Maximum Percentage Error Research Articles

Investigation of the dynamic behavior of hybrid composites using finite element and experimental method

ABSTRACT Composite materials are used extensively nowadays because of their excellent strength-to-weight ratio, high hardness, and good tensile strength. The characteristics of the composite material are a function of the type of the fiber, i.e. natural or synthetic, size, a resin used and so on. Most industries are using hybrid composites nowadays to have the desired properties. In the present analysis, four different fibers, i.e. Carbon, E-Glass, Flax and Bamboo, are used, their composites hybrid composites have been created with ASTM D3039, and their mechanical behavior is investigated using Finite Element Analysis. The eigenvalues and vibrational behavior of the hybrid composites are analyzed through Finite Element Analysis. The eigenvalues and mechanical behavior are also determined through experimental testing and compared with the same obtained through Finite Element Analysis. The results obtained from the two methods agree well; therefore, the analytical model is validated. The optimum hybrid composite material is decided based on the mechanical properties, natural frequencies and damping ratio. The six hybrid composites show the natural frequency varies from 17 hz to 32 hz. Similar frequencies are predicted from Finite Element Analysis, and the maximum percentage error of experimental and Finite Element Analysis models varies up to 4%.

Performance prediction of flame-retardant clothing using correlations and artificial neural networks: Optimizing firefighter safety

Flame-Retardant Clothing serves as a protective shield for firefighters that safeguards them from exposure to heat, flames and other thermal hazards. To achieve an optimal design for the clothing, it is essential to simultaneously account for all the factors affecting the performance of the clothing. The present study employs data from a numerical model to explore heat and moisture transport through clothing subjected to flame exposure. Seventeen non-dimensionless parameters associated with the heat and moisture transport in flame-retardant clothing are obtained. A correlation is developed to link the dimensionless second-degree burn time with other non-dimensional parameters. This correlation provides a means to predict the thermal protective performance (TPP) of the clothing. Additionally, an Artificial Neural Network (ANN) method is employed to determine the TPP of the clothing. Multi-layer feedforward backpropagation networks are utilized to predict the TPP under specified exposure conditions. The findings indicate that both the correlations and the ANN approach adopted in the present study demonstrated promising results. However, the ANN model predictions show better agreement with model data in comparison to the results derived from the developed correlation. The maximum percentage error in the predicted non-dimensional second degree burn time using ANN is limited to 10 %.

A New Approach to Characterize Superplastic Materials from Free-Forming Test and Inverse Analysis

For about 60 years, the aerospace industry has been strongly interested in superplastic forming processes to produce extremely light and complex-shaped components. Superplastic characteristics are found in lightweight metallic materials such as titanium-based, aluminum-based, and, more recently, magnesium-based alloys. Since the high ductility exhibited by superplastic materials is two orders of magnitude higher than that of conventional materials, complex-shaped components can be obtained. If made with conventional materials, they require expensive assembly operations. The behaviour of superplastic materials is summarized by a constitutive equation commonly obtained via tensile testing that subjects the tested material to a one-dimensional stress state. On the contrary, free-forming tests allows us to test the material by subjecting it to a stress state similar to that determined during a real superplastic-forming process. The aim of this work is to define the characteristic parameters of superplastic materials by free-forming tests. The behaviour of superplastic materials is commonly modelled using a power law which puts the material into a stress-to-strain-rate relationship. This law needs to identify two parameters characterizing superplastic materials: the strain rate sensitivity index and the strength coefficient. In this work, a new procedure is presented that implies the two material parameters vary with strain. It allows for a reduction in the number of constants needed to determine the material constitutive equation, thus requiring low simulation time compared to models that adopt the multiple-objective optimization based on genetic algorithms (GAs). It is more suitable to be used in the industrial field. Furthermore, the proposed procedure is compared with a conventional procedure which is also based on the inverse analysis carried out through the use of a finite element analysis. The results of the conventional procedure, based on the inverse analysis, which is conducted through the use of a finite element analysis, are used to calculate the material constants, and are compared with those coming from the procedure proposed in this work. The proposed procedure appears equally simple and gives more accurate results compared to the conventional procedure. In fact, the maximum percentage error, regarding the prediction of the forming times of a free-forming process, was reduced from 20% to 8%. The development of the proposed procedure, as well as the comparison of the results with a conventional procedure, required the development of an experimental activity. This activity consists of free-forming tests conducted at a constant pressure (the pressures employed vary from 0.2 to 0.4 MPa), at a temperature of 753 K, and on circular sheets (thickness 1.0 mm and radius 40 mm) in superplastic magnesium alloy AZ31.

