Open lilyniu opened 5 days ago
Thank you for your interest in our research. This demonstration primarily focuses on efficiently feeding all the data into the pipeline for prediction.
During the presentation, we will use the following settings:
cell_number = 9 # battery ID
phase = 99 # which age of feature to predict
start_cycle = 500 # starting cycle
append = 5 # attach n cycle features
These settings control the prediction conditions, ensuring that only information from the start_cycle and the following 5 appended cycles is used for the target prediction. Performing individual predictions without contamination from data outside the target range is also possible. You only need to refer to the padding method used in process2predict for handling the data, then input it into the model for prediction.
In the inferring_voltage code file, the file for charge_data and discharge_data used for the input values of the model uses information that is not from the first 100 cycles, for example print(charge_data[0].shape) is (687, 4, 500). How to predict the whole complete using only the first 100 cycle's, would like to have your answer thank you!
`cell_feature={}
for i in tqdm(range(len(charge_data))): charge_feature=feature_selector(feature_selector_ch, charge_data[i],charge_norm) discharge_feature=feature_selector(feature_selector_dis, discharge_data[i],discharge_norm) cell_feature['%d'%(i)]=concat_data(charge_feature,discharge_feature, summary_data[i])
x_in1,x_in2,y_in1,y_in2,y_in3=process2predict(cell_feature,section) in_x1=np.vstack(x_in1[cell_number]).reshape(-1,input_data_len,12) in_x2=np.vstack(x_in2[cell_number]).reshape(-1,1) predict_voltage,predict_capacity,predict_power=predictor.predict([in_x1,in_x2],batch_size=256)`