Genetic Algorithm Optimized Back Propagation Research Articles

Monitoring Soybean Soil Moisture Content Based on UAV Multispectral and Thermal-Infrared Remote-Sensing Information Fusion.

By integrating the thermal characteristics from thermal-infrared remote sensing with the physiological and structural information of vegetation revealed by multispectral remote sensing, a more comprehensive assessment of the crop soil-moisture-status response can be achieved. In this study, multispectral and thermal-infrared remote-sensing data, along with soil-moisture-content (SMC) samples (0~20 cm, 20~40 cm, and 40~60 cm soil layers), were collected during the flowering stage of soybean. Data sources included vegetation indices, texture features, texture indices, and thermal-infrared vegetation indices. Spectral parameters with a significant correlation level (p < 0.01) were selected and input into the model as single- and fuse-input variables. Three machine learning methods, eXtreme Gradient Boosting (XGBoost), Random Forest (RF), and Genetic Algorithm-optimized Backpropagation Neural Network (GA-BP), were utilized to construct prediction models for soybean SMC based on the fusion of UAV multispectral and thermal-infrared remote-sensing information. The results indicated that among the single-input variables, the vegetation indices (VIs) derived from multispectral sensors had the optimal accuracy for monitoring SMC in different soil layers under soybean cultivation. The prediction accuracy was the lowest when using single-texture information, while the combination of texture feature values into new texture indices significantly improved the performance of estimating SMC. The fusion of vegetation indices (VIs), texture indices (TIs), and thermal-infrared vegetation indices (TVIs) provided a better prediction of soybean SMC. The optimal prediction model for SMC in different soil layers under soybean cultivation was constructed based on the input combination of VIs + TIs + TVIs, and XGBoost was identified as the preferred method for soybean SMC monitoring and modeling, with its R2 = 0.780, RMSE = 0.437%, and MRE = 1.667% in predicting 0~20 cm SMC. In summary, the fusion of UAV multispectral and thermal-infrared remote-sensing information has good application value in predicting SMC in different soil layers under soybean cultivation. This study can provide technical support for precise management of soybean soil moisture status using the UAV platform.

Prediction of Geometric Characteristics of Laser Cladding Layer Based on Least Squares Support Vector Regression and Crested Porcupine Optimization.

The morphology size of laser cladding is a crucial parameter that significantly impacts the quality and performance of the cladding layer. This study proposes a predictive model for the cladding morphology size based on the Least Squares Support Vector Regression (LSSVR) and the Crowned Porcupine Optimization (CPO) algorithm. Specifically, the proposed model takes three key parameters as inputs: laser power, scanning speed, and powder feeding rate, with the width and height of the cladding layer as outputs. To further enhance the predictive accuracy of the LSSVR model, a CPO-based optimization strategy is applied to adjust the penalty factor and kernel parameters. Consequently, the CPO-LSSVR model is established and evaluated against the LSSVR model and the Genetic Algorithm-optimized Backpropagation Neural Network (GA-BP) model in terms of relative error metrics. The experimental results demonstrate that the CPO-LSSVR model can achieve a significantly improved relative error of no more than 2.5%, indicating a substantial enhancement in predictive accuracy compared to other methods and showcasing its superior predictive performance. The high accuracy of the CPO-LSSVR model can effectively guide the selection of laser cladding process parameters and thereby enhance the quality and efficiency of the cladding process.

Smart Cushions with Machine Learning-Enhanced Force Sensors for Pressure Injury Risk Assessment.

Prolonged sitting can easily result in pressure injury (PI) for certain people who have had strokes or spinal cord injuries. There are not many methods available for tracking contact surface pressure and shear force to evaluate the PI risk. Here, we propose a smart cushion that uses two-dimensional force sensors (2D-FSs) to measure the pressure and shear force in the buttocks. A machine learning algorithm is then used to compute the shear stresses in the gluteal muscles, which helps to determine the PI risk. The 2D-FS consists of a ferroelectret coaxial sensor (FCS) unit placed atop a ferroelectret film sensor (FFS) unit, allowing it to detect both vertical and horizontal forces simultaneously. To characterize and calibrate, two experimental approaches are applied: one involves simultaneously applying two perpendicular forces, and one involves applying a single force. To separate the two forces, the 2D-FS is decoupled using a deep neural network technique. Multiple FCSs are embedded to form a smart cushion, and a genetic algorithm-optimized backpropagation neural network is proposed and trained to predict the shear strain in the buttocks to prevent PI. By tracking the danger of PI, the smart cushion based on 2D-FSs may be further connected with home-based intelligent care platforms to increase patient equality for spinal cord injury patients and lower the expense of nursing or rehabilitation care.