Measurement and analysis of residual stress of multilayer carbon fiber laminate based on incremental drilling method

In the automotive and aerospace industries, carbon fiber is widely used in structures such as chassis and wings due to its lightweight and excellent mechanical properties, and the residual stresses are crucial for product design. In this paper, six kinds of carbon fiber laminates with different layup directions are investigated by using the integral method, and the calibration coefficient matrix is numerically solved by using ABAQUS finite element software, and the effects of laminate direction and incremental step on the three-direction residual stresses are experimentally analyzed. The residual stresses in different oriented layers were predicted by classical lamination theory (CLT), and the longitudinal, transverse and in-plane shear stresses were compared by combining incremental drilling experiments with calibration factors. The results showed that the ply with [45/90/−45/0/45/90/−45/0] laminate had the best performance,the difference between theoretical and experimental values for σx was 0.421MPa. The maximum τxy measurement was 1.7103MPa, with a maximum percentage error of 25.3%. The maximum difference for σy was 0.7306MPa. These findings indicate that the method is more accurate in predicting the residual stresses in the first four layers of composite material. This provides important theoretical and experimental support for the optimal design of composites.

Domain knowledge-guided machine learning framework for state of health estimation in Lithium-ion batteries

Accurate estimation of battery state of health is crucial for effective electric vehicle battery management. Here, we propose five health indicators that can be extracted online from real-world electric vehicle operation and develop a machine learning-based method to estimate the battery state of health. The proposed indicators provide physical insights into the energy and power fade of the battery and enable accurate capacity estimation even with partially missing data. Moreover, they can be computed for portions of the charging profile and real-world driving discharging conditions, facilitating real-time battery degradation estimation. The indicators are computed using experimental data from five cells aged under electric vehicle conditions, and a linear regression model is used to estimate the state of health. The results show that models trained with power autocorrelation and energy-based features achieve capacity estimation with maximum absolute percentage error within 1.5% to 2.5%.

Analytical and Numerical Approaches to Optimize the Air Vents Location of Office Space at Different Surrounding Conditions

Optimization becomes an essential tool in all aspects of life. Air entrances and exits locations in any building are one of them that need to be optimized to reduce the energy consumption. Hence, a numerical and analytical investigations of two Models were performed to a teaching assistantship (TA’s) office. The original office design (Model 1) has all the hot and cold air inlets beside the air outlet in the mid-position of the room ceiling. The suggested model of the office (Model 2) is designed by putting the cold air inlet in the middle of the ceiling, whereas the hot air inlet in the middle of the floor close to the widow of the office. The air outlet was kept in the middle of the room ceiling. These models analysed as a function of the ambient temperatures in the 21st of four months (December, March, June, and September) within the periods 9:00, 12:00, and 16:00 h. For more validation, the comparison between the analytical and numerical results of Model 2 is validated with a maximum variation of about 7%. In addition, the analytical room temperature results of Model 1 are higher than Model 2 by 10.3%. Finally, the best average comfortable temperature for humans had been discovered in Model 2 for all cases around 23 ˚C with a maximum error percentage of less than 10%.

Control and monitor bottle filling, capping and labelling machine using a programmable logic controller and human machine interface

Bottle filling, closing, and labellingmachine control and monitoring tool using Programmable Logic Controller (PLC) and Human-Machine Interface (HMI). This tool is designed to optimize the production process in the bottled beverage industry,with dimensions of 200 cm × 70 cm × 100 cm. The conveyor transports bottles through the filling, capping, and labeling units. In each unit, sensors are installed to detect the presence of bottles. Bottle liquid volume is determined through time settings inputted through the HMI,and the filling is performed by a water pump. After filling, bottles are capped and tightened mechanically using an air impact system. Labeling is handled by two DC motors which unroll the label and rotate the bottle, ensuringthe label adheres evenlyto all sides of the bottle. The number of bottles produced is calculated using a capacitive proximity sensor and displayed on the HMI. Testing results show that the system effectivelycontrols and monitors the bottle filling, capping, and labeling process viaPLC and HMI. Based on the test results on the bottle filling unit, the system can fill water into the bottle with a maximum percentage error of 0.5% on a 1000 ml volume bottle. In the capping unit, the success rate achieved is 70%, while the success rate in the bottle labeling unit is 80%.

