Target Set Point Research Articles

Reinforcement learning for closed-loop regulation of cardiovascular system with vagus nerve stimulation: a computational study

Objective. Vagus nerve stimulation (VNS) is being investigated as a potential therapy for cardiovascular diseases including heart failure, cardiac arrhythmia, and hypertension. The lack of a systematic approach for controlling and tuning the VNS parameters poses a significant challenge. Closed-loop VNS strategies combined with artificial intelligence (AI) approaches offer a framework for systematically learning and adapting the optimal stimulation parameters. In this study, we presented an interactive AI framework using reinforcement learning (RL) for automated data-driven design of closed-loop VNS control systems in a computational study. Approach. Multiple simulation environments with a standard application programming interface were developed to facilitate the design and evaluation of the automated data-driven closed-loop VNS control systems. These environments simulate the hemodynamic response to multi-location VNS using biophysics-based computational models of healthy and hypertensive rat cardiovascular systems in resting and exercise states. We designed and implemented the RL-based closed-loop VNS control frameworks in the context of controlling the heart rate and the mean arterial pressure for a set point tracking task. Our experimental design included two approaches; a general policy using deep RL algorithms and a sample-efficient adaptive policy using probabilistic inference for learning and control. Main results. Our simulation results demonstrated the capabilities of the closed-loop RL-based approaches to learn optimal VNS control policies and to adapt to variations in the target set points and the underlying dynamics of the cardiovascular system. Our findings highlighted the trade-off between sample-efficiency and generalizability, providing insights for proper algorithm selection. Finally, we demonstrated that transfer learning improves the sample efficiency of deep RL algorithms allowing the development of more efficient and personalized closed-loop VNS systems. Significance. We demonstrated the capability of RL-based closed-loop VNS systems. Our approach provided a systematic adaptable framework for learning control strategies without requiring prior knowledge about the underlying dynamics.

Safety, Metabolic and Psychological Outcomes of Medtronic MiniMed 780G™ in Children, Adolescents and Young Adults: A Systematic Review.

The MiniMed™ 780G is a second-generation automated insulin delivery system that implements a modified proportional-integral-derivative algorithm with some features of an MD-Logic artificial pancreas algorithm. The system may deliver automatic correction boluses up to every 5min, and it allows the user to choose between three glucose target setpoints (100, 110 and 120mg/dL). We aimed to review the current evidence on this device in children, adolescents, and young adults living with type 1 diabetes. We screened 783 papers, but only 31 manuscripts were included in this review. Data on metabolic outcomes show that this system is safe as regards severe hypoglycaemia and diabetic ketoacidosis. The glycated haemoglobin may drop to levels about 7%, with CGM reports showing a time in range of 75-80%. The time above range and the time below range are within the recommended target in most of the subjects. Few studies evaluated the psychological outcomes. This system seems to be more effective than the first-generation automated insulin delivery systems. The MiniMed™ 780G has been associated with an improvement in sleep quality in subjects living with diabetes and their caregivers, along with an improvement in treatment satisfaction. Psychological distress is as reduced as the glucose control is improved. We also discuss some case reports describing particular situations in clinical practice. Finally, we think that data show that this system is a further step towards the improvement of the treatment of diabetes as concerns both metabolic and psychological outcomes.

A Proportional–Integral Feedback Controlled Automatic Flow Chemistry System to Produce On‐Demand AgAu Alloy Nanoboxes

