Offshore doubly fed wind power generator fault judging method based on GRA-LSTM-stacking model

Abstract

The invention relates to an offshore doubly fed wind power generator fault judging method based on a GRA-LSTM-stacking model. The method includes the following steps that firstly, a GRA is used for analyzing the SCADA state variable, and the state variable relevant to the temperature of a generator is screened out to serve as the input of an LSTM network; secondly, an LSTM is used for predicting the temperature of the generator under the normal state, and a predicated value is obtained; thirdly, the residual error absolute value between the practical value and the predicated value is calculated so that a statistical method can be used for setting an alarming threshold value, the early failure of the generator is distinguished accordingly, and a failure sample is extracted; and fourthly, astacking fusion algorithm is used for conducting data processing on the extracted failure sample, and a final accurate judgment result is output. Compared with the prior art, the offshore doubly fed wind power generator fault judging method has the beneficial effects of being accurate in failure judgment, high in universality, simple in failure sample obtaining and the like.

Description

technical field [0001] The invention relates to the field of fault diagnosis of offshore wind power generators, in particular to a fault diagnosis method for offshore doubly-fed wind power generators based on the GRA-LSTM-stacking model. Background technique [0002] Offshore wind power has become the focus of global renewable energy development. With the rapid development of offshore wind power technology, European offshore wind power powers have begun to move towards large-scale and deep-sea development. The currently planned Dogger Bank offshore wind farm in the UK is 123-289km offshore, with a total installed capacity of 4.8GW. Germany has planned 13 offshore wind farms with a total installed capacity of 21.3GW within 370.4km outside its territorial waters. China's offshore wind farms are also moving towards large-scale and deep-sea development. As one of the mainstream models of offshore wind turbines, doubly-fed asynchronous generators face problems such as harsh op...

AI Technical Summary

Problems solved by technology

However, once the early warning time of the SCADA system is insufficient, it will lead to insufficient fault samples collected, making fault diagnosis impossible

In addition, when a wind turbine has an early minor fault, the monitored state quantity usually does not exceed the system threshold. At this time, the SCADA system cannot give an effective early warning, resulting in the failure to obtain fault samples.

Images