Enhanced PINNs with augmented Lagrangian method and transfer learning for hydrodynamic lubrication analysis

PurposeBy seamlessly integrating physical laws, physics-informed neural networks (PINNs) have flexibly solved a wide variety of partial differential equations (PDEs). However, encoding PDEs and constraints as soft penalties in the loss function can cause gradient imbalances, leading to training and accuracy issues. This study aims to introduce the augmented Lagrangian method (ALM) and transfer learning to address these challenges and enhance the effectiveness of PINNs for hydrodynamic lubrication analysis.Design/methodology/approachThe loss function was reformatted by ALM, adaptively adjusting the loss weights during training. Transfer learning was used to accelerate the convergence of PINNs under similar conditions. Additionally, the iterative process for load balancing was reframed as an inverse problem by extending film thickness as a trainable variable.FindingsALM-PINNs significantly reduced the maximum absolute boundary error by almost 80%. Transfer learning accelerated PINNs for solving the Reynolds equation, reducing training epochs by an order of magnitude. The iterative process for load balancing was effectively eliminated by extending the thickness as a trainable parameter, achieving a maximum percentage error of 2.31%. These outcomes demonstrated strong agreement with FDM results, analytical solutions and experimental data.Originality/valueThis study proposes a PINN-based approach for hydrodynamic lubrication analysis that significantly improves boundary accuracy and the training process. Additionally, it effectively replaces the load balancing procedure. This methodology demonstrates considerable potential for broader applications across various boundary value problems and iterative processes.Peer reviewThe peer review history for this article is available at: https://publons.com/publon/10.1108/ILT-07-2024-0277/

Performance optimization for solar photovoltaic thermal system with spiral rectangular absorber using Taguchi method

Solar collector systems efficiently transform sunlight into energy that may be used to meet various needs. This research aimed to use the Taguchi method to determine the ideal operating parameters for a solar thermal collector with a rectangular spiral absorber. Controllable parameters including mass flow rate, solar radiation, and absorber design were manipulated during the energy recovery process, and features like PV temperature and outlet water temperature were used to assess the system’s effectiveness. The findings indicate that certain criteria significantly affect response indicators. The observed percentage contribution of absorber design, solar radiation, and the mass flow rate was 69.19%, 27.99%, and 2.83% in PV surface temperature. In comparison, the individual percentage contributions were 73.63%, 13.51%, and 10.57% for absorber design, solar radiation and mass flow rate for water output temperatures. The present model’s R2 values for PV and outlet water temperatures are 97.24% and 99.67%, respectively. The Predictive regression model was found in fine harmony and the maximum percentage error is limited to 0.68%. The maximum analytical electrical efficiency was observed with a spiral rectangular absorber of 14.57% at the lowest mass flow rate of 0.04 kg/s at the lowest radiation level of 600 W/m2. In comparison, maximum analytical thermal efficiency was observed with a spiral rectangular thermal absorber of 63.56% at the highest flow rate of 0.06 kg/s and the highest solar radiation level of 1000 W/m2. The analytical and experiment findings were in better agreement in this study, with the highest relative error of 7.52%. According to the study’s findings, the rectangular absorber-based PVT system is at its best at a higher mass flow rate to lower PV temperature and boost thermal energy recovery via water. The present research work can be extended for exergy, environmental, and economic feasibility analysis.

Forward design method for the design of panda polarization-maintaining few-mode optical fiber based on artificial neural network.

Panda polarization-maintaining few-mode optical fiber (PPMFMOF) has important research significance in the short distance optical transmission field owing to its advantages of weak nonlinear effects, which is benefit to reduce the use of digital signal processing equipment. Designing a high-performance PPMFMOF quickly and efficiently is expected and yet challenging. In this article, we demonstrated a forward design method for the design of PPMFMOF based on artificial neural network (ANN) to solve the problems of inefficient and time-consuming PPMFMOF design in traditional design method. By studying the influence of different ANN models on the fiber performance, the approximate range of the optimal value was obtained in advance, then the minimum effective refractive index difference (Δneff,min) between adjacent LP modes was used as the optimization object, finally design of PPMFMOF supporting 10 LP modes in C + L band was successfully realized. This method provided low time-consuming, high-efficiency and high-accuracy for the fast design of PPMFMOF and the maximum mean absolute percentage error (MAPE) of the ANN model to predict the effective refractive index (neff) of 10 LP modes is only 3.2211 × 10-7. We believe that the proposed method could also be quickly and accurately applied to other functional optical fiber designs.

