Sensitivity Of Model Parameters Research Articles

Experimental investigation of the creep behaviour of remoulded loess under different levels of compactness.

Generally, a loess high fill project will undergo a time-dependent deformation and settlement process for a long time after the initial fill. Understanding the creeping behaviour of compacted loess is an important part of determining the stability of a compacted loess foundation. To study the creep behaviour of remoulded loess under different levels of compactness, we performed triaxial shear and triaxial creep tests using Q2 loess specimens obtained from the new district of Yan’an city. Based on laboratory test results, the triaxial shear and creep characteristics of remoulded loess under different levels compactness are summarised. The regularity of instantaneous strain, creep strain, total accumulated strain and initial shear modulus were analysed and the relationship between the compactness and long-term strength of remoulded loess is provided. It was found that the remoulded loess becomes harder and its long-term strength increases with an increase in compactness. Furthermore, we propose a new creep model (HD), based on the hardening-damage mechanism, and have derived one-dimensional (1D) and three-dimensional (3D) creep equations based on this new creep model. This new creep model is flexible enough to fit the typical creep test curves of remoulded loess, while perfectly describing the tertiary creep stage. Finally, the sensitivity of the HD creep model parameters was analysed; the results indicate that the parameters denoted as α, γ, and β significantly affect the morphological changes and various stage characteristics are represented by the creep curve.

The sensitivity of simulated streamflow to individual hydrologic processes across North America

Streamflow sensitivity to different hydrologic processes varies in both space and time. This sensitivity is traditionally evaluated for the parameters specific to a given hydrologic model simulating streamflow. In this study, we apply a novel analysis over more than 3000 basins across North America considering a blended hydrologic model structure, which includes not only parametric, but also structural uncertainties. This enables seamless quantification of model process sensitivities and parameter sensitivities across a continuous set of models. It also leads to high-level conclusions about the importance of water cycle components on streamflow predictions, such as quickflow being the most sensitive process for streamflow simulations across the North American continent. The results of the 3000 basins are used to derive an approximation of sensitivities based on physiographic and climatologic data without the need to perform expensive sensitivity analyses. Detailed spatio-temporal inputs and results are shared through an interactive website.

An Accelerated Thrombosis Model for Computational Fluid Dynamics Simulations in Rotary Blood Pumps

PurposeThrombosis ranks among the major complications in blood-carrying medical devices and a better understanding to influence the design related contribution to thrombosis is desirable. Over the past years many computational models of thrombosis have been developed. However, numerically cheap models able to predict localized thrombus risk in complex geometries are still lacking. The aim of the study was to develop and test a computationally efficient model for thrombus risk prediction in rotary blood pumps.MethodsWe used a two-stage approach to calculate thrombus risk. The first stage involves the computation of velocity and pressure fields by computational fluid dynamic simulations. At the second stage, platelet activation by mechanical and chemical stimuli was determined through species transport with an Eulerian approach. The model was compared with existing clinical data on thrombus deposition within the HeartMate II. Furthermore, an operating point and model parameter sensitivity analysis was performed.ResultsOur model shows good correlation (R2 > 0.93) with clinical data and identifies the bearing and outlet stator region of the HeartMate II as the location most prone to thrombus formation. The calculation of thrombus risk requires an additional 10–20 core hours of computation time.ConclusionThe concentration of activated platelets can be used as a surrogate and computationally low-cost marker to determine potential risk regions of thrombus deposition in a blood pump. Relative comparisons of thrombus risk are possible even considering the intrinsic uncertainty in model parameters and operating conditions.

Urban Traffic Congestion State Recognition Supporting Algorithm Research on Vehicle Wireless Positioning in Vehicle–Road Cooperative Environment

