The invention belongs to the technical field of fault detection and
recovery, and particularly relates to a fault self-diagnosis and self-adaptive
recovery method for an
embedded controller, which comprises the following steps of: starting a fault self-diagnosis process by adopting a periodic triggering mode and an event triggering mode, and acquiring,
processing and generating a diagnosis
data source with
time sequence characteristics in real time; step-by-step anomaly recognition is carried out through a three-level anomaly recognition mechanism, an anomaly
feature set is extracted, cooperative fault positioning is carried out in combination with a weighted
cosine similarity algorithm and
fault tree analysis, fault types and severity levels are judged, and a fault positioning report is generated; the optimal self-adaptive
recovery strategy is matched to carry out preliminary recovery, the strategy is divided into light, medium and deep levels according to serious levels and is provided with priorities, and the scheme with the minimum influence on the core service is preferentially selected; the
recovery effect is verified through multi-dimensional comparison, and
processing is carried out according to the
recovery effect. According to the method, the real-time accurate diagnosis, intelligent positioning and self-adaptive recovery of the fault of the
embedded controller are realized, and the reliability and self-healing capability of the
system are improved.