Prediction of thick plate properties of CLT and innovative panels through machine learning algorithms

Determining the membrane, bending, and shear stiffness of cross-laminated timber (CLT) and innovative panels requires advanced finite element computations and homogenization techniques based on thin and thick plate theories. These computations may not be easily implemented by engineers. To address this issue, this paper implements and makes accessible a machine-learning model that directly predicts the linear elastic properties of CLT and innovative panels. First, a large database of plate stiffness moduli is built from finite element computations taking into account microstructural characteristics such as the layers’ thicknesses, board width, gaps width, longitudinal modulus of elasticity, and longitudinal and rolling shear stiffness moduli of timber. Second, three approximations are built and investigated: closed-form solutions, an artificial neural network trained on the database, and another artificial neural network that uses the closed-form solutions as prior knowledge before being trained on the data. The results demonstrate the superiority of artificial neural networks with prior knowledge and a very satisfying accuracy for engineering applications. The root mean square percentage error (RMSPE) values range from 0.23% to 3.01%, and the maximum absolute percentage error (MaxAPE) values range from 1.12% to 20%.

Combining Multi-Indirect Features Extraction and Optimized GPR Algorithm for Online State of Health Estimation of Lithium-ion Batteries

Abstract With the wide application of lithium batteries (LIBs) in electrified transportation and smart grids, especially in the pure electric vehicle industry, the accurate health maintenance monitor of LIBs has emerged as critical for safe battery operation. Although many data-driven methods with state of health (SOH) estimation for LIBs have been proposed, the problems of industrial generation and computational cost still need to be improved further. In contrast, this paper carried out a low-complexity SOH estimation method for LIBs. Specifically, the seven health indicators are extracted firstly to characterize battery health status from voltage, current, temperature and other data that can be obtained online. Then, the optimized gaussian process regression (GPR) algorithm is proposed with proper computational cost. Ultimately, combining a multi-indirect features extraction and optimized GPR algorithm, the online SOH estimation for LIBs was established and verified with NASA experiment data. The experimental results show that the maximum Mean Absolute Percentage Error (MAPE) of SOH estimation from the proposed method is 1.4496 and the minimum MAPE only reaches 0.5635. More importantly, the optimized GPR for SOH estimation can achieve a maximum 65.37% improvement under multiple evaluation criteria compared to traditional GPR. The method proposed in this paper is helpful for realizing on-line SOH estimation in battery management systems.

Optimizing High-Performance Predictive Modeling of the Medium-Speed WEDM Processing of Inconel 718

The purpose of this research was to create a predictive model for a medium-speed wire electrical discharge machine (WEDM) utilizing an artificial neural network (ANN). Medium-speed WEDM experiments were developed based on the I-optimal mixture design for machining, the Inconel 718 superalloy. During the experiment, the input parameters were the spark ontime, spark offtime, wire feed, and current, with the material removal rate (MRR) and surface roughness (Ra) selected as performance indicators. The ANN model was trained on experimental data and built using a feed-forward backpropagation neural network with a (4-8-2) structure and the Bayesian regularization (BR) learning approach. The model correctly predicted the relationship between the medium-speed WEDM’s primary process parameters and machining performance. An integrated ANN model and the Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) were used to determine the ideal parameters for the MRR and Ra, resulting in a set of Pareto-optimal solutions. The confirmation experiment revealed that the mean prediction error between the experimental and ideal solutions had a maximum error percentage of 1% for the MRR and 2% for the Ra, which are within acceptable ranges. This showed that the best process–parameter combinations were better for the MRR and Ra.

