Comprehensive safety risk evaluation of fireworks production enterprises using the frequency-based ANP and BPNN

Neural network-based prediction of auto-ignition temperature of ternary mixed liquids

“…In summary, both 2 models are able to successfully complete the work of predicting the spontaneous combustion temperature of mixed combustible liquids. However, in terms of the predictive ability and stability of the models, the BPNN model is stronger than the 1DCNN model, which is due to the stronger nonlinear modelling ability and adaptability of BPNN, for different types of data and problems, BPNN is able to adaptively adjust the weights and biases, and improve the generalization ability of the model [ 21 ]. And in terms of the architecture of the model, the architecture of the BPNN model is relatively simple, and the architecture of the 1DCNN model is more complex.…”

“…In the experiment, the activation function is ReLU [ 21 ], the loss function adopts mean square error (MSE), and the optimizer is Adam [ 22 ]. The input of the network is 205 groups of mixed molecular descriptors, and the output is the measured value of ternary mixed liquid AIT.…”

“…The BP algorithm is a gradient descent algorithm that makes learning of feedforward networks easier [ 21 , 22 ]. Fig.…”