Aiming　at　the　problem　of　the　high　failure　rate　of　hydraulic　pumps　of　agricultural　machinery　and　the　lack　of　effective　means　and　methods　for　fault　determination,　this　paper　proposes　a　hydraulic　pump　fault　diagnosis　method　based　on　empirical　wavelet　transform　(EWT)　and　kernel　limit　learning　machine　for　solving　the　diagnosis　of　hydraulic　pump　faults　in　agricultural　machinery.　Firstly,　k-means　is　used　to　improve　EWT,　which　makes　the　signal　decomposition　more　accurate.　Then,　the　decomposed　sub-signals　are　fed　into　the　seagull　optimization　algorithm　(SOA)　improved　kernel　limit　learning　machine　for　fault　classification.　From　the　experimental　results,　it　is　known　that　the　proposed　method　has　an　accuracy　of　97.67%　for　hydraulic　pump　fault　diagnosis　and　can　effectively　diagnose　the　faults　of　hydraulic　pumps　of　agricultural　machinery.　©　2023　SPIE.