Mixed‐metal nanoparticles present a challenging task for synthesis with high precision and reproducibility due to the complex interplay of compositional and reaction condition factors. The present study demonstrates a highly precise and automated flow chemistry technique for synthesizing AgAu alloy nanoboxes with tailored optical properties. The synthesis process involves a proportional–integral (PI) feedback control mechanism that enables accurate regulation of reaction parameters, such as the flow rate, to achieve the target AgAu nanoboxes having a desired UV–vis absorbance wavelength. The PI control algorithm is built on the first‐order plus dead‐time model, which correlates the flow rate of the precursors with the maximum absorbance peaks of the resultant nanoalloy products. Based on the difference between the real‐time measured UV–vis absorbance wavelength and the target wavelength, the flow rate of the precursor (i.e., reagent concentration) is tuned via an iterative process until the real‐time absorbance wavelength of the AgAu alloy nanoboxes is matched with the target setpoint. The implementation of a PI feedback control mechanism in a flow chemistry system can offer a highly versatile and universal strategy for generating on‐demand complex nanomaterials with significantly enhanced consistency and reliability by mitigating concentration variations and minimizing the need for human intervention.

An automated high inoculation density fed-batch bioreactor, enabled through N-1 perfusion, accommodates clonal diversity and doubles titers.

An important consideration for biopharmaceutical processes is the cost of goods (CoGs) of biotherapeutics manufacturing. CoGs can be reduced by dramatically increasing the productivity of the bioreactor process. In this study, we demonstrate that an intensified process which couples a perfused N-1 seed reactor and a fully automated high inoculation density (HID) N stage reactor substantially increases the bioreactor productivity as compared to a low inoculation density (LID) control fed-batch process. A panel of six CHOK1SV GS-KO® CHO cell lines expressing three different monoclonal antibodies was evaluated in this intensified process, achieving an average 85% titer increase and 132% space-time yield (STY) increase was demonstrated when comparing the 12-day HID process to a 15-day LID control process. These productivity increases were enabled by automated nutrient feeding in both the N-1 and N stage bioreactors using in-line process analytical technologies (PAT) and feedback control. The N-1 bioreactor utilized in-line capacitance to automatically feed the bioreactor based on a capacitance-specific perfusion rate (CapSPR). The N-stage bioreactor utilized in-line Raman spectroscopy to estimate real-time concentrations of glucose, phenylalanine, and methionine, which are held to target set points using automatic feed additions. These automated feeding methodologies were shown to be generalizable across six cell lines with diverse feed requirements. We show this new process can accommodate clonal diversity and reproducibly achieve substantial titer uplifts compared to traditional cell culture processes, thereby establishing a baseline technology platform upon which further increases bioreactor productivity and CoGs reduction can be achieved.

Raman-controlled pyruvate feeding to control metabolic activity and product quality in continuous biomanufacturing.

Despite technological advances ensuring stable cell culture perfusion operation over prolonged time, reaching a cellular steady-state metabolism remains a challenge for certain manufacturing cell lines. This study investigated the stabilization of a steady-state perfusion process producing a bispecific antibody with drifting product quality attributes, caused by shifting metabolic activity in the cell culture. A novel on-demand pyruvate feeding strategy was developed, leveraging lactate as an indicator for tricarboxylic acid (TCA) cycle saturation. Real-time lactate monitoring was achieved through in-line Raman spectroscopy, enabling accurate control at predefined target setpoints. The implemented feedback control strategy resulted in a three-fold reduction of ammonium accumulation and stabilized product quality profiles. Stable and flat glycosylation profiles were achieved with standard deviations below 0.2% for high mannose and fucosylation. Whereas galactosylation and sialylation were stabilized in a similar manner, varying lactate setpoints might allow for fine-tuning of these glycan forms. The Raman-controlled pyruvate feeding strategy represents a valuable tool for continuous manufacturing, stabilizing metabolic activity, and preventing product quality drifting in perfusion cell cultures. Additionally, this approach effectively reduced high mannose, helping to mitigate increases associated with process intensification, such as extended culture durations or elevated culture densities.

