Proper Net considering Multiple Input Variables and the Whole Engine’s Degradation

Analytical redundancy of variable cycle engine based on variable-weights neural network

On the Use of Artificial Neural Networks to Model the Performance and Emissions of a Heavy-Duty Natural Gas Spark Ignition Engine

Research on the Analytical Redundancy Method for the Control System of Variable Cycle Engine

Prediction of Efficient Operating Conditions Inside a Heavy-Duty Natural Gas Spark Ignition Engine Using Artificial Neural Networks

Analytical Redundancy Design for Aeroengine Sensor Fault Diagnostics Based on SROS-ELM

Estimation of Small Failure Probability Based on Adaptive Subset Simulation and Deep Neural Network

An Artificial Neural Network Model to Predict Efficiency and Emissions of a Gasoline Engine

A Multi-Branch Redundant Adversarial Net for Intelligent Fault Diagnosis of Multiple Components under Drastically Variable Speeds.

Variable-Geometry Rotating Components Modeling Based on Reference Characteristic Curves for the Variable Cycle Engine

Noise-Resistant Adaptive Gain Recurrent Neural Network for Visual Tracking of Redundant Flexible Endoscope Robot with Time-Varying State Variable Constraints

A Study on Aeroengine Conceptual Design Considering Multi-Mission Performance Reliability

An intelligent estimation-based large-scale envelope control for variable cycle engine with packet dropout

Model of Turbine's Performance Analysis Based on Error Back Propagation Artificial Neural Network

A Study on the Artificial Neural Network Model for an Atkinson Cycle Engine

Matrix Completion Based Aeroengine Analytical Redundancy

Structural-Guided Interaction and Attention-Enhancing Compensation Network for Machine Remaining Useful Life Prediction

Improved Similarity Based Prognostics Method for Turbine Engine Degradation with Degradation Consistency Test

Artificial neural network for sensor failure detection in an automotive engine

High Pressure Common Rail Diesel Engine Modeling for Calibration and Optimization

Neural Network Prediction of Cycle-to-cycle Power Variability in a Spark-Ignited Internal Combustion Engine