Fuel cell voltage prediction method based on long short-term memory neural network model

Abstract

The invention discloses a fuel cell voltage prediction method based on a long short-term memory neural network model, and the method comprises the steps: firstly obtaining a plurality of historical detection signals of a fuel cell, carrying out the preprocessing, and decomposing a predicted voltage signal; normalizing the predicted voltage signal and the historical detection signal to obtain a characteristic signal and filtering the characteristic signal; inputting the filtered characteristic signal into the established combined long-short-term memory neural network model, and repeatedly training to obtain a fuel cell voltage prediction model; and finally, carrying out normalization processing and Kalman filtering on the screened characteristic signals after real-time acquisition, inputting the screened characteristic signals into a fuel cell voltage prediction model, and carrying out reverse normalization processing on the obtained result to obtain the predicted voltage of the fuel cell. Particularly, the real-time voltage state of the fuel cell power generation system working for a long period is accurately predicted, so that the fault of the fuel cell can be sensed in advance, and the safe, stable and long-term operation of equipment is ensured.

Description

technical field [0001] The invention relates to the technical field of fuel cells, in particular to a fuel cell voltage prediction method based on a long short-term memory neural network model. Background technique [0002] Although fuel cells have multiple advantages such as cleanliness and high efficiency, their commercialization process is still in its infancy, and its service life is an important factor limiting its large-scale application. In view of the poor durability of fuel cells, prediction and health management (PHM) technology has become a popular method for current fuel cell health assessment and remaining working life prediction. The health status of the fuel cell can be represented by the voltage of the fuel cell. Using the historical detection signal of the fuel cell to predict the voltage of the fuel cell, the fuel cell management and control strategy can be formulated in a targeted manner, which can effectively improve the reliability of the fuel cell power...

AI Technical Summary

Problems solved by technology

Document CN111413626 A regularly measures the polarization curve of the fuel cell and builds a model of polarization demand decay to predict the life of the fuel cell. Polarization curves, making the method face difficulties in practical application

Document CN 109683093 B utilizes data to obtain a model of voltage attenuation obtained by fitting attenuation coefficients under different working conditions. This document obtains voltage attenuation based on statistics and model regression, but it is difficult to classify working conditions in practical applications.

Images