Adaptive Gain Tuning Rule for Nonlinear Sliding-Mode Speed Control of Encoderless Three-Phase Permanent Magnet Assisted Synchronous Motor

In this paper, an adaptive gain tuning rule is designed for the nonlinear sliding mode speed control (NSMSC) in order to enhance the dynamic performance and the robustness of the permanent magnet assisted synchronous reluctance motor (PMa-SynRM) with considering the parameter uncertainties. A nonlinear sliding surface whose parameters are altering with time is designed at first. The proposed NSMSC can minimize the settling time without any overshoot via utilizing a low damping ratio at starting along with a high damping ratio as the output approaches the target set-point. In addition, it eliminates the problem of the singularity with the upper bound of an uncertain term that is hard to be measured practically as well as ensures a rapid convergence in finite time, through employing a simple adaptation law. Moreover, for enhancing the system efficiency throughout the constant torque region, the control system utilizes the maximum torque per ampere technique. The nonlinear sliding surface stability is assured via employing Lyapunov stability theory. Furthermore, a simple sliding mode estimator is employed for estimating the system uncertainties. The stability analysis and the experimental results indicate the effectiveness along with feasibility of the proposed speed estimation and the NSMSC approach for a 1.1-kW PMa-SynRM under different speed references, electrical and mechanical parameters disparities, and load disturbance conditions.

A Computationally Efficient Algorithm for Perturbed Dynamic Programs (A-PDP)

This letter proposes an algorithm for solving perturbed dynamic programs, where first-order corrections are applied to a pre-computed optimal strategy and its corresponding value function using local quadratic approximation. The technique is developed to handle perturbations of external inputs and parameters affecting system dynamics, objective and constraint functions, allowing the application to a wide variety of perturbed problems. The method is applied to the energy optimization of an electric vehicle thermal management system, formulated as a constrained dynamic program where external conditions and target set-point are perturbed from their nominal values. Simulations result show 80% reduction in computation time from a Dynamic Programming (DP) solution, with only minimal effect on the overall controller performance.

Design and Validation of Automated Sensor-Based Artificial Ripening System Combined with Ultrasound Pretreatment for Date Fruits

Climate change affects fruit crops’ growth and development by delaying fruit ripening, reducing color development, and lowering fruit quality and yield. The irregular date palm fruit ripening in the past few years is assumed to be related to climatic change. The current study aimed to design and validate an automated sensor-based artificial ripening system (S-BARS) combined with ultrasound pretreatment for artificial ripening date fruits cv. Khalas. A sensor-based control system was constructed to allow continuous real-time recording and control over the process variables. The impact of processing variables, i.e., the artificial ripening temperature (ART-temp) and relative humidity (ART-RH) using the designed S-BARS combined with ultrasound pretreatment variables, i.e., time (USP-Time) and temperature (USP-Temp) on the required time for fruit ripening (RT), the percentage of ripened fruits (PORF), the percentage of damaged fruits (PODF), and the electrical energy consumption (EEC) were investigated. The quadratic predictive models were developed using the Box–Behnken Design (B-BD) to predict the RT, PORF, PODF, and EEC experimentally via Response Surface Methodology(RSM). Design Expert software (Version 13) was used for modeling and graphically analyzing the acquired data. The artificial ripening parameter values were determined by solving the regression equations and analyzing the 3D response surface plots. All parameters were simultaneously optimized by RSM using the desirability function. The Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE) between the predicted and actual experimental values were used to evaluate the developed models. The physicochemical properties of the ripened fruit were assessed under the optimization criteria. The results indicated that the pretreated unripe date fruits with 40 kHz ultrasound frequency, 110 W power, and USP-Temp of 32.49 °C for 32.03 min USP-Time under 60 °C ART-Temp and 59.98% ART-RH achieved the best results. The designed S-BARS precisely controlled the temperature and relative humidity at the target setpoints. The ultrasound pretreatment improved the color and density of the artificially ripened date fruits, decreased the RT and EEC, and increased the PORF without negatively affecting the studied fruit quality attributes. The developed models could effectively predict the RT, PORF, PODF, and EEC. The designed S-BARS combined with ultrasound pretreatment is an efficient approach for high-quality ripening date fruits.

