To address issues such as insufficient fault response sensitivity and poor coordination between upper and lower-level protections caused by fixed parameters of traditional power grid relay protection equipment

a relay protection fault detection and adaptive coordination scheme integrating multi-sensor data

time-series imaging technology

and deep learning is proposed.The scheme first collects power grid operation data through voltage and current sensors

and sets a sampling frequency of 4800Hz to ensure data timeliness and accuracy.It then converts one-dimensional time-series signals acquired by the sensors into two-dimensional images using Gramian angular summation field(GASF)

effectively preserving the time-domain correlation features of the signals.Furthermore

a fault classification model is constructed based on an improved AlexNet convolutional neural network(CNN)

which can accurately identify eight types of faults including single-phase grounding

two-phase short circuit

three-phase short circuit and normal operation state.Based on fault classification

the LINKNET method is introduced to identify the pairing relationship between main and backup protection relays

and relay parameters are dynamically adjusted to ensure the coordination between upper and lower-level protections.Experimental results based on a 9-node power grid test system demonstrate that the proposed fault classification model achieves 100% classification accuracy on the test set

and the optimization solution time is only 0.047s.By adaptively adjusting relay parameters

the protection action time in fault scenarios is significantly shortened

and the coordination time interval meets the specification requirements.