Estimation of Winter Wheat Chlorophyll Content Based on Wavelet Transform and the Optimal Spectral Index

Hyperspectral remote sensing technology plays a vital role in advancing modern precision agriculture due to its non-destructive and efficient nature. To achieve accurate monitoring of winter wheat chlorophyll content, this study utilized 68 sets of chlorophyll content data and hyperspectral measurements collected during the jointing stage of winter wheat over two consecutive years (2019–2020), under various fertilization types and nitrogen application levels. Continuous wavelet transform was applied to transform the original reflectance, ranging from 21 to 210, and the correlation matrix method was utilized to identify the spectral index at each scale, with the highest correlation to winter wheat chlorophyll content as the optimal spectral index combination input. Subsequently, winter wheat chlorophyll content prediction models were developed using three machine learning methods: random forest (RF), support vector machine (SVM), and a genetic algorithm-optimized backpropagation neural network (GA-BP). The results indicate that the spectral data processed through continuous wavelet transform at seven scales, from 21 to 27, show the highest correlation with winter wheat chlorophyll content at a scale of 26, with a correlation coefficient of 0.738, compared with the correlation of 0.611 of the original reflectance, and the accuracy is improved by 20.7%. The average highest correlation value between the spectral index at scale 26 and winter wheat chlorophyll content is 0.752. As the scale of wavelet transform increases, the correlation between the spectral index and winter wheat chlorophyll content and the accuracy of the predictive model show a trend of first increasing and then decreasing. The optimal input variables for predicting winter wheat chlorophyll content and the best machine learning method are the spectral data at a scale of 26 processing combined with the GA-BP model. The optimal predictive model has a validation set coefficient of determination (R2) of 0.859, root mean square error (RMSE) of 1.366, and mean relative error (MRE) of 2.920%. The results show that the prediction model can provide a technical basis for improving the hyperspectral inversion accuracy of winter wheat chlorophyll and modern precision agriculture.

Towards post-curing parameters optimization of phthalonitrile composites through the synergy of experiment and machine learning

Phthalonitrile composites are highly anticipated in fields such as aerospace, marine, and electronics due to their exceptional heat resistance, excellent high-temperature mechanical properties, and outstanding processability. The conversion of nitrile functional groups to macrocyclic structures during post-curing is essential for performance enhancement. However, the influence of post-curing conditions on the mechanical properties of phthalonitrile composites is complicated, and the underlying mechanism remains unclear. In this study, an experimental and machine learning synergic strategy was employed to optimize the post-curing parameters of the phthalonitrile composites, including temperature, time, and pressure. Through mechanical experiments conducted at both room temperature and 400 °C, scanning electron microscope, and dynamic mechanical analysis, the underlying mechanism of the influence of post-curing parameters on the composite mechanical properties was revealed. The enhancement of the polymerization degree brought about by post-curing is the decisive factor for high-temperature performance, while the room-temperature properties are a tradeoff between the polymerization degree improvement and the defect proliferation. The genetic algorithm-optimized backpropagation (GA-BP) neural network was trained for the efficient and accurate prediction of the optimal post-curing parameters. The tendency of mechanical properties with the post-curing parameters predicted by machine learning is consistent with the mechanism revealed by experiments.

Amphibious vehicle's resistance optimization through neural networks and genetic algorithms

Amphibious vehicles, as a new type of aquatic and terrestrial transport platform, are increasingly involved in the existing transportation system. Resistance is a key factor that affects the efficiency and energy consumption of vehicles in aquatic sailing. Resistance reduction optimization design is a focal point and challenge in the design process of amphibious vehicles. In this paper, a resistance performance optimization method has been proposed based on neural networks and genetic algorithms. First, key parameters for the shape design are extracted based on a thorough understanding of the vehicle's performance. These parameters are used to construct a parameterized design space. Second, a training set is obtained based on the Latin hypercube sampling method and numerical calculation methods, and a test set is randomly generated. To achieve better resistance prediction performance, a method based on the genetic algorithm-optimized backpropagation Neural Network is proposed. Next, the resistance performance of the two operating conditions is optimized through the non-dominated sorting genetic algorithm II, and optimized configuration parameters are obtained, which has a 22.71% energy-saving ratio at cruising speed. Finally, the optimized configuration is analyzed using numerical calculation methods to validate the resistance prediction and optimization methods.