Linear Tracking MPC for Nonlinear Systems—Part I: The Model-Based Case

We develop a tracking model predictive control (MPC) scheme for nonlinear systems using the linearized dynamics at the current state as a prediction model. Under reasonable assumptions on the linearized dynamics, we prove that the proposed MPC scheme exponentially stabilizes the optimal reachable equilibrium w.r.t. a desired target setpoint. Our theoretical results rely on the fact that, close to the steady-state manifold, the prediction error of the linearization is small and hence, we can slide along the steady-state manifold towards the optimal reachable equilibrium. The closed-loop stability properties mainly depend on a cost matrix which allows us to trade off performance, robustness, and the size of the region of attraction. In an application to a nonlinear continuous stirred tank reactor, we show that the scheme, which only requires solving a convex quadratic program online, has comparable performance to a nonlinear MPC scheme while being computationally significantly more efficient. Further, our results provide the basis for controlling nonlinear systems based on data-dependent linear prediction models, which we explore in our companion paper.

Water chemistry challenges in pulping and papermaking – fundamentals and practical insights: Part 1: Water chemistry fundamentals and pH

Water is an essential component of the papermaking process. Nevertheless, papermakers often overlook its importance compared to fibers and chemical additives. A better understanding of water properties and chemical interactions associated with water at the wet end leads to a sound foundation for high-quality paper production and smooth operation. Not all fresh water and process water is the same. Fresh water varies from mill to mill, primarily due to the location and availability of water sources. Some industrial trends, such as enhancing water conservation and production yield, gradually shift process water quality over time. The current work serves as a primer on water and water chemistry fundamentals to help the papermaker prepare for the future challenges of increased contamination of process water associated with reduced fresh water usage. This paper focuses on basic water chemistry definitions and discusses the impact of pH on wet-end operation. It is clear that pH is a fundamental factor that directly affects the process and impacts other factors relevant to the papermaking process. It is crucial to understand what pH represents, how it is measured, how to select the proper pH and carefully control it, and how to closely maintain the process at target setpoints. Understanding the sensitivity of operation to pH change will lead to an appropriate focus on these issues. In addition to basic theory, we also review onsite experience and practical mill cases. It is imperative to stress that, although critical, pH is not the only chemical parameter impacting papermaking operations. Other factors, such as ionic concentration measured by conductivity, surface, soluble charge, and hardness, are critical and will be discussed in Part II of this series. As pH is a primary and independent factor that impacts various forms of charge and conductivity, the authors decided to start the current series of papers by discussing pH.

Clinical Evaluation of an Automatic Oxygen Control System for Premature Infants Receiving High-Flow Nasal Cannula for Respiratory Support: A Pilot Study.

A pilot clinical study was conducted that compared the peripheral oxygen saturation (SpO2) targeting performance of an automatic oxygen control system with manual oxygen control, which is the standard of care for preterm and low birth weight infants on high-flow nasal cannula (HFNC). The new oxygen control device studied was used to automatically adjust the fraction of inspired oxygen (FiO2) according to a desired SpO2 target setpoint and measured feedback signals including the SpO2 and other signals. A crossover study was designed with several endpoints including the comparison of the percentage of time that the SpO2 was within the target range with the automatic oxygen control device versus manual oxygen control. Other metrics were also compared to assess the performance of the system including the number of bradycardia events. The pilot study included six patients that fit the inclusion criteria. The results showed that there were improvements in all of the measured outcomes considered including statistically significant improvements in the number of bradycardia events during the period when the automatic oxygen control device was used.

Adaptive charging control using ANN-PID controllers on multiple DC loads with varying battery voltages

Various rechargeable electronic devices currently have batteries with different capacities and voltages, while the available chargers are generally fixed for one device. This is considered less effective because different types of electronic devices will require different battery chargers. Therefore, the adaptive power charge is needed to recharge batteries with different voltages and capacities through a single port by adjusting the type of load connected. This system uses buck converter with duty cycle settings through microcontrollers to lower the input voltage to variable output voltage. When the load is connected, the limit switch will be depressed and the system will start the duty cycle tracking process. The voltage will be increased gradually until the current is read at a certain value to identify the load. After the current reads the duty cycle stops tracking, then the current and voltage characteristics are used as input variables for the artificial neural network (ANN) algorithm to determine the target setpoint voltage to be executed by the proportional, integral and derivative (PID) controller. The designed adaptive power charge can identify the connected load accurately. The average ANN output error is 1.46e-4% and the average PID controller error is 6.4e-2%. The system can reach a steady state at 0.01 s.

