Model Predictive Stability Research Articles

A method for seismic fault identification based on self-training with high-stability pseudo-labels

The imaging principle of seismic image is different from that of natural image. There are many problems on seismic images, such as limited resolution, complex reflection feature and strong uncertainty, which leads to significant difference in the research emphases of computer vision on seismic image. In response to the above issues, the semi-supervised semantic segmentation method based on self-training with high-stability pseudo-labels is proposed for seismic fault identification. Specifically, the self-training strategy Fault-Seg-ST is proposed to decouple Teacher–Student model predictions by adding random noise to unlabeled images. The selective retraining strategy Fault-Seg-SST is proposed to evaluate the reliability of pseudo-labels based on the stability of model predictions. The contrastive loss function is proposed to learn pixel-level feature representation with intra class affinity and inter class separability. The Fault-Seg-SST has achieved the optimal segmentation performance on both DeepLabV2 (Dice 86%, mIoU 77%) and UNet with ResNet 18 (Dice 83%, mIoU 76%). Experimental results show that the fault identification method based on self-training with high-stability pseudo-labels demonstrates the superiority of selective retraining strategy, and the fault identification model trained with Fault-Seg-SST strategy can achieve the fine-grained fault segmentation.

Global self-sustaining and local inheritance for source-free unsupervised domain adaptation

In this paper, we investigate a practical problem called source-free unsupervised domain adaptation, which adapts a source-trained model to the target domain with unlabeled target data. To address this problem, we propose a novel GlObal self-sustAining and Local inheritance (GOAL) method. GOAL contains three components. (1) A backbone follows a mean teacher scheme. The teacher model serves as a smoothing functionality, facilitating a more consistent convergence of the student model. This capability alleviates the student model’s sensitivity to minor input data variations and enhances the overall robustness of the model. Additionally, disparities in predictions between the student and teacher models can be leveraged to identify potential noise in the data. (2) A Global Consistency Self-Sustaining mechanism for learning a stable, discriminative, and diverse prediction space. On the one hand, we employ neighbor samples and mean-teacher schemes to enhance the discriminability and stability of model predictions. On the other hand, non-neighbor samples are leveraged to augment the diversity of model predictions. Furthermore, to mitigate the impact of potential negative neighbors, we derive a weighting factor by incorporating both neighbor entropy and the top-nd similarity of features. (3) A Local Topology Inheritance mechanism to improve the semantic structure of the feature space. We construct a semantic topology graph based on the output predictions of the teacher model and subsequently transmit the teacher topology to the feature space of the student utilizing a local topology inheritance loss. Combining these three components, GOAL can effectively solve the source-free unsupervised domain adaptation. To the best of our knowledge, GOAL is the first attempt to perform topology inheritance for global consistency domain adaptation. Comprehensive experiments illustrate the effectiveness and superiority of GOAL in addressing source-free unsupervised domain adaptation.

Combining Disease Mechanism and Machine Learning to Predict Wheat Fusarium Head Blight

Wheat Fusarium head blight (FHB) can be effectively controlled through prediction. To address the low accuracy and poor stability of model predictions of wheat FHB, a prediction method of wheat FHB that couples a logistic regression mechanism-based model and k-nearest neighbours (KNN) model is proposed in this paper. First, we selected predictive factors, including remote sensing-based and meteorological factors. Then, we quantitatively expressed the factor weights of the disease occurrence and development mechanisms in the disease prediction model by using a logistic model. Subsequently, we integrated the obtained factor weights into the predictive factors and input the predictive factors with weights into the KNN model to predict the incidence of wheat FHB. Finally, the accuracy and generalizability of the models were evaluated. Wheat fields in Changfeng, Dingyuan, Fengyuan, and Feidong counties, Anhui Province, where wheat FHB often occurs, were used as the study area. The incidences of wheat FHB on 29 April and 10 May 2021 were predicted. Compared with a model that did not consider disease mechanism, the accuracy of our model increased by approximately 13%. The overall accuracies of the models for the two dates were 0.88 and 0.92, and the F1 index was 0.86 and 0.94, respectively. The results show that the predictions made with the logistic-KNN model had higher accuracy and better stability than those made with the KNN model, thus achieving remote sensing-based high-precision prediction of wheat FHB.

Association between systolic blood pressure course and outcomes after stroke thrombectomy

BackgroundSystolic blood pressure (SBP) after endovascular thrombectomy (EVT) for large artery occlusive stroke is dynamic, requiring adaptable early prediction tools for improving outcomes. We investigated if post-EVT SBP course was...

We shape our buildings, but do they then shape us? A longitudinal analysis of pedestrian flows and development activity in Melbourne.

Cities are increasingly promoting walkability to tackle climate change, improve urban quality of life, and address socioeconomic inequities that auto-oriented development tends to exacerbate, prompting a need for predictive pedestrian flow models. This paper implements a novel network-based pedestrian flow model at a property-level resolution in the City of Melbourne. Data on Melbourne's urban form, land-uses, amenities, and pedestrian walkways as well as weather conditions are used to predict pedestrian flows between different land-use pairs, which are subsequently calibrated against hourly observed pedestrian counts from automated sensors. Calibration allows the model extrapolate pedestrian flows on all streets throughout the city center based on reliable baseline observations, and to forecast how new development projects will change existing pedestrian flows. Longitudinal data availability also allows us to validate how accurate such predictions are by comparing model results to actual pedestrian counts observed in following years. Updating the built-environment data annually, we (1) test the accuracy of different calibration techniques for predicting foot-traffic on the city's streets in subsequent years; (2) assess how changes in the built environment affect changes in foot-traffic; (3) analyze which pedestrian origin-destination flows explain observed foot-traffic during three peak weekday periods; and (4) assess the stability of model predictions over time. We find that annual changes in the built environment have a significant and measurable impact on the spatial distribution of Melbourne's pedestrian flows. We hope this novel framework can be used by planners to implement "pedestrian impact assessments" for newly planned developments, which can complement traditional vehicular "traffic impact assessments".

On pre-emptive vehicle stability control

Future vehicle localisation technologies enable major enhancements of vehicle dynamics control. This study proposes a novel vehicle stability control paradigm, based on pre-emptive control that considers the curvature profile of the expected path ahead in the computation of the reference direct yaw moment and braking control action. The additional information allows pre-emptive trail braking control, which slows down the vehicle if the predicted speed profile based on the current torque demand is deemed incompatible with the reference trajectory ahead. Nonlinear model predictive control is used to implement the approach, in which also the steering angle and reference yaw rate provided to the internal model are varied along the prediction horizon, to account for the expected vehicle path. Two pre-emptive stability control configurations with different levels of complexity are proposed and compared with the passive vehicle, and two state-of-the-art nonlinear model predictive stability controllers, one with and one without non-pre-emptive trail braking control. The performance is assessed along obstacle avoidance tests, simulated with a high-fidelity model of an electric vehicle with in-wheel motors. Results show that the pre-emptive controllers achieve higher maximum entry speeds – up to ∼34% and ∼60% in high and low tyre-road friction conditions – than the formulations without preview.

A Multigene Model for Predicting Tumor Responsiveness After Preoperative Chemoradiotherapy for Rectal Cancer

Although preoperative chemoradiotherapy (PCRT) is regarded as a standard treatment for locally advanced rectal cancer, there is no reliable biomarker for predicting responsiveness to PCRT. We aimed to develop a biomarker model for predicting response to PCRT. We included 184 patients who received PCRT followed by surgical resection and categorized them as good responders (complete or near-complete regression) or poor responders (all other patients). Candidate gene mRNAs were isolated from formalin-fixed paraffin-embedded tumor specimens and analyzed using the NanoString nCounter gene expression assay. Stepwise logistic regression analysis was used to select genes in discovery and training phases. A quantitative radio-responsiveness prediction model was developed and validated using internal cross-validation groups, and the model's predictive value was assessed based on the area under the receiver operating characteristic curve (AUC). By comparing the gene expressions between good and poor responders, we created a multigene mRNA model using FZD9, HRAS, ITGA7, MECOM, MMP3, NKD1, PIK3CD, and PRKCB. This panel showed good ability to predict treatment response (AUC: 0.846 for the whole data set). Internal cross-validation was performed to evaluate the model's predictive stability among 3 cohorts, which provided AUC values of 0.808-0.909. The satisfactory diagnostic performance of the radio-response prediction index persisted regardless of other clinicopathologic features such as clinical T or N stage, interval between radiation and surgery, and pretreatment carcinoembryonic antigen levels (P = .001, 95% CI, 0.686-0.905). We developed a multigene mRNA-based biomarker model that allows prediction of rectal cancer response to PCRT, which may help identify patients who will benefit most from PCRT.

Model Predictive Stabilization Control of High-Speed Autonomous Ground Vehicles Considering the Effect of Road Topography

This paper presents a model predictive control (MPC) scheme for the stabilization of high-speed autonomous ground vehicles (AGVs) considering the effect of road topography. Accounting for the road curvature and bank angle, a single-track dynamic model with roll dynamics is derived. Variable time steps are utilized for vehicle model discretization, enabling collision avoidance in the long-term without compromising the prediction accuracy in the near-term. Accordingly, safe driving constraints including the sideslip envelope, zero-moment-point and lateral safety corridor are developed to handle stability and obstacle avoidance. Taking these constraints into account, an MPC problem is formulated and solved at each step to determine the optimal steering control commands. Moreover, feedback corrections are integrated into the MPC to compensate the unmodeled dynamics and parameter uncertainties. Comparative simulations validate the capability and real-time ability of the proposed control scheme.

Model Predictive Stabilization of Constrained Underactuated Autonomous Underwater Vehicles With Guaranteed Feasibility and Stability

This paper studies the model predictive stabilization problem of underactuated autonomous underwater vehicles (AUVs) with control input constraints, where the full dynamic and kinematic model including Coriolis and centripetal matrix, the hydrodynamic added mass, and damping is utilized, and a four-control-input configuration is considered. A novel model predictive control algorithm is proposed based on the homogeneity of system model and an existing time-varying control law. Theoretical results on ensuring algorithm feasibility and closed-loop stability are provided. It is shown that if the terminal set level and the cost weights are designed appropriately, the developed algorithm is iteratively feasible and the closed-loop system is asymptotically stabilized at the equilibrium point. Comparison studies verify the effectiveness of the developed results.

Continuous-time model predictive control of under-actuated spacecraft with bounded control torques

The stabilization problem of rigid spacecraft is essential for space explorations and operations. This paper studies the model predictive stabilization problem of a class of underactuated rigid spacecrafts with two bounded control torques. A novel model predictive control (MPC) algorithm is designed by making use of the homogeneity of the system dynamics. In addition, a local homogeneous Lyapunov function is constructed based on which the approach to designing the terminal set and other parameters are developed. Finally, the conditions for ensuring algorithm feasibility and closed-loop stability are provided. We show that under the given conditions, the designed MPC algorithm is feasible, and the closed-loop system is asymptotically stable. Simulation and comparison studies verify that the developed results are effective and valid, and the designed controller fulfills the constraint satisfaction and achieves much faster convergence rate in comparison with conventional continuous time-varying controllers.

Exploring the modeling of spatiotemporal variations in ambient air pollution within the land use regression framework: Estimation of PM10 concentrations on a daily basis

Estimation of daily average exposure to PM10 (particulate matter with an aerodynamic diameter <10 μm) using the available fixed-site monitoring stations (FSMs) in a city poses a great challenge. This is because typically FSMs are limited in number when considering the spatial representativeness of their measurements and also because statistical models of citywide exposure have yet to be explored in this context. This paper deals with the later aspect of this challenge and extends the widely used land use regression (LUR) approach to deal with temporal changes in air pollution and the influence of transboundary air pollution on short-term variations in PM10. Using the concept of multiple linear regression (MLR) modeling, the average daily concentrations of PM10 in two European cities, Vienna and Dublin, were modeled. Models were initially developed using the standard MLR approach in Vienna using the most recently available data. Efforts were subsequently made to (i) assess the stability of model predictions over time; (ii) explores the applicability of nonparametric regression (NPR) and artificial neural networks (ANNs) to deal with the nonlinearity of input variables. The predictive performance of the MLR models of the both cities was demonstrated to be stable over time and to produce similar results. However, NPR and ANN were found to have more improvement in the predictive performance in both cities. Using ANN produced the highest result, with daily PM10 exposure predicted at R2 = 66% for Vienna and 51% for Dublin. In addition, two new predictor variables were also assessed for the Dublin model. The variables representing transboundary air pollution and peak traffic count were found to account for 6.5% and 12.7% of the variation in average daily PM10 concentration. The variable representing transboundary air pollution that was derived from air mass history (from back-trajectory analysis) and population density has demonstrated a positive impact on model performance.Implications: The implications of this research would suggest that it is possible to produce a model of ambient air quality on a citywide scale using the readily available data. Most European cities typically have a limited FSM network with average daily concentrations of air pollutants as well as available meteorological, traffic, and land-use data. This research highlights that using these data in combination with advanced statistical techniques such as NPR or ANNs will produce reasonably accurate predictions of ambient air quality across a city, including temporal variations. Therefore, this approach reduces the need for additional measurement data to supplement existing historical records and enables a lower-cost method of air pollution model development for practitioners and policy makers.

Model predictive stabilizer for T–S fuzzy recurrent multilayer neural network models with general terminal weighting matrix

This paper investigates the model predictive stabilization problem for Takagi–Sugeno (T–S) fuzzy multilayer neural networks with general terminal weighting matrix. A new set of linear matrix inequality (LMI) conditions on the general terminal weighting matrix of receding horizon cost function is presented such that T–S fuzzy multilayer neural networks with model predictive stabilizer are asymptotically stable. The general terminal weighting matrix of receding horizon cost function can be obtained by solving a set of LMIs. A numerical example is given to illustrate the effectiveness of the proposed stabilization scheme.

Influences of Spatial and Temporal Variation on Fish-Habitat Relationships Defined by Regression Quantiles

Abstract We used regression quantiles to model potentially limiting relationships between the standing crop of cutthroat trout Oncorhynchus clarki and measures of stream channel morphology. Regression quantile models indicated that variation in fish density was inversely related to the width:depth ratio of streams but not to stream width or depth alone. The spatial and temporal stability of model predictions were examined across years and streams, respectively. Variation in fish density with width:depth ratio (10th-90th regression quantiles) modeled for streams sampled in 1993-1997 predicted the variation observed in 1998-1999, indicating similar habitat relationships across years. Both linear and nonlinear models described the limiting relationships well, the latter performing slightly better. Although estimated relationships were transferable in time, results were strongly dependent on the influence of spatial variation in fish density among streams. Density changes with width:depth ratio in a single st...