Vehicle–road cooperative technology applies wireless communication and a new generation of internet technology to urban traffic management, providing an effective way to solve urban traffic congestion and improve traffic efficiency. This article researches the vehicle wireless positioning fusion algorithm, suitable for the actual vehicle–road collaborative environment, which is an important step of urban traffic congestion state recognition. First, based on the error correction of existing wireless positioning algorithms, a weighting indicator considering distance and positioning compound errors is designed, and a vehicle wireless positioning fusion algorithm based on error weighting to eliminate line-of-sight (LOS) and non-line-of-sight (NLOS) error is proposed. Secondly, the wireless positioning fusion algorithm is verified based on accuracy evaluation indicators such as root mean square error (RMSE), Cramer Rao lower bound (CRLB), geometric differentiation of precision (GDOP), and cumulative distribution probability (CDP), and the sensitivity of the distance propagation model parameters to the positioning error is analyzed. The verification results show that the local vehicle wireless positioning fusion algorithm proposed in this article could be useful to locate vehicles in an actual vehicle–road collaborative environment. The positioning accuracy could reach 46.31 m with 67% probability, while the positioning accuracy could reach 122.53 m with 95% probability. The average positioning accuracy could reach 39.97 m. Compared with the two types of wireless positioning methods based on ranging and non-ranging methods, the positioning accuracy is improved by 7.74% and 17.69%. The algorithm can either use the roadside base stations to carry out the individual vehicle positioning or cooperate with GPS positioning and trilateral positioning to make up for the positioning blind spots caused by the lack of signal, interference, or base station overload in the urban complex road environment and, furthermore, improves the robustness of vehicle positioning. The results could assist in all-day real-time traffic congestion state recognition and other actual scenarios.

Hydrological modeling for streamflow and sediment yield simulation using the SWAT model in a forest-dominated watershed of north-eastern Himalayas of Kashmir Valley, India

Abstract In this study, the Soil and Water Assessment Tool (SWAT) model was used to examine the spatial variability of sediment yield, quantify runoff, and soil loss at the sub-basin level and prioritize sub-basins in the Sindh watershed due to its computational efficiency in complex watersheds. The Sequential Uncertainty Fitting-2 approach was used to determine the sensitivity and uncertainty of model parameters. The parameter sensitivity analysis showed that Soil Conservation Services Curve Number II is the most sensitive model parameter for streamflow simulation, whereas linear parameters for sediment re-entrainment is the most significant parameter for sediment yield simulation. This study used daily runoff and sediment event data from 2003 to 2013; data from 2003 to 2008 were utilized for calibration and data from 2009 to 2013 were used for validation. In general, the model performance statistics showed good agreement between observed and simulated values of streamflow and sediment yield for both calibration and validation periods. The noticed insights of this research show the ability of the SWAT model in simulating the hydrology of the Sindh watershed and its reliability to be utilized as a decision-making tool by decision-makers and researchers to influence strategies in the management of watershed processes.

A Mathematical Model of the Transmission Dynamics of Bovine Schistosomiasis with Contaminated Environment.

Schistosomiasis, a vector-borne chronically debilitating infectious disease, is a serious public health concern for humans and animals in the affected tropical and sub-tropical regions. We formulate and theoretically analyze a deterministic mathematical model with snail and bovine hosts. The basic reproduction number [Formula: see text] is computed and used to investigate the local stability of the model's steady states. Global stability of the endemic equilibrium is carried out by constructing a suitable Lyapunov function. Sensitivity analysis shows that the basic reproduction number is most sensitive to the model parameters related to the contaminated environment, namely: shedding rate of cercariae by snails, cercariae to miracidia survival probability, snails-miracidia effective contact rate and natural death rate of miracidia and cercariae. Numerical results show that when no intervention measures are implemented, there is an increase of the infected classes, and a rapid decline of the number of susceptible and exposed bovines and snails. Effects of the variation of some of the key sensitive model parameters on the schistosomiasis dynamics as well as on the initial disease transmission threshold parameter [Formula: see text] are graphically depicted.

Uncertainty and sensitivity analysis of algal-bacterial model under different ranges of parameter variation

Based on the Activated Sludge Model No.3, a comprehensive model consisting of autotrophic bacteria, algae, and heterotrophic bacteria was presented in this work to describe the relationship among microorganisms and substrates in photo-sequencing batch reactors. The proposed model, which is called the algal-bacteria model, has involved a total of 31 parameters. The standard regression coefficient method was used to conduct global sensitivity analysis of the algal-bacteria model proposed was used to reduce the experiment time and energy of the experimenters, which determined that eight sensitivity parameters from analysis results according to the absolute value of sensitivity indices SRC. Analyzing the influence of different parameter value ranges on the sensitivity analysis, it provides a reference for the parameter sensitivity and uncertainty analysis of the algae-bacteria model. The results show that the algal-bacteria model simulates the removal process of complex nutrients well, and the parameter values affect the sensitivity analysis, but the identification of the most sensitive parameters will not be affected. Moreover, in terms of predicting the removal of dissolved oxygen and pollutants, the simulation results were proved to be in good agreement with the experimental data. It was demonstrated that the established model could be used as a valuable tool for optimizing the parameters of photo-sequencing batch reactors. Key findings of this paper include (1) representation of a comprehensive model to describe the interactions among bacteria, algae, and substrates; (2) assessment of the reliability of model parameters; (3) adoption of a novel method to estimate the most sensitive parameters and evaluate the influence of the change of parameters on the sensitivity analysis results. Our findings suggest that the application of the algal-bacterial consortia system may provide a viable technology in wastewater treatment.

