Deep learning-based stop control method and system for high-speed train

A deep learning-based stop control system for a high-speed train, the system comprising:

• a training data set obtaining module, configured to obtain a training data set, wherein the training data set comprises a corresponding position, speed, and stop position of a train at each time point in a time period, and the time period comprises a current time point and a time point before the current time point;
• a convolutional neural network (CNN) establishment module, configured to establish a CNN, wherein the CNN comprises an input layer, A convolutional layers, an expansion layer, B hidden layers, and an output layer, wherein A is a positive integer greater than 0 and less than or equal to 2, and B is a positive integer greater than or equal to 2 and less than or equal to 3, the input layer comprises (N−1)*2 two-dimensional input neurons, wherein N represents a quantity of time points comprised in the time period, each convolutional layer comprises C convolution kernels, wherein C is a positive integer greater than or equal to 10 and less than or equal to 30, and each hidden layer comprises D neurons, wherein D is a positive integer greater than or equal to 5 and less than or equal to 20:
• a training and optimization module, configured to train and optimize the CNN by using the training data set and a Dropout method, to obtain an optimized CNN, wherein the Dropout method comprises:
• step 1:
• randomly discarding a proportion of neurons during model training;
• and
• step 2:
• changing an output yl of the hidden layer from yl=f(wlzl−1+bl) to yl=f(wlzl−1rl−1+bl) where wl represents a weight matrix from an l−1th hidden layer to an lth hidden layer, bl represents a bias of the lth hidden layer, zl−1 represents the input of the (l−1)th hidden layer, rl−1=Bernoulli(p), representing that a vector (0, 1) is randomly generated based on a probability of p;
• an actual running data obtaining module, configured to obtain actual running data of a to-be-controlled train, wherein the actual running data comprises a corresponding position and speed of the to-be-controlled train at each time point in the time period;
• a prediction module, configured to input the actual running data into the optimized CNN to obtain a stop position of the to-be-controlled train;
• a determining module, configured to determine whether the stop position of the to-be-controlled train is 0;
• a breaking module, configured to output a breaking command when an output result of the determining module is that the stop position of the to-be-controlled train is 0;
• and
• a return module, configured to return to execute the actual running data obtaining module when the output result of the determining module is that the stop position of the to-be-controlled train is not 0.