Long-term water demand forecasting using artificial intelligence models in the Tuojiang River basin, China.

Accurate forecasts of water demand are a crucial factor in the strategic planning and judicious use of finite water resources within a region, underpinning sustainable socio-economic development. This study aims to compare the applicability of various artificial intelligence models for long-term water demand forecasting across different water use sectors. We utilized the Tuojiang River basin in Sichuan Province as our case study, comparing the performance of five artificial intelligence models: Genetic Algorithm optimized Back Propagation Neural Network (GA-BP), Extreme Learning Machine (ELM), Gaussian Process Regression (GPR), Support Vector Regression (SVR), and Random Forest (RF). These models were employed to predict water demand in the agricultural, industrial, domestic, and ecological sectors using actual water demand data and relevant influential factors from 2005 to 2020. Model performance was evaluated based on the Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE), with the most effective model used for 2025 water demand projections for each sector within the study area. Our findings reveal that the GPR model demonstrated superior results in predicting water demand for the agricultural, domestic, and ecological sectors, attaining R2 values of 0.9811, 0.9338, and 0.9142 for the respective test sets. Also, the GA-BP model performed optimally in predicting industrial water demand, with an R2 of 0.8580. The identified optimal prediction model provides a useful tool for future long-term water demand forecasting, promoting sustainable water resource management.

DBN-GABP model for estimation of aircraft wake vortex parameters using Lidar data

Aircraft wake turbulence is an inherent outcome of aircraft flight, presenting a substantial challenge to air traffic control, aviation safety and operational efficiency. Building upon data obtained from coherent Doppler Lidar detection, and combining Dynamic Bayesian Networks(DBN) with Genetic Algorithm-optimized Backpropagation Neural Networks(GA-BPNN), this paper proposes a model for the inversion of wake vortex parameters. During the wake vortex flow field simulation analysis, the wind and turbulent environment were initially superimposed onto the simulated wake velocity field. Subsequently, Lidar-detected echoes of the velocity field are simulated to obtain a data set similar to the actual situation for model training. In the case study validation, real measured data underwent preprocessing and were then input into the established model. This allowed us to construct the wake vortex characteristic parameter inversion model. The final results demonstrated that our model achieved parameter inversion with only minor errors. In a practical example, our model in this paper significantly reduced the mean square error of the inverted velocity field when compared to the traditional algorithm. This study holds significant promise for real-time monitoring of wake vortices at airports, and is proved a crucial step in developing wake vortex interval standards.

Intelligent prediction model of a polymer fracture grouting effect based on a genetic algorithm-optimized back propagation neural network

Polymer grouting can effectively improve the stability of surrounding rock fractures. However, in practical construction, it is difficult to judge the degree of coupling between the slurry and the rock, and the effective grouting range after grouting. Therefore, early prediction of the effect of grouting on the surrounding rock is crucial. In this paper, a new artificial intelligence method is proposed to predict the polymer fracture grouting effect. The genetic algorithm optimized back propagation neural network (GA-BP) is employed to construct an intelligent prediction model. To acquire a substantial dataset for constructing the model, an easily assembled/disassembled test apparatus for polymer fracture grouting is designed. The maximum coupling degree of the fractures and slurry diffusion distance are chosen as the evaluation metrics for the grouting effectiveness. The influences of the fracture characteristic parameters and grouting volume on the grouting effect are investigated. Furthermore, a comprehensive analysis is conducted on the spatiotemporal diffusion characteristics and slurry-rock coupling mechanism of polymer grouting. Compared to traditional BP neural networks, and three other machine learning algorithms (decision trees, random forests and gradient boosting decision trees), the GA-BP model outperforms them in terms of R2 (coefficient of determination), MSE (mean squared error), MBE (mean bias error), MAE (mean absolute error) and RMSE (root mean squared error) in both the test and training sets. The GA algorithm significantly improves the accuracy and robustness of the prediction model. The optimized model demonstrates significant accuracy in predicting grouting results and assessing efficiency, providing a practical reference for grouting construction.