A Differentially Private Framework for Deep Learning With Convexified Loss Functions

Differential privacy (DP) has been applied in deep learning for preserving privacy of the underlying training sets. Existing DP practice falls into three categories - objective perturbation, gradient perturbation and output perturbation. They suffer from three main problems. First, conditions on objective functions limit objective perturbation in general deep learning tasks. Second, gradient perturbation does not achieve a satisfactory privacy-utility trade-off due to over-injected noise in each epoch. Third, high utility of the output perturbation method is not guaranteed because of the loose upper bound on the global sensitivity of the trained model parameters as the noise scale parameter. To address these problems, we analyse a tighter upper bound on the global sensitivity of the model parameters. Under a black-box setting, based on this global sensitivity, to control the overall noise injection, we propose a novel output perturbation framework by injecting DP noise into a randomly sampled neuron (via the exponential mechanism) at the output layer of a baseline non-private neural network trained with a convexified loss function. We empirically compare the privacy-utility trade-off, measured by accuracy loss to baseline non-private models and the privacy leakage against black-box membership inference (MI) attacks, between our framework and the open-source differentially private stochastic gradient descent (DP-SGD) approaches on six commonly used real-world datasets. The experimental evaluations show that, when the baseline models have observable privacy leakage under MI attacks, our framework achieves a better privacy-utility trade-off than existing DP-SGD implementations, given an overall privacy budget $\epsilon \leq 1$ for a large number of queries.

Stability Evaluation of Proppant in Fractures of Gas Storage in Yulin Gas Field

In nearly a hundred years of construction, underground gas storage has become the main natural gas storage and peak regulation means in the world. For the gas storage, the large production and high flow rate of the gas well in the actual production process will cause the backflow of proppant filled in the supporting fracture, which will bring great harm to the gas field production. In this article, when the proppant fracture reaches a stable state in the process of gas injection and production, the stress of proppant particles is analyzed, the critical velocity of proppant reflux is calculated, and then the critical production model is established; calculate the permeability change during proppant migration, then calculate the fluid velocity and production, and determine the fluid velocity range of injection and production wells in gas storage. The parameter sensitivity of velocity and flow model is analyzed. The results show that with the increase of closure stress, the critical gas flow and critical gas velocity of proppant backflow gradually increase, and the proppant filling layer is more stable. The smaller the thickness and width of the filling layer, the greater the critical gas flow and critical gas velocity of proppant backflow, the more stable the proppant filling layer, and the lesser the chance of backflow. The higher the saturation, the lower the critical gas velocity, and the more prone the proppant to reflux. It has important guiding significance for realizing the optimization of gas well production and maintaining efficient production efficiency.

Global Sensitivity Analysis of the MEMLS Model for Retrieving Snow Water Equivalent

Sensitivity analysis (SA) of model parameters is of great importance for understanding, development, and application of models. However, the influence of snow microstructure variability on snow water equivalent retrieval from passive microwave measurements is still unclear. This article explores the parameter sensitivity of the microwave emission model of layered snowpacks (MEMLS) with improved born approximation (IBA) by using a quantitative global SA method, the extended Fourier amplitude sensitivity test (EFAST) algorithm. A deep analysis is conducted, including the sensitivity of passive microwave emission to snow parameters, the sensitivity variation analysis for different snow conditions, and the temporal properties of the parameter sensitivity. The results show the exponential correlation length, snow depth, and snow density are the three most sensitive parameters for snow without salt in the MEMLS model for the brightness temperature gradient at 18.7 and 36.5 GHz. For snow with a small salt content, the exponential correlation length, snow depth, snow temperature, and snow density are the four most sensitive parameters. Second, snow parameter variability highly affects the microwave radiation. The sensitivity values of microwave brightness temperature to snow depth gradually increase when the exponential correlation length is less than 0.25 mm and then slightly decreases with the increase of exponential correlation length and decreases along with the increase of snow density. Finally, our analysis highlights the importance to include the snow density, especially for deep snow depth, in the combination of sensitive factors in future multiparameter retrievals.

