CNN-based Compressor Mass Flow Estimator in Industrial Aircraft Vapor Cycle System

TL;DR

This paper tackles the challenge of estimating compressor mass flow $\dot{m}$ in aircraft Vapor Cycle Systems where physical sensors are costly and vibration-sensitive. It proposes a causal Conv1D CNN-based virtual sensor that predicts $\dot{m}$ from available measurements and compares it to a Polynomial Regression baseline, using a real industrial dataset from LIEBHERR Aerospace. A semi-automatic time-series segmentation method is developed to compute Engineering Performance (EP) metrics on real data, enabling robust evaluation of dynamic behavior. The results show that the CNN substantially improves standard MSE and EP metrics, supporting practical deployment for real-time monitoring and control in aerospace cooling loops; future work will integrate physical knowledge into ML models to further enhance performance.

Abstract

In Vapor Cycle Systems, the mass flow sensor playsa key role for different monitoring and control purposes. However,physical sensors can be inaccurate, heavy, cumbersome, expensive orhighly sensitive to vibrations, which is especially problematic whenembedded into an aircraft. The conception of a virtual sensor, basedon other standard sensors, is a good alternative. This paper has twomain objectives. Firstly, a data-driven model using a ConvolutionalNeural Network is proposed to estimate the mass flow of thecompressor. We show that it significantly outperforms the standardPolynomial Regression model (thermodynamic maps), in terms of thestandard MSE metric and Engineer Performance metrics. Secondly,a semi-automatic segmentation method is proposed to compute theEngineer Performance metrics for real datasets, as the standard MSEmetric may pose risks in analyzing the dynamic behavior of VaporCycle Systems.

Figures (8)

• Figure 1: Vapor Cycle System (VCS)
• Figure 2: Vapor Cycle on Pressure-Enthalpy diagram 5991005
• Figure 3: Scaled measures
• Figure 4: Feedforward CNN Conv1D (kernel size, number of output channels)
• Figure 5: Templates of patterns