ATTD 2021 Invited Speaker Abstracts.

ATTD 2021 Invited Speaker Abstracts.

Автоматический уровнемер жидкости для взрывоопасных зон

The purpose of this work is to develop an automatic liquid level meter for hazardous zones. The analysis of currently known level meters, which are widely used in the oil and oil refining industry, showed that their cost is very high. The proposed level gauge cost is significantly lower than that of industrial analogues for manufacture and operation. This paper describes the capabilities and operating principles of the developed liquid level meter for hazardous zones. The proposed level gauge is designed to measure the level of petroleum products for technological accounting and control of petroleum products. The flowchart and algorithm of the level gauge operation are given. This paper presents the "upper" level of automatic process control on personal computers. This level is based on the Trace Mode software package by AdAstrA Research Group, Ltd (Russia). The software package provides for the transmission of target setpoints from a computer, preliminary alarm, archiving of data of technological parameters, and a mnemonic diagram for passing technological parameters in real time. The level meter is designed to measure various levels with appropriate settings, the measurement accuracy ranges from 0.5–1 %.

Continuous Mixing Technology: Characterization of a Vertical Mixer Using Residence Time Distribution.

Continuous powder mixing technology (CMT) application during continuous direct compression has emerged as a leading technology used in the development and manufacture of solid oral dosage forms. The critical quality attributes of the final product are heavily dependent on the performance of the mixing step as the quality of mixing directly influences the drug product quality attributes. This study investigates the impact of blend material properties (bulk density, API particle size distribution) and process parameters (process throughput, hold up mass and impeller speed) on the mixing performance. Mixing of the blend was characterized using the Residence Time Distribution (RTD) of the process by trending the outlet stream of the mixer using a near-infrared (NIR) probe after the injection of a small mass of tracer at the inlet stream. The outcomes of this study show that the RTDs of the mixer with throughput ranging between 15 and 30kg/h; impeller speed ranging between 400 and 600rpm and hold up mass (HUM) ranging between 500 and 850g can be described by a series of two ideal Continuous Stirred Tank Reactors (CSTRs) with different volumes, and correspondingly, different mean residence times. It is also observed that the mixing is mainly occurring in the lower chamber of the CMT and the normalized RTDs of the mixer are similar across the range of process conditions and material attributes studied. The results also showed that the formulation blend with different API particle sizes and bulk properties, like bulk density and flowability, provide insignificant impact on the mixing performance. The CMT allows independent selection of target set points for HUM, impeller rotational speed and line throughput and it shows great robustness and flexibility for continuous blending in solid oral dose manufacturing.

Improved Glycemic Outcomes With Medtronic MiniMed Advanced Hybrid Closed-Loop Delivery: Results From a Randomized Crossover Trial Comparing Automated Insulin Delivery With Predictive Low Glucose Suspend in People With Type 1 Diabetes.