Risk-based flood adaptation assessment for large-scale buildings in coastal cities using cloud computing

• A risk-based flood adaptation model was developed and investigated in the cloud computing system • Both global and local sensitivity analysis were conducted to show sensitivity of model parameters • Adaptation scenarios were designed to incorporate life-cycle cost and benefits of adaptation, social vulnerability, and public adaptation policies • Public adaptation with seawall and enforced building elevation policy within flood zones could substantially reduce building damages Flood risk management (FRM) in coastal cities is a challenging task due to uncertain climate hazards under sea-level rise (SLR) and large-scale vulnerable buildings within in the floodplain. This study presents a building-level adaptation framework to evaluate alternative community adaptation strategies relying on cloud computing. We incorporated multiple sources of model uncertainties and randomly generated storm surges in each year of simulation using the extreme value distribution (GEV) theory. Based on a case study in Miami-Dade County, Florida, four adaptation scenarios were designed to evaluate their effectiveness in adaptation. Our sensitivity analysis suggested a positive linear relationship between community flood risk reduction and total community adaptation costs in the life-cycle cost-benefit (LCCB) model. Our results showed that uncertainties of the total community damage based on the LCCB model ranges from $ 221 million to $ 2.75 billion. However, when considering social vulnerability, the total community damage increased substantially, ranging from $ 244 million to $ 3.44 billion. Nevertheless, a 6ft public seawall based on the upper bound of the GEV distribution with the enforced building elevation policy in flood zones could substantially reduce community flood damage under uncertain sea-level rises.

Separation of succinic acid from fermentation broth: Dielectric exclusion, Donnan effect and diffusion as the most influential mass transfer mechanisms

Nanofiltration can be used to separate succinic acid (SA) from other organic acids in a fermentation broth. However, commercial nanofiltration membranes can be insufficiently selective since some produced organic acids have similar molecular weights. SA production from biomass fermentation entails separation of succinic acid from the other concomitantly produced organic acids such as acetic acid, formic acid, pyruvic acid etc. In this study, five commercial nanofiltration membranes were tested for SA rejection and selectivity under dead-end and crossflow modes of SA recovery from selected synthetic solutions and a real fermentation broth. SA rejections up to 94.1 ± 1.5% and 95.5 ± 3.5% were observed for the synthetic broth and the fermentation broth, respectively. The SA fractions obtained in the retentate were nearly 70% and 60% for synthetic and fermentation broth, respectively. The highest rejections and SA fraction values were achieved with DK and NF270 membranes, which also showed the highest permeate flux. Model evaluation using the Donnan-steric pore model with dielectric exclusion (DSPM-DE) showed that 99% of the SA rejection occurs at the membrane interface and not inside the membrane. Based on the modelling, the most important mechanism affecting separation was dielectric exclusion at both sides of the membrane surfaces. Moreover, identified sensitive model parameters, i.e. pore dielectric constant (εpore), charge density (Xd) and pore radius (rp), were estimated through model fitting and were found to be highly correlated with significant variations. Finally, for the first time, a calibrated model with synthetic broth was successfully used to predict organic acids rejection for a real fermentation broth (with 12% and 3% prediction error for succinate rejection by NF270 and DK, respectively). Such prediction was achieved only by measuring organic acid concentrations in the fermentation broth and using the three sensitive parameters. Furthermore, the potential contribution of different transport mechanisms of SA was quantified for the first time.

Enabling high-fidelity electrochemical P2D modeling of lithium-ion batteries via fast and non-destructive parameter identification

Physics-based electrochemical models provide insight into the battery internal states and have shown great potential in battery design optimization and automotive and aerospace applications. However, the complexity of the electrochemical model makes it difficult to obtain parameter values accurately. In this study, a novel non-destructive parameter identification method is proposed to parameterize the most commonly used electrochemical pseudo-two-dimensional model. The whole identification process consists of three key steps. First, in order to find the optimal identification conditions, the sensitivity of model parameters is analyzed, and parameters are classified into three types according to their most sensitive conditions. Second, feasible initial guess values of these unknown parameters are obtained using a deep learning algorithm, which can not only help avoid the divergence problem of the identification algorithm but also speed up the subsequent identification process. Finally, two different approaches are combined and used for parameter identification, and parameters that have high sensitivity are estimated in a step-wise manner. We show that 14 electrochemical parameters can be estimated accurately within 1 h using simulation and experimental data. After estimating the model parameters, the root-mean-square error of the predicted voltage from the model is less than 14 mV.

