Design and Application of Scientific Research Project Management System based on Convolutional Neural Network and Stochastic Gradient Descent with Multilayer Perceptron

The scientific research project management system build on web technology, addresses the increasing data generated by stakeholders throughout the product life cycle. However, while project management in energy technology sector is typically evaluated from the perspectives of energy economics and environmental sustainability, there are challenges in improving and standardizing quality management processes. In this paper, proposed Convolutional Neural Network (CNN) and Stochastic Gradient Descent with Multilayer Perceptron (SGD-MLP) to enhance the efficiency and accuracy of project management system. Data is obtained from Republic of Slovenia for Public Legal Records and Related Services (AJPES), spanning a yearlong period. The ResNet-50 model is used for feature extraction, which is effectively mange’s project specific feature and classifies the data. The CNN is responsible for classifying various tasks within a research project, while the MLP is used for resource utilization of project, ensuring efficient management of resources. The proposed CNN-SGD-MLP technique achieve better accuracy in 90.99%, precision of 98.13%, recall of 89.42% and F1-score of 89.46% in cost phase, 90.99% of accuracy in involved the time and overall average is 91.06% of accuracy, 90.16% of precision, 90.43% of recall and 90.12% of f1-score in PM and exiting method Structure Equation Modelling (SEM) sample.