“…The inertia weight will gradually decrease with the iteration. Then initialize the particle swarm: Randomly initialize the position of the particle swarm within the corresponding range: ( , , , , ) i X k n l lr b k represents the size of convolution kernel in TCN, with the range of [2,8] n represents the number of convolution kernels in TCN network, with the range of [10,200] l represents the number of layers in TCN network, with the range of [2,6] lr represents the learning rate, with the value range of [0.001, 0.01] b represents the batch processing size, with a value range of [64,256] Randomly initialize velocity of particles in each dimension. And the velocity of particles in a certain dimension is always limited to 20% of the range of the dimension Then update the particle swarm: In each iteration, the network determined by the position of each particle is trained.…”