Inverse design of cellular structures with the geometry of triply periodic minimal surfaces using generative artificial intelligence algorithms

Triply periodic minimal surfaces (TPMS) exhibit excellent mechanical and energy absorption properties due to their structural advantages. However, existing porous TPMS structural design methods are constrained to a forward process from structural parameters to mechanical properties. This study proposed an inverse design method that combines bidirectional generative adversarial networks (BiGAN) and mechanical performance targets, resulting in a combined TPMS structure of Primitive and IWP types with superior buffering and energy absorption capabilities. The results show that under a single load value target condition of the designed structure, the minimum deviation index (R2) between the load value corresponding to the displacement point and the target load value is only 0.987, and the maximum mean absolute percentage error (MAPE) is only 5.92 %. When considering the elastic modulus target, the approach successfully conducts two sets of combined structural designs meeting the requirements of both high and low elastic moduli. When targeting the specified load-displacement curve conditions, specifically when combining high elastic modulus with ascending plasticity, the designed structures exhibit an error of only 2.2 % compared to the target property. Moreover, the quasi-static uniaxial compression experiments conducted on additively manufactured designed structures confirm that the experimental curves match the target curves in terms of deformation trends and load value ranges. The success of this inverse design approach for cellular TPMS structures has the potential to expedite new structural material development processes.

Regression prediction of critical exhaust volumetric flow rate in tunnel fire with two-point extraction ventilation based on neural network method

In this study, a Genetic Algorithm-Backpropagation Neural Network (GA-BPNN) model was developed to predict critical exhaust volumetric flow rate in tunnel fires with two-point extraction ventilation system. Seven influencing factors served as inputs for the model, while the critical exhaust volumetric flow rate was the output. Experimental data from two reduced-scale tunnels were utilized to both train and validate the GA-BPNN model. The results demonstrate a satisfactory prediction performance, with key metrics such as Mean Absolute Percentage Error (MAPE), Relation Coefficient (R2), and Root Mean Square Error (RMSE) achieving values of 0.0384, 0.9989, and 3.5258, respectively. Importantly, the model maintains a maximum Absolute Percentage Error (APE) below 10 %. In contrast, the traditional BPNN model exhibited an APE of approximately 18.9 %. Moreover, the predictive results of the GA-BPNN model were compared with those of a previously semi-empirical formula. It was observed that the semi-empirical predictions exhibited APE exceeding 20 % for certain data points. This further underscores the superiority of the GA-BPNN model in predicting critical exhaust volumetric flow rates in tunnel fires with two-point ventilation system. These findings affirm the feasibility and effectiveness of GA-BPNN regression model as a reliable method for predicting critical exhaust volumetric flow rates under multiple factors.

Enhanced fractional-order total variation regularization-based velocity field reconstruction for CUP-VISAR diagnostic system.

The fusion of a velocity interferometer system for any reflector with compressed ultrafast photography systems in recent literature can achieve two-dimensional spatiotemporal diagnosis of shock wave velocities. Addressing the limitations posed by 7 × 7 coded aperture sampling, this study introduces an enhanced three-dimensional reconstruction algorithm grounded in fractional-order total variation regularization (E-3DFOTV). Simulated reconstructions and analysis were conducted on 80 frames of 350 × 800 fringes. The results show that compared with TWIST, ADMM, and E-3DTV, the average PSNR of the E-3DFOTV algorithm is increased by 16.81 dB, 14.46 dB, and 2.98 dB, respectively, and the average SSIM of the E-3DFOTV algorithm is increased by 53.20%, 27%, and 3.19%, respectively. Moreover, the reconstruction time consumption of E-3DFOTV is reduced by 33.48% compared with the E-3DTV algorithm and 2.94% compared with the ADMM algorithm. The two-dimensional distribution of shock wave velocity fields reconstructed using E-3DFOTV exhibits minimal errors, with percentages within 1.67%, 1.00%, and 2.14% at different slices, respectively. Moreover, the experiment was conducted on the ShenGuang-III prototype laser facility and VISAR data has been reconstructed in 1.25 ns range. Reconstruction results from experimental data demonstrate that the percentage errors at maximum velocity location for ADMM, E-3DTV, and E-3DFOTV are 12.08%, 19.27%, and 3.59%, and the maximum percentage error for E-3DFOTV is 6.65%, underscoring the feasibility of the algorithm.

