Prediction and Control of Input and Output for Industry–University–Research Collaboration Network in Construction Industry

An unreasonable allocation of resources has led to a low rate output in the industry–university–research collaboration network. A solution this problem is control and predict input output. However, network characteristics strong nonlinearity insufficient samples. It difficult for existing methods migrate networks because traditional methods, including Proportional–Integral–Derivative (PID) Model Predictive Control (MPC), are usually not applied system with controlled needs have specific parameters, while modern feedforward feedback control, their limitations both parameters other aspects. In addition, there lack research on prediction networks, no effective applicable scheme prediction. Considering samples network, Feedforward Control–Feedback based Multi-Layer Perceptron (FCFCM-MLP) proposed paper. Adopting controller structure Grid Search-Multilayer (GS-MLP), block diagram, controller, such as Harris Hawk Optimization-Support Vector Regression (HHO-SVR) designed FCFCM-MLP, which effectively realizes inputs outputs. paper, simulation tests output-feedback tracking conducted real statistics papers jointly produced by construction industry. The results show that model obvious effectiveness. Specifically, compared composed structures optimal Particle Dynamic Multiple Perturbation_Butterfly Optimization Algorithm-Support Regression_Grid Search-Multi-Layer (PDM_BOA-SVR_GS-MLP) obtained paper can minimize predictive error improve accuracy.