Fuel cell system state assessment method based on random matrix eigenspectrum analysis

Abstract

The invention discloses a fuel cell system state evaluation method based on random matrix spectrum analysis, and belongs to the field of fuel cell system state evaluation; the method comprises the following steps: firstly, according to a pressure control subsystem, a temperature control subsystem, an XBO driving interaction subsystem and a voltage monitoring subsystem contained in the fuel cell system, obtaining a multi-physical quantity monitoring signal set to carry out characteristic signal transformation and constructing a fuel cell system running state correlation random matrix; secondly,calculating a sample covariance matrix corresponding to the random matrix, calculating an eigenvalue spectrum distribution of the sample covariance matrix, and constructing a linear spectrum statistical index and a mean spectrum function index based on an average spectrum radius, an information entropy, an likelihood ratio and Wasserstein distance; finally, according to Marchenko-Pastur theorem and the ring theorem, determining the statistical threshold value of the characteristic value spectrum distribution of the fuel cell system, and achieving the effective evaluation of the operating state of the fuel cell system.

Description

technical field [0001] The invention belongs to the field of fuel cell system state evaluation, and in particular relates to a fuel cell system state evaluation method based on random matrix characteristic spectrum analysis. Background technique [0002] As the most promising alternative power source to conventional internal combustion engines, fuel cell systems have been extensively researched and applied in the past decades. Among them, the proton exchange membrane fuel cell (PEMFC) is currently the most promising type of fuel cell for commercial application due to its outstanding advantages such as high power density, low operating temperature, fast dynamic response, and environmental friendliness. With the development of commercial PEMFC integration technology and the continuous improvement of the service life of integrated high-power PEMFC, PEMFC systems have broad application prospects in many fields such as transportation and distributed power generation. [0003] A ...

AI Technical Summary

Problems solved by technology

Among them, the existing model-driven method mainly constructs the residual between monitoring state variables and estimating the system state by analyzing the state space observer, and further realizes the evaluation of the system state through the statistical analysis of the residual, but due to the complexity of the model and the high-order Observability constraints, which usually require simplification or reduction of the model, may lead to unreasonable estimated residuals and cause system false alarms or missed alarms; while existing data-driven methods are mainly based on normal / abnormal states obtained by simulation or actual measurement The following feature samples are used to identify the operating state based on black-box models such as fuzzy logic, artificial neural network, and support vector machine. Hydration status, abnormal peroxygen ratio, etc., did not evaluate the system operating status from an overall perspective

[0006] Since the high-power integrated fuel cell system is a typical closed-loop system, if its potential failure risks and minor state abnormalities cannot be effectively detected and identified during the normal operation of the fuel cell system, these minor abnormal signals will gradually accumulate and be Spread to the whole system, which will further lead to abnormal shutdown or fault damage of the system; and the acquisition of the above-mentioned characteristic samples still depends on the fault prior of expert experience, which is not sensitive to potential risks and minor abnormalities

Therefore, based on existing models and data-driven methods, it is impossible to achieve reliable state assessment of complex fuel cell systems. At present, in actual fuel cell systems, only the independent process variable monitoring function based on simple threshold judgment is configured for the monitoring variable set, and the threshold setting is redundant. The margin is high, and it is impossible to reasonably evaluate the current overall operating status from the perspective of the system as a whole

Images