To study the MiniMed Advanced Hybrid Closed-Loop (AHCL) system, which includes an algorithm with individualized basal target set points, automated correction bolus function, and improved Auto Mode stability. This dual-center, randomized, open-label, two-sequence crossover study in automated-insulin-delivery-naive participants with type 1 diabetes (aged 7-80 years) compared AHCL to sensor-augmented pump therapy with predictive low glucose management (SAP + PLGM). Each study phase was 4 weeks, preceded by a 2- to 4-week run-in and separated by a 2-week washout. The study was completed by 59 of 60 people (mean age 23.3 ± 14.4 years). Time in target range (TIR) 3.9-10 mmol/L (70-180 mg/dL) favored AHCL over SAP + PLGM (70.4 ± 8.1% vs. 57.9 ± 11.7%) by 12.5 ± 8.5% (P < 0.001), with greater improvement overnight (18.8 ± 12.9%, P < 0.001). All age-groups (children [7-13 years], adolescents [14-21 years], and adults [>22 years]) demonstrated improvement, with adolescents showing the largest improvement (14.4 ± 8.4%). Mean sensor glucose (SG) at run-in was 9.3 ± 0.9 mmol/L (167 ± 16.2 mg/dL) and improved with AHCL (8.5 ± 0.7 mmol/L [153 ± 12.6 mg/dL], P < 0.001), but deteriorated during PLGM (9.5 ± 1.1 mmol/L [17 ± 19.8 mg/dL], P < 0.001). TIR was optimal when the algorithm set point was 5.6 mmol/L (100 mg/dL) compared with 6.7 mmol/L (120 mg/dL), 72.0 ± 7.9% vs. 64.6 ± 6.9%, respectively, with no additional hypoglycemia. Auto Mode was active 96.4 ± 4.0% of the time. The percentage of hypoglycemia at baseline (<3.9 mmol/L [70 mg/dL] and ≤3.0 mmol/L [54 mg/dL]) was 3.1 ± 2.1% and 0.5 ± 0.6%, respectively. During AHCL, the percentage time at <3.9 mmol/L (70 mg/dL) improved to 2.1 ± 1.4% (P = 0.034) and was statistically but not clinically reduced for ≤3.0 mmol/L (54 mg/dL) (0.5 ± 0.5%; P = 0.025). There was one episode of mild diabetic ketoacidosis attributed to an infusion set failure in combination with an intercurrent illness, which occurred during the SAP + PLGM arm. AHCL with automated correction bolus demonstrated significant improvement in glucose control compared with SAP + PLGM. A lower algorithm SG set point during AHCL resulted in greater TIR, with no increase in hypoglycemia.

An optimal hierarchical control scheme for smart generation units: An application to combined steam and electricity generation

Optimal management of thermal and energy grids with fluctuating demand and prices requires to orchestrate the generation units (GU) among all their operating modes. A hierarchical approach is proposed to control coupled energy nonlinear systems. The high level hybrid optimization defines the unit commitment, with the optimal transition strategy, and best production profiles. The low level dynamic model predictive control (MPC), receiving the set-points from the upper layer, safely governs the systems considering process constraints. To enhance the overall efficiency of the system, a method to optimal start-up the GU is here presented: a linear parameter-varying MPC computes the optimal trajectory in closed-loop by iteratively linearizing the system along the previous optimal solution. The introduction of an intermediate equilibrium state as additional decision variable permits the reduction of the optimization horizon, while a terminal cost term steers the system to the target set-point. Simulation results show the effectiveness of the proposed approach.

Transition from double-leg to single-leg stance in the assessment of postural stability

Kinematic characteristics of the double-leg stance (DLS) to a single-leg stance (SLS) transition were analyzed in a group of young adolescent girls to assess their postural stability control. Twenty volunteers participated in a single experimental session during which their postural stability was assessed based upon the center of pressure (COP) trajectories during the transitions in two typical sensory conditions: with eyes open (EO) and with eyes closed (EC). To quantify the postural control we applied Fitts' model treating the postural sway as the noise at the initial and the target setpoint control. Results showed that in young healthy subjects characteristics of the transition to either left or right single-leg stance were quite symmetrical. The postural sway at the target posture was characterized by the double increase of postural sway when tested with EO and by the almost quadrupled amount of sway in EC trials. The sway at the target resulted in the decline of the COP mean and peak velocity proportionally to the movement index of difficulty (ID). The estimated ID value increased by 74% in EC trials while the probability of instability increased to 70%. The DLS-SLS test can be recommended for clinical and laboratory assessment of postural stability.