Evaluation of the Number of Visits to Chinese Medical Institutions Using a Logistic Differential Equation Model

In this study, we established a two-dimensional logistic differential equation model to study the number of visits in Chinese PHCIs and hospitals based on the behavior of patients. We determine the model's equilibrium points and analyze their stability and then use China medical services data to fit the unknown parameters of the model. Finally, the sensitivity of model parameters is evaluated to determine the parameters that are susceptible to influence the system. The results indicate that the system corresponds to the zero-equilibrium point, the boundary equilibrium point, and the positive equilibrium point under different parameter conditions. We found that, in order to substantially increase visits to PHCIs, efforts should be made to improve PHCI comprehensive capacity and maximum service capacity.

Reconstruction of airflow path parameters in multizone models based on Bayesian inference and measured data

The accurate inference of model parameters is important for predicting contaminant distributions in multizone environments, which is critical for ensuring indoor personnel health and formulating an optimistic ventilation strategy. However, several important airflow path parameters (flow coefficients, flow exponents, etc.) are not directly observed or observable. Therefore, in this study, we proposed a reconstruction method to estimate the parameters for the multizone model based on measured data. MATLAB codes were programmed to call the CONTAM (a multizone simulation program developed by the US National Institute of Standards and Technology (NIST)) engine to simulate the pollutant transport in a multizone scaled building model. The sensitivities of different model parameters were evaluated through a Global Sensitivity Analysis (GSA). Subsequently, a Bayesian inference algorithm was applied to iteratively determine the unknown parameters. Finally, multizone scaled experiments with different forms of pollutant sources were conducted to validate the reconstruction method. The pollutant concentrations predicted using the measured parameters were compared with the those obtained using the reconstructed parameters. The results showed that the concentrations predicted using the reconstructed parameters agreed well with the measured data under constant source (CS) release and most dynamic source (DS) release in different experimental cases. The simulation results based on the reconstructed parameters were more accurate than those based on the directly measured parameters. The proposed reconstruction method can conveniently construct digital models that are equivalent to physical buildings, and these models can be further used to conduct analysis of built environments.

Comparison of big-leaf and two-leaf light use efficiency models for GPP simulation after considering a radiation scalar

Light use efficiency (LUE) models, mainly including the big-leaf (BL) and two-leaf (TL) categories, are efficient approaches to simulate gross primary productivity (GPP). Recently, a TL LUE model considering a radiation scalar (RTL-LUE) was developed, which improved the GPP simulation and unified the same model structure with the BL models using only one maximum LUE. However, whether the radiation scalar is suitable for BL models is still unknown yet, and the core parameters of maximum LUE in the BL and TL models have not been fairly compared before. In this study, we step forward to modifying the MOD17 model using the radiation scalar (RMOD17), and compare the GPP simulations and maximum LUEs in BL (MOD17, RMOD17) and TL (RTL-LUE) models at global 169 FLUXNET eddy covariance (EC) sites. Results indicate that the GPP estimation from RMOD17 (R2=0.72) matches better with EC GPP than those from the original MOD17 (R2=0.65), because the radiation scalar effectively corrects the underestimations or overestimations in low or high photosynthetically active radiation (PAR) ranges caused by the radiation-independent LUE in MOD17. The RTL-LUE can further improve the accuracy (R2=0.74) by alleviating the GPP underestimation of BL RMOD17 in high productivity ranges, which is mainly caused by the insufficient consideration of the shaded leaves contributions. The maximum LUE from RMOD17 and RTL-LUE both show more reasonable values and lower diurnal variations than MOD17, and further analysis proves the radiation scalar is the main reason for this. Besides, RTL-LUE presents a lower model parameter sensitivity compared to RMOD17, indicating that TL is a more robust strategy than BL to simulate GPP. This study highlights the importance of considering a radiation scalar and two-leaf strategy in GPP simulation in LUE models, to better describe the multi-order impacts of radiation on vegetation photosynthesis.

