.png){kind=icon}
Return
Gate recurrent unit neural network inverse model prediction-based dynamic multi-objective evolutionary algorithm and application
.png){kind=icon}
PRE
AI
DOI：10.1016/j.engappai.2026.114633.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Dynamic Multi-Objective Evolutionary Algorithms (DMOEAs) generate the initial population for future environment based on prediction model. When the prediction model mismatches the evolutionary regularity of decision variables in Dynamic Multi-objective Optimization Problems (DMOPs), prediction-based DMOEAs are difficult to generate high-quality initial populations. For this problem, a DMOEAs based on the inverse model prediction of Gate Recurrent Unit (GRU) neural network is proposed. GRU neural network is used to fit the mapping relationship between the objective function of DMOPs and the decision variables, and is taken as the inverse model of DMOPs. The objective function of predicting future environment through Kalman filtering as the input of inverse model to output the decision variables of future environment. The posterior improvement is carried out based on the evolution direction of decision variables output by the inverse model in historical environment, and the result is used as the initial population. The proposed method can generate high-quality initial populations in most DMOPs, and is not limited matching the variation law of prediction model with the decision variables, thereby has higher solving efficiency. The effectiveness of proposed method was verified through experiments and the application in environmental economic power dispatching.
Keywords:
Dynamic Multi-Objective Evolutionary Algorithm
Inverse Model Prediction
Gate Recurrent Unit
Kalman Filtering
Dynamic Multi-Objective Optimization
Journal
IF:
8
Papers:
5.7K
Citations:
3.5W
Organization
No organization information available
Cited Papers
No cited papers available