A Study on the Calibration of Steam Flowmeter Readings

In the reading correction of steam instruments, traditional methods face the challenges of nonlinear errors and complex environmental changes. In this paper, a neural network-based calibration method is proposed to construct a PSO-CNN-LSTM model by combining particle swarm optimisation (PSO), convolutional neural network (CNN) and long short-term memory network (LSTM). The hyperparameters of the CNN are optimised using PSO to extract key features in the steam gauge readings, and the features are processed by CNN for deep learning to capture the complex nonlinear relationships in the data. Finally, the LSTM network is used to model the time-series data to correct the reading drift due to time variation. The experimental results show that the proposed model is significantly better than the traditional method in terms of correction accuracy and robustness, and can effectively improve the reading accuracy of steam instrumentation.