Accounting for taxi service conditions in estimating route travel time from floating car data using Markov chain model

Recognising the variations in driving behaviour between taxis in the empty and carry conditions is pivotal for enhancing the accuracy of route travel time estimations using floating car data. However, existing methods largely overlook this distinction. In light of this, this study aims to harness these variations for more precise estimations. Utilising taxi data, we segmented the information by service conditions and executed distinct estimations for each segment. The route travel time was deduced through convolutional operation, complemented by a Markov chain model to discern correlations between travel times across various links. Our innovative approach realised a substantial enhancement in accuracy. Notably, when accounting for distinct service conditions, there was a reduction of 51.44% in mean absolute error and a 46.83% decline in maximum percentage error. By providing more accurate and reliable travel time predictions, our methodology enables better-informed traffic management decisions. Accurate travel time estimations are essential for optimising traffic signal timings, planning efficient routing strategies, and managing road network usage. These improvements in traffic management can lead to smoother traffic flow, reduced travel times, and ultimately, diminished congestion in urban areas.

New insight into groundwater 4He ages based on Ne isotopic equilibrium in Jianghan Plain, Central China

The change of hydrostatic pressure caused by the fluctuation of groundwater table in the aquifer will lead to the partial dissolution of excess 4He gas, resulting in the isotope imbalance of 3He/4He-4He. The dissolved Ne in groundwater is mainly derived from the atmosphere, and its isotopic composition can correct the isotopic imbalance of 3He/4He-4He. We collected thirty-eight groundwater samples from the second aquifer of the Jianghan Plain, and the isotopic concentrations and ratios of He and Ne were measured. 21Ne/22Ne-20Ne/22Ne illustration is proposed to estimate the shares of atmospheric and mantle components. The 21Ne content and isotopic ratios of atmospheric and mantle components are used to estimate a calculated Ne content. The difference between the calculated Ne content and the measured Ne content (∆Ne) is used to evaluate the percentage of error estimated“excess air”. The accumulation of crustal 4He is corrected with the measured 4He content and the percentage of error estimated “excess air”. We found the maximum percentage of error estimated “excess air” was 7.57 % occurring in the groundwater samples from the second aquifer of Jianghan Plain, and the disequilibrium of 3He/4He-4He led to overestimation of the share of mantle He. The percentage of mantle He in total dissolved components is reassessed and range from 0.03 % to 0.74 %, indicating the mantle component is minor. The reassessed 4He ages (1.79 ka to 21.90 ka) were uniformly older than those estimated by traditional method which only use the measured 3He/4He ratio to distinguish the crust 4He (1.28 ka to 18.74 ka). 4He age is significantly underestimated up to 47.05 %.

Bayesian Approach to Estimate Proglacial Lake Volume (BE‐GLAV)

AbstractWe present a new model called Bayesian Estimated Glacial Lake Volume (BE‐GLAV) to estimate the volume of proglacial lakes. Presuming the lake cross‐section as trapezoidal, BE‐GLAV uses a Bayesian calibration approach to adjust the cross‐sectional geometry to match modeled and observed lake surface widths. We validated our model using bathymetric measurements from lakes spread across High Mountain Asia (specifically, the Himalaya and Tien‐Shan), with aerial extents ranging from 0.01 to 5.5 km2. The modeled lake volumes agreed with the measured lake volume with a root‐mean‐square absolute uncertainty of ∼14%. With minimum and maximum errors of ∼0.3% and ∼61.2%, BE‐GLAV performed well compared to 10 other models in a model inter‐comparison experiment. Using the measured set of volumes, our model can constrain both the root mean square (RMS) error and the maximum percentage error in modeled lake volume, unlike other models, some of which can compute just the RMS uncertainty.

A fast wide-range air balancing control method based on deep reinforcement learning

Air balancing is a crucial technology for reducing energy consumption in ventilation ducts and improving indoor environmental quality. Existing air balancing methods face challenges due to the complex and diverse internal duct topologies of ventilation duct systems, as well as the strong coupling of airflow between branches, making air balancing tuning difficult. To address the slow convergence speed and difficulty in precise control of air balancing in multi-branch ventilation systems, this paper proposes a fast wide-range air balancing control method based on deep reinforcement learning (DRL-AB). A system airflow control process with Markovian properties is designed, with state, action, and reward functions oriented towards reducing airflow errors. A dynamic single-branch target training mechanism is designed to reduce training costs while improving the agent’s training efficiency and its ability to generalize to different target airflow levels. The proposed method’s performance under various target airflow conditions is verified through experimental platforms. The results indicate that, in all test cases, the maximum percentage error is controlled within 3.33%, ensuring rapid and accurate convergence of a wide range of target airflows. This demonstrates strong universality, prevents waste of airflow energy, and enhances energy utilization efficiency.