Modeling of the shape of Taylor cone in EHD printing based on backpropagation neural network and genetic algorithm

Electrohydrodynamic (EHD) printing has been recognized as a promising additive manufacturing technology with superior pattern resolution and economic viability. The shape of Taylor cone is deemed a pivotal element for the amplification of deposition efficiency, and the maintenance of consistent operational steadiness in EHD printing. The correlations between diverse operating parameters and the shape of Taylor cone are presently not well investigated. In this paper, modeling of relationships between operating parameters and the shape of Taylor cone was conducted with a backpropagation neural network (BPNN) and a genetic algorithm optimized backpropagation (GA-BP) neural network. Taylor cone semi-vertical angle and the meniscus height were employed as two indexes to characterize the shape of Taylor cone. The prediction accuracies of BPNN model and GA-BP model were 92.79 % and 95.46 %, respectively. The GA-BP model demonstrated higher precision in forecasting the shape of Taylor cone. A predictive framework for the shape of Taylor cone was proposed, which provided a practical tool for process optimization in EHD printing.

Prediction modeling of cigarette ventilation rate based on genetic algorithm backpropagation (GABP) neural network

AbstractThe ventilation rate of cigarettes is an important indicator that affects the internal quality of cigarettes. When producing cigarettes, the unit may experience unstable ventilation rates, which can lead to a decrease in cigarette quality and pose certain risks to smokers. By establishing the ventilation rate prediction model, guide the design of unit parameters in advance, to achieve the goal of stabilizing unit ventilation rate, improve the stability of cigarette ventilation rate, and enhance the quality of cigarettes. This paper used multiple linear regression networks (MLR), backpropagation neural networks (BPNN), and genetic algorithm-optimized backpropagation (GABP) to construct a model for the prediction of cigarette ventilation rate. The model results indicated that the total ventilation rate was significantly positively correlated with weight (P &lt; 0.01), circumference, hardness, filter air permeability, and open resistance. The results showed that the MLR models' (RMSE = 0.652, R2 = 0.841) and the BPNN models’ (RMSE = 0.640, R2 = 0.847) prediction ability were limited. Optimization by genetic algorithm, GABP models were generated and exhibited a little better prediction performance (RMSE = 0.606, R2 = 0.873). The results indicated that the GABP model has the highest accuracy in the prediction of predicting ventilation rate and can accurately predict cigarette ventilation rate. This method can provide theoretical guidance and technical support for the stability study of the ventilation rate of the unit, improve the design and manufacturing capabilities and product quality of short cigarette products, and help to improve the quality of cigarettes.

Short-term power grid load forecasting based on variable weight combination hybrid model

Abstract The power grid load exhibits non-linearity and volatility, posing challenges to power grid dispatching. To enhance the precision of power grid load forecasting, a variable weight combination forecasting model is suggested to address the issue of inadequate forecasting efficacy of individual algorithms. Considering the impact of various environmental factors on power grid load, a load influence feature dataset is formulated. Initially, support vector machines, genetic algorithm-optimized back propagation (BP) neural networks and radial basis neural networks are employed to forecast individual loads. Subsequently, a variance–covariance weight dynamic distribution method is utilized to merge the prediction results of the three individual algorithms, thereby establishing a short-term power grid load prediction model with variable weight combination. Taking a regional power grid as an example, the simulation results show that the prediction accuracy of the variable weight combination model is higher than that of the single algorithm. Taking the evaluation index MAPE as an example, compared with the three single algorithms, the prediction accuracy is improved by 42.31%, 48.56% and 65.33%. The practice proves that the proposed variable weight combination forecasting model greatly improves the accuracy of power network load forecasting.

Investigation on compressive strength of coral aggregate concrete: Hybrid machine learning models and experimental validation

The compressive strength of coral aggregate concrete (CAC-CS) holds importance in structural engineering and architectural design. Thus, this study establishes machine learning models for the prediction of CAC-CS and assesses their accuracy and generalization capability. The importance of input variables is analyzed by using Shapley additive explanations (SHAP) and single-variable fluctuations. A user-friendly graphical user interface (GUI) for CAC-CS prediction is developed to facilitate practical use. The accuracy of the GUI is subsequently validated through experiments. Results demonstrate that the genetic algorithm-optimized backpropagation neural network (GA-BP) model provides predictions that are closely aligned with experimental values. The GA-BP model minimizes the mean and standard deviation of its residual distribution. The performance indicators, including correlation coefficient, mean absolute error, mean absolute percentage error, mean square error, and root mean square error, of the GA-BP model are 0.98, 3.78, 10.95, 11.08, and 3.33, respectively, and are superior to those of the other tested models. The analysis of SHAP values and single-variable fluctuations reveals that among the factors influencing CAC-CS, water-binder ratio and curing age are the most sensitive. Reducing the water-binder ratio and increasing the curing age enhances CAC-CS. Additionally, the sensitivity analysis of mineral admixtures and the water-binder ratio in experiments aligns with the analysis of SHAP values and single-variable fluctuations. The error between the experimental values and GUI predictions for 14 mixture designs is less than 5 %, thus validating the accuracy and generalization capability of the GUI proposed in this study. In conclusion, this research provides practical guidelines for preparing CAC and contributes to the conservation and development of island resources.

