Research on residential building duration prediction model based on mean clustering and neural network

Abstract

Abstract The duration target will directly affect the effectiveness of the implementation of the residential building project, so it is necessary to predict a reasonable residential building duration. In this work, genetic algorithm (GA) is used to optimize and improve the weights and thresholds of the BP neural network to form a GA-BP neural network model, and seven parameters, such as aboveground building structure, number of building floors, building area, decoration standards, foundation structure, number of underground building floors, and underground building area, are taken as the input parameters of the neural network model, and the residential building duration is taken as the output parameter. 111 sets of residential building duration data were collected and divided into 90 training sets and 21 test sets, and the model was validated and analyzed in comparison using root mean square error (RMSE), correlation coefficient (R) and average error rate, which were used to validate that the GA-BP neural network model helps in predicting the duration of residential buildings. In order to improve the prediction accuracy of GA-BP neural network model, this work uses artificial bee colony improved K-means clustering algorithm to classify 111 sets of experimental data and 33 sets of new data, according to the classification of the training and testing results show that the ABC-K-means-GA-B model all have a strong generalization ability and strong prediction accuracy, and at the same time, it proves that this paper's proposed ABC- K-means-GA-BP neural network helps to predict the construction period of residential buildings, which is of great practical significance for improving construction efficiency.