Gas turbine fault diagnosis expert system method based on GA and L-M combined optimization
A fault diagnosis and gas turbine technology, applied in the direction of gas turbine devices, gas turbine engine testing, reasoning methods, etc., can solve problems such as prone to failures and production accidents, and achieve the effect of ensuring diversity and improving stability
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0045] Fault diagnosis, in short, is to judge the current fault state of the system according to the characteristic parameters of the system. Therefore, the fault diagnosis space of the system can be divided into two parts, namely feature space and state space. The feature space is all possible values of the measurable characteristic parameters of the diagnostic object, and the state space is all possible states of the system (including fault state and normal state). When the diagnosed object is state S, we can determine the feature y, that is, there is a mapping G relationship:
[0046] G:S→y
[0047] Similarly, when the characteristic value of the object to be diagnosed is known, its corresponding state can also be determined, that is, there is a mapping F relationship:
[0048] F: y→S.
[0049] Such asfigure 1 As shown, there is a mapping relationship between the state space and the feature space of the system, and the state space will correspond to the corresponding f...
Embodiment 2
[0098] The number of neurons in the input layer of the BP network is 5, the number of neurons in the hidden layer is 11, and the number of neurons in the output layer is 10. The network error is in the form of a sum of squares, and the target error is set to 10 -3 , the maximum training times is 10000. In the present invention, three algorithms are respectively adopted to train gas turbine gas circuit fault samples, these three algorithms are: GA optimized BP algorithm, L-M optimized BP algorithm and GA and L-M combined optimized BP algorithm. The five neurons in the input layer of the BP network are respectively expressed as: the variation of the rotational speed of the low-pressure compressor (δn 1 %), high pressure compressor speed variation (δn 2 %), change in compression ratio of high pressure compressor (δπ HC %), change in compression ratio of low-pressure compressor (δπ LC %) and fuel flow rate change (δwf%) these five variables. In order to make the diagnostic re...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