Identification and evaluation of processes for joint disposal of high level radioactive waste and low to intermediate level radioactive waste

Abstract. The on-going research project „Identification and evaluation of processes that can arise by disposing of both high level radioactive waste (HAW) and low to intermediate level radioactive waste (LAW/MAW) at the same site“ (GemEnd, FKZ 4719F10401), commissioned by the Federal Office for the Safety of Nuclear Waste Management (BASE), is concerned with the question which thermal, hydraulic, mechanical, chemical and biological (THMCB) processes could be of importance for the long-term safety of the geological repository for high level radioactive waste. The focus of the project is on mutual influences between the HAW and LAW/MAW repositories, which should be constructed separately according to the Safety Regulations (Federal Ministry for the Environment, Nature Conservation and Nuclear Safety, BMU, 2020). A second point of emphasis is on processes that could result from the disposal of small amounts of LAW/MAW within the HAW repository. The analyses carried out for each of the potential host rocks clay rock, rock salt and crystalline rock as well as for a combination of clay rock above crystalline rock at a generic site are divided into a qualitative and a quantitative part. As for the qualitative analyses, all potentially relevant processes are identified and evaluated as to whether they are negligible or principally relevant for the exemplary repository configurations considered and according to the current state of knowledge. With regard to the quantitative analyses, the possible extent of potentially safety-relevant processes is illustrated by means of coupled numerical simulations. Of special interest are the effects of particularly sensitive model approaches and/or parameters and notably of the distances between the HAW and LAW/MAW repositories in the different exemplary repository configurations considered. From the results of the quantitative and qualitative analyses, knowledge gaps will be identified and the possibility of their reduction by research and development activities will be discussed. Furthermore, aspects of the transferability of the results to the German site selection procedure will be illuminated. At the interdisciplinary research symposium safeND selected preliminary results of both the qualitative and quantitative analyses will be presented.

Estimation of The Main Effect and Total Effect of a PBPK Model Based on The Uncertainty of Individual Parameter for Treatment Planning in PSMA Therapy

The purpose of this study was to identify the most important physiologically-based pharmacokinetic (PBPK) model parameters determining the absorbed dose (AD) in prostate-specific membrane antigen (PSMA) therapy. The extended-Sobol’ global sensitivity analysis method was used to analyze the sensitivity of the PBPK model parameters obtained from 3 patients. The investigated PBPK model parameters were the blood flow to the organs, PSMA binding rate, biological release rates, and density of organs receptor. The outputs of extended Sobol method were the main effect Si and the total effect STi of the parameter of interests for each ADs. The sampling strategy of extended Sobol has been implemented based on the mean and covariance matrix of the parameters. From the simulations, the most important parameters which determine the ADs to the kidney was the kidney receptor density (Si=0,4, STi=0,8). For tumors, it was shown that tumor receptor density was the most essential parameter (Si=0,7, STi=0,8). In conclusion, measurement of the blood flow and organ receptor densities might be of interest to improve individualized treatment of PSMA therapy.

Sensitivity Analysis of a Cardio-respiratory Model in Preterm Newborns for the Study of Patent Ductus Arteriosus.

This paper proposes an integrated model of cardio-respiratory interactions in preterm newborns, focused on the study of the patent ductus arteriosus (PDA). A formal model parameter sensitivity analysis on blood flow through the PDA is performed. Results show that the proposed model is capable of simulating hemodynamics in right-to-left and left-to-right shunts. For both configurations, the most significant parameters are associated with mechanical ventricular properties and circulatory parameters related to left ventricle loading conditions. These results highlight important physiological mechanisms involved in PDA and provide key information towards the definition of patient-specific parameters.

Prediction Modeling and Sensitivity Analysis of Robot Bone Milling Temperature Operated by a Doctor

Bone milling is a common method in robot orthopedic surgery. However, excessive milling temperature will cause thermal necrosis of bone cells and tissues. It is necessary to carry out further research and analysis on the robot bone milling process considering the lamina milling skills of spinal surgeons and clinical practice to reduce the damage to bone cells and nearby tissues and obtain good cutting surface quality. Considering the randomness of milling parameters during operation, a prediction method of milling temperature model for ball milling cutter considering the doctor’s surgical skills was proposed based on response surface method. Because of material anisotropy and microstructure difference between the cortical bone and cancellous bone, this paper would analyze the influencing factors in different bone layers to establish the prediction model of milling temperature in the segments of cortical bone and cancellous bone. Also, the influence and distribution of milling parameters on temperature in three cutting modes such as parallel cutting mode, cross cutting mode, and vertical cutting mode in the cortical bone region were analyzed. The parameter sensitivity of the milling temperature prediction model was analyzed by the Sobol method, and the influence of the input parameters on the output milling temperature was analyzed quantitatively.