What question did this study set out to answer?

This research aims to develop an efficient energy consumption prediction framework for evaluating green buildings in rural revitalization contexts.

April 30, 2026Open Access

Performance evaluation of green building in the process of rural revitalization driven by Artificial Intelligence and BP neural network

Key Points

This research aims to develop an efficient energy consumption prediction framework for evaluating green buildings in rural revitalization contexts.
Constructed a hybrid prediction model using Back Propagation Neural Network optimized by Marine Predators Algorithm.
Utilized a public building energy efficiency dataset with eight building features as inputs.
Trained and tested the optimized MPA-BPNN model for Heating Load and Cooling Load predictions.
The MPA-BPNN model achieved a Coefficient of Determination (R²) of 0.9995 for Heating Load and 0.9989 for Cooling Load.
Root Mean Square Errors (RMSE) for Heating Load and Cooling Load were reduced to 0.2118 and 0.3175, respectively.

Abstract

To address the problems of complex models, high costs, and insufficient universality in green building performance evaluation under the background of rural revitalization, this study proposes an Artificial Intelligence (AI)-driven efficient energy consumption prediction framework. The framework constructs a hybrid prediction model of Back Propagation Neural Network (BPNN) optimized by Marine Predators Algorithm (MPA). During the research process, the unique exploration and exploitation strategies of the MPA are utilized to globally optimize and search for the initial weights and thresholds of the BPNN model. This process effectively overcomes the inherent defects of traditional BPNN, such as sensitivity to initial parameters and ease of falling into local optimal solutions, and significantly improves the model’s convergence speed and optimization ability. The experiment uses a public building energy efficiency dataset; it takes eight building features as inputs: relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, and glazing area distribution. The outputs are Heating Load (HL) and Cooling Load (CL). These inputs and outputs are used to train and test the optimized MPA-BPNN model. The research results show that the constructed MPA-BPNN model performs excellently in HL and CL prediction. The Coefficients of Determination (R²) for HL and CL prediction are as high as 0.9995 and 0.9989, respectively, while the Root Mean Square Errors (RMSE) are reduced to 0.2118 and 0.3175, respectively. This intelligent evaluation model can provide efficient and reliable decision support for the planning and design of rural green buildings, and is of great significance for promoting the green and low-carbon transformation of rural buildings and assisting in the realization of the rural revitalization strategy.

Ask AI

Mark Helpful

Bookmark

Relay

View Full Paper