97-LB: Safety and Glycemic Outcomes of the MiniMed Advanced Hybrid Closed-Loop (AHCL) System in Subjects with T1D

Background: Automated insulin delivery systems improve glycemic control including time spent in, below and above range (TIR, TBR and TAR, respectively) in people with T1D. A next generation system, the MiniMedTM AHCL system, that offers a target set point (SP) of 100 or 120mg/dL, autocorrects to 120mg/dL every 5 mins and has fewer Auto Mode exits, was evaluated. Methods: A 16-site, single-arm, in-home trial of the AHCL system in 39 adolescents (14-21yrs) and 118 adults (≥22yrs) with T1D was conducted. After a baseline period of sensor-integrated pump, HCL feature, or predictive low glucose feature use (~14 days), the AHCL feature was enabled with a 100 or 120mg/dL SP for ~45 days and then the other SP for ~45 days. Study endpoints included safety events, change in A1C, sensor glucose (SG), %TIR, %TBR and %TAR. Results: There were no DKA or severe hypoglycemia episodes during the study period. Auto Mode was used ≥95% of the time; autocorrection was 22% of daily bolus. The table shows outcomes (mean±SD) for the overall group and each age group at baseline and study period (100 and 120mg/dL SP), and at the 100mg/dL SP. Best glycemic results (SG of 141±8.8mg/dL, TIR of 78.8±5.5% and TBR of 2.6±2.0%; N=29) were seen with a 100mg/dL SP and active insulin time of 120 minutes. Conclusion: These pivotal trial data demonstrate that AHCL is safe and significantly improved A1C and SG in subjects aged ≥14 years with T1D. Disclosure A.L. Carlson: Consultant; Self; Medtronic. Research Support; Self; Abbott, Dexcom, Inc., Eli Lilly and Company, Insulet Corporation, Medtronic, Novo Nordisk A/S, Sanofi, UnitedHealth Group. B.W. Bode: Advisory Panel; Self; Medtronic, Novo Nordisk A/S. Consultant; Self; Eli Lilly and Company, Medtronic, Novo Nordisk A/S. Research Support; Self; Abbott, Advance, Boehringer Ingelheim Pharmaceuticals, Inc., Dexcom, Inc., Diasome Pharmaceuticals, Inc., Eli Lilly and Company, Insulet Corporation, Janssen Pharmaceuticals, Inc., MannKind Corporation, Medtronic, National Institutes of Health, Nova Biomedical, Novo Nordisk A/S, Provention Bio, Inc., Sanofi, Senseonics. Speaker’s Bureau; Self; AstraZeneca, Boehringer Ingelheim Pharmaceuticals, Inc., Eli Lilly and Company, Janssen Pharmaceuticals, Inc., MannKind Corporation, Medtronic, Novo Nordisk A/S, Sanofi, Senseonics. Stock/Shareholder; Self; Aseko. R.L. Brazg: None. M.P. Christiansen: Research Support; Self; Abbott, Ascensia Diabetes Care, Biolinq, Dexcom, Inc., Eli Lilly and Company, Lexicon Pharmaceuticals, Inc., Medtronic, Merck &amp; Co., Inc., Novo Nordisk A/S, Sanofi-Aventis, Senseonics, ViroMed Laboratories, Xeris Pharmaceuticals, Inc. S.K. Garg: Advisory Panel; Self; Boehringer Ingelheim Pharmaceuticals, Inc., Eli Lilly and Company, Medtronic, Novo Nordisk A/S, Roche Diagnostics France. Research Support; Self; Dexcom, Inc., Eli Lilly and Company, Medtronic. K. Kaiserman: Advisory Panel; Self; Medtronic. Consultant; Self; Medtronic. Research Support; Self; Medtronic. Speaker’s Bureau; Self; Medtronic. M. Kipnes: None. D.R. Liljenquist: None. A. Philis-Tsimikas: Advisory Panel; Self; Lilly Diabetes, Medtronic, Novo Nordisk A/S, Sanofi. Employee; Spouse/Partner; Ionis Pharmaceuticals, Inc. