Model-based Rules Research Articles

A Hybrid Approach: Dynamic Diagnostic Rules for Sensor Systems in Industry 4.0 Generated by Online Hyperparameter Tuned Random Forest

In this work, a hybrid component Fault Detection and Diagnosis (FDD) approach for industrial sensor systems is established and analyzed, to provide a hybrid schema that combines the advantages and eliminates the drawbacks of both model-based and data-driven methods of diagnosis. Moreover, it shines light on a new utilization of Random Forest (RF) together with model-based diagnosis, beyond its ordinary data-driven application. RF is trained and hyperparameter tuned using three-fold cross validation over a random grid of parameters using random search, to finally generate diagnostic graphs as the dynamic, data-driven part of this system. This is followed by translating those graphs into model-based rules in the form of if-else statements, SQL queries or semantic queries such as SPARQL, in order to feed the dynamic rules into a structured model essential for further diagnosis. The RF hyperparameters are consistently updated online using the newly generated sensor data to maintain the dynamicity and accuracy of the generated graphs and rules thereafter. The architecture of the proposed method is demonstrated in a comprehensive manner, and the dynamic rules extraction phase is applied using a case study on condition monitoring of a hydraulic test rig using time-series multivariate sensor readings.

A Hybrid Approach: Dynamic Diagnostic Rules for Sensor Systems in Industry 4.0 Generated by Online Hyperparameter Tuned Random Forest

In this work, a hybrid component Fault Detection and Diagnosis (FDD) approach for industrial sensor systems is established and analyzed, to provide a hybrid schema that combines the advantages and eliminates the drawbacks of both model-based and data-driven methods of diagnosis. Moreover, it shines the light on a new utilization of Random Forest (RF) together with model-based diagnosis, beyond its ordinary data-driven application. RF is trained and hyperparameter tuned using three-fold cross validation over a random grid of parameters using random search, to finally generate diagnostic graphs as the dynamic, data-driven part of this system. This is followed by translating those graphs into model-based rules in the form of if-else statements, SQL queries or semantic queries such as SPARQL, in order to feed the dynamic rules into a structured model essential for further diagnosis. The RF hyperparameters are consistently updated online using the newly generated sensor data to maintain the dynamicity and accuracy of the generated graphs and rules thereafter. The architecture of the proposed method is demonstrated in a comprehensive manner, and the dynamic rules extraction phase is applied using a case study on condition monitoring of a hydraulic test rig using time-series multivariate sensor readings.

A Hybrid Approach: Dynamic Diagnostic Rules for Sensor Systems in Industry 4.0 Generated by Online Hyperparameter Tuned Random Forest

In this work, a hybrid component Fault Detection and Diagnosis (FDD) approach for industrial sensor systems is established and analyzed, to provide a hybrid schema that combines the advantages and eliminates the drawbacks of both model-based and data-driven methods of diagnosis. Moreover, it shines the light on a new utilization of Random Forest (RF) together with model-based diagnosis, beyond its ordinary data-driven application. RF is trained and hyperparameter tuned using three-fold cross validation over a random grid of parameters using random search, to finally generate diagnostic graphs as the dynamic, data-driven part of this system. This is followed by translating those graphs into model-based rules in the form of if-else statements, SQL queries or semantic queries such as SPARQL, in order to feed the dynamic rules into a structured model essential for further diagnosis. The RF hyperparameters are consistently updated online using the newly generated sensor data to maintain the dynamicity and accuracy of the generated graphs and rules thereafter. The architecture of the proposed method is demonstrated in a comprehensive manner, and the dynamic rules extraction phase is applied using a case study on condition monitoring of a hydraulic test rig using time-series multivariate sensor readings.

Applying rule-based model-checking to construction site layout planning tasks

Abstract Building Information Modeling (BIM) is a widely established method in the architecture, engineering, construction, and facilities management (AEC/FM) industry. Although BIM focuses on processes throughout the lifecycle of the built environment, the applications in the planning phase, e.g. the generation of construction site layouts, have not reached their full potential yet. One important example herein is the allocation and dimensioning of resources (e.g., building materials and equipment) which is typically carried out by humans according to clearly defined rules and best practices. This paper presents model-based rule checking for the planning of construction site layouts. We demonstrate that existing Business Rule Management Systems (BRMS), such as the open-source rule engine Drools, can be used. We combine Drools with the Industry Foundation Classes (IFC) to retrieve data from a building information model and use the information within the rule engine. We define general model requirements and implement a sample set of prototype rules. We also introduce the concept of offset geometry for rules that, for example, demand a known safety distance between temporary construction site elements. The developed approaches are explained and evaluated in field-realistic, practical case studies. Finally, we present a discussion how the application of the developed rule-based system may assist human decision making in tasks such as safe construction sites layout planning.

A bigraphical multi-scale modeling methodology for system of systems

In this paper, we present a multi-scale modeling methodology for software System of Systems (SoS) using the formal technique of bigraphical reactive system. This methodology provides a correct by design approach ensuring the correctness of the SoS architectures. A first scale is defined by the designer. Then, it is refined by successively adding lower scale details. The transition between scales is implemented following a rule-oriented refinement process. The executed rules respect the system constraints ensuring, in this way, the correctness of the obtained scale architectures. Moreover, we address the dynamic aspect of SoS by providing model-based rules of reconfiguration actions. We illustrate our approach with a Smart Buildings case study.

Abstract 13636: Derivation and Validation of a Clinical Prediction Rule for Sleep Disordered Breathing in Patients With Cardiovascular Disease

Introduction: There is a growing demand for polysomnography (PSG) indispensable for the diagnosis of sleep apnea which is associated with the development of cardiovascular disease. Hypothesis: We assessed the hypothesis that a prediction rule without PSG could identify obstructive sleep apnea (OSA) patients in which the continuous positive airway pressure (CPAP) treatment was recommended. Methods: Consecutive 326 cardiovascular disease patients simultaneously underwent a full-night PSG and a digital-installed built-in memory pulse-oximeter. A logistic regression-based prediction rule was developed in patients at university hospital (n=278) and validated in patients at community hospital (n=48). Results: In the derivation group, apnea-hypopnea index (AHI) was correlated with 3% oxygen desaturation index (ODI) (r=0.94, P<0.0001) and 223 patients were diagnosed as sleep disordered breathing (SDB) (AHI >=5). Four independent parameters were identified to distinguish between OSA (n=161) and central sleep apnea patients (n=62): pulse pressure, left atrial diameter, minimum oxygen saturation and B-type natriuretic peptide levels. The observed probability of OSA, by calculated score, was comparable between derivation and validation groups (Figure 1). The model-based rule was well calibrated (Hosmer-Lemeshow χ2=4.23) and demonstrated good discrimination in the derivation group. Discriminatory ability was also reasonable in the validation group and the difference between these groups was limited (Figure 2). Provided that cutoff levels of 3%ODI and 4 independent parameters were entirely satisfied, positive predictive value was 100% in both groups for the diagnosis of OSA necessary for CPAP treatment. Conclusions: A prediction rule for the type of sleep apnea without PSG can quantify the difference of pathophysiological mechanism in SDB patients. This index may be useful in detecting eligible candidates for CPAP in cardiovascular disease patients.

The PID+p controller structure and its contextual model-based tuning

AbstractA recently published paper addressed the so called “PI+p” controller (i.e., a PI augmented with an additional, stable pole) and demonstrated that said controller yields interesting results particularly if tuned with the “contextual” approach, where model-based rules are used in conjunction with a relay experiment so as to provide at the same time the tuned regulator, and the tuning model. In this manuscript the scope is extended to the “PID+p” (with obvious meaning) controller structure, discussing its potentialities with reference to some benchmark cases.

Derivation and Validation of a Clinical Prediction Rule for Nosocomial Pneumonia after Coronary Artery Bypass Graft Surgery

Nosocomial pneumonia is an important cause of morbidity and mortality among surgical patients in the United States. The emergence of effective but potentially costly or risky preventive interventions makes perioperative risk stratification desirable. We sought to develop a prediction rule for pneumonia after coronary artery bypass grafting (CABG), a common surgical procedure. Data on individuals undergoing CABG at 32 hospitals in 6 states were extracted from Tenet Healthcare's Quality and Resource Management System. A logistic regression-based prediction rule was developed in half of the study sample and validated in the remaining patients. Of 17,143 individuals undergoing CABG from January 1999 through February 2004, 361 (2%) developed pneumonia without a known aspiration etiology. Thirteen independent predictors of pneumonia were identified in the derivation subset of the sample: body mass index <18.5 (defined as the weight in kilograms divided by the square of the height in meters), smoking history, admission from a nonresidential setting, cancer history, chronic obstructive pulmonary disease, Canadian Cardiovascular Society score 3, prior internal mammary artery CABG, emergency status, serum creatinine level >1.2 mg/dL, percutaneous transluminal coronary angioplasty, blood transfusion, preoperative vancomycin administration, and receipt of mechanical ventilation for >1 day. The model-based rule was well calibrated (Hosmer-Lemeshow X(2)=5.51; P=.70) and demonstrated good discrimination (area under the receiver-operating characteristic curve [ROC AUC], 0.78) in the derivation group. Discriminatory ability was also reasonable in the validation cohort (ROC AUC, 0.75; P=.18, for difference in ROC AUC between groups). Using a large cohort of patients treated at community and teaching hospitals, we derived and validated a prediction rule for pneumonia after CABG. This index may prove to be useful in prioritizing receipt of preventive interventions.

Auto-tune system using single-run relay feedback test and model-based controller design

In this paper, a systematic approach for auto-tune of PI/PID controller is proposed. A single run of the relay feedback experiment is carried out to characterize the dynamics including the type of damping behavior, the ultimate gain, and ultimate frequency. Then, according to the estimated damping behavior, the process is classified into two groups. For each group of processes, model-based rules for controller tuning are derived in terms of ultimate gains and ultimate frequencies. To classify the processes, the estimation of an apparent deadtime is required. Two artificial neural networks (ANNs) that characterize this apparent deadtime using the ATV data are thus included to facilitate this estimation of this apparent deadtime. The model-based design for this auto-tuning makes uses of parametric models of FOPDT (i.e. first-order-plus-dead-time) and of SOPDT (i.e. second-order-plus-dead-time) dynamics. The results from simulations show that the controllers thus tuned have satisfactory results compared with those from other methods.

Methods for knowledge acquisition and refinement in second generation expert systems

First generation expert systems rely on the use of surface knowledge, such as associational or heuristic. Second generation technology is characterized by two additional features: deep knowledge and machine learning. Three second generation methods for knowledge acquisition are reviewed: learning rules from examples, model-based rule learning, and semi-automatic model acquisition. The man-machine process of acquiring and refining knowledge extends the role of expert systems to expert support systems, since both man and machine learn through repeated knowledge refinement cycles. Explanation of solutions and of the knowledge base itself is crucial for this man-machine learning process. An extended expert system shell schema is presented that includes a knowledge acquisition and a knowledge explanation module.