Retraction: Designing an intelligent monitoring system for corn seeding by machine vision and Genetic Algorithm-optimized Back Propagation algorithm under precision positioning.

Retraction: Designing an intelligent monitoring system for corn seeding by machine vision and Genetic Algorithm-optimized Back Propagation algorithm under precision positioning.

Prediction of elastic properties of 3D4D rotary braided composites with voids using multi-scale finite element and surrogate models

The conventional evaluation of 3D braided composites' mechanical properties through numerical and experimental methodologies serves as a hindrance to material application owing to the considerable expenses, time constraints, and laborious efforts involved. Moreover, the presence of void defects induced during the processing exacerbates this challenge. In this study, a multi-scale finite element model (FEM) and a surrogate model are established for predicting elastic properties of three dimensional four directional (3D4D) rotary braided composites with voids for the first time. Based on the established FEM, a comprehensive dataset containing 768 data points is formed, covering the ranges of both design parameters and void defect parameters. The influence of braiding angle, yarn width, and porosity, on the elastic constants of 3D4D rotary braided composites is accurately analyzed. A genetic algorithm-optimized back propagation neural network (GABPNN) model is developed, which possess the capability to replicate FEM outcomes with a commendable R-value of 0.99. The remarkable concordance between the anticipated outcomes and experimental datasets corroborates the triumphant implementation of the present method in unraveling the interconnections between microstructure and properties in 3D4D rotary braided composites containing voids. Consequently, this offers a propitious instrument for expediting the intelligent conception and refinement of composite materials.

Spatial Downscaling of Soil Moisture Based on Fusion Methods in Complex Terrains

Large-area soil moisture (SM) data with high resolution and precision are the foundation for the research and application of hydrological and meteorological models, water resource evaluation, agricultural management, and warning of geological disasters. It is still challenging to downscale SM products in complex terrains that require fine spatial details. In this study, SM data from the Soil Moisture Active and Passive (SMAP) satellite were downscaled from 36 to 1 km in the summer and autumn of 2017 in Sichuan Province, China. Genetic-algorithm-optimized backpropagation (GABP) neural network, random forest, and convolutional neural network were applied. A fusion model between SM and longitude, latitude, elevation, slope, aspect, land-cover type, land surface temperature, normalized difference vegetation index, enhanced vegetation index, evapotranspiration, day sequence, and AM/PM was established. After downscaling, the in situ information was fused through a geographical analysis combined with a spatial interpolation to improve the quality of the downscaled SM. The comparative results show that in complex terrains, the GABP neural network better captures the soil moisture variations in both time and space domains. The GDA_Kriging method is able to merge in situ information in the downscaled SM while simultaneously maintaining the dynamic range and spatial details.

Multi‐model predictive control ofSCRflue gas denitrification system in coal‐fired power plant based on kernel fuzzy c‐means clustering and integrated model

AbstractA multi‐model predictive control strategy based on kernel fuzzy c‐means (KFCM) clustering and integrated model is proposed for the complex problem of rapid and accurate control of ammonia injection in selective catalytic reduction (SCR) denitrification systems of coal‐fired power plants under a wide range of variable load conditions. First, the SCR data samples are clustered using the KFCM clustering algorithm, and the number of clusters is determined by introducing the Xie‐Beni index. Second, the prediction model of the SCR denitrification system is established by an integrated modelling approach, and the sub‐learners of the integrated model are the genetic algorithm optimized back propagation (GA‐BP) neural network model and the least squares support vector machine (LSSVM) model. Third, a multi‐model prediction controller based on the particle swarm optimization (PSO) algorithm and the integrated model is designed and developed. To ensure the stability of the system, a model‐switching strategy based on the minimum Euclidean distance is proposed. Finally, simulation verification and industrial field application verification are fulfilled by comparing with proportion integral differential (PID) control and single model predictive control (MPC). The results show that the multi‐model predictive control method proposed in this paper can obtain higher control accuracy and better control stability and meet the control requirements for the long‐term operation of the SCR denitrification system.

