The invention discloses a method for designing a neural network pooling layer based on an interval algorithm and relates to the field of artificial intelligence. The method uses the interval algorithm to define the input value and the output value of a pooling layer into interval data types to form an interval pooled layer. The interval pooling layer can deal with uncertain input values of a certain range, that is, the interval data types, so as to avoid the problem that some high-precision floating-point values cannot be accurately expressed by a computer, and can output reliable interval values to improve reliability. The interval pooling layer and the corresponding interval neural network technology are used to build an interval neural network, which can make up for the inability of the neural network to process interval data and solve the practical problems with the interval data by the technology of the convolutional neural network. The method can be applied to the actual scene.