Temperature field test and prediction using a GA-BP neural network for CRTS II slab tracks

AbstractThe CRTS II slab track, which is connected in a longitudinal direction, is one of the main ballastless tracks in China, with approximately 7365 km of operational track. Temperature loading is a very vital factor leading to slab track damages such as warping and cracking. While existing research on temperature distribution rests on either site tests in special environments or theoretical analysis, the long-term temperature field characteristics are not clear. Therefore, a long-term temperature field test for the CRTS II slab track on bridge-subgrade transition section was conducted to analyze the temperature field. A GA-BP (genetic algorithm optimized back propagation) neural network was trained on the test data to predict the temperature field. The vertical and lateral temperature distributions in four typical days were carried out. We found that the temperature along the track was distributed in a nonlinear manner. This was particularly distinct in the vertical direction for depths of less than 300 mm. The highest and lowest daily temperatures and the daily range of the temperature were analyzed. With the increasing depth, the daily highest temperatures and range of the temperature were smaller, the daily lowest temperatures were higher, and the time corresponding to this peak value appeared later in the day. Both the highest and lowest daily temperature could be predicted using the GA-BP neural network, though the accuracy in predicting the highest temperature was higher than that in predicting the lowest temperature.

Machine learning-assisted correlations of heat/mass transfer and pressure drop of microchannel membrane-based desorber/absorber for compact absorption cycles

Microchannel membrane-based desorbers and absorbers are key components in efficient and compact absorption refrigeration systems. To improve the accuracy of current empirical correlations for describing the heat/mass transfer and solution pressure drop characteristics, more advanced correlation models are urgently needed. This study applies three machine learning algorithms, namely Random Forest (RF), Least-Squares Support Vector Machine (LS-SVM), and Genetic Algorithm-optimized Back Propagation Artificial Neural Network (GABPNN), to develop new models for Nusselt number (Nu), Sherwood number (Sh), and friction factor (Fr) of microchannel membrane-based desorber and absorber, respectively, based on experimental results. These machine learning-assisted models effectively improve the prediction accuracy for both the desorber and absorber. Among them, the Random Forest model performs best, improving the prediction accuracy by 12.37% (Nu), 21.49% (Sh), and 14.24% (Fr) for the desorber and 24.28% (Nu), 27.82% (Sh), and 30.47% (Fr) for the absorber, compared to the conventional empirical correlations. Moreover, in relative to the correlations obtained from the literature, the Random Forest model predicts the overall heat transfer coefficient (U), sorption rate (J), and solution pressure drop (DP) with accuracies higher by 44.75%, 64.66%, and 83.52% for the desorber, and 79.66%, 81.52%, and 69.53% for the absorber. This study is expected to facilitate the simulation, design, and optimization of microchannel membrane-based desorbers/absorbers towards more efficient and compact absorption cycles.

Improved artificial neural networks (ANNs) for predicting the gas separation performance of polyimides

This study aimed to establish a quantitative structure–property relationship (QSPR) model for predicting the gas separation performance of polyimide membranes using neural networks combined with the repeat unit structure of materials. Using a data bank based on 125 polyimides, we calculated a total of 20 descriptors for all polyimides using Yampolskii's group contribution method, which divides polyimides' repeat units into their smallest groups. The number of groups contained in each polyimide is taken as the network input, and the gas permeability as the network output. Two neural network models, back-propagation (BP) and genetic algorithm-optimized back-propagation (GABP) algorithms, were used as the prediction model, and the prediction results were compared. When compared with the previous models used to predict the gas separation performance for all polymers and other machine learning (ML) models, the prediction results obtained using the GABP model are encouraging, showing a root mean squared error (RMSE) of 0.44 for CO2, indicating that the model is applicable to polyimide. In addition, the GABP model is easy to operate, requires few parameters, it is also applicable to copolyimides. The GABP model based on the group contribution method can thus satisfactorily predict polyimides' gas separation. This is expected to be used to guide the synthesis and structure screening of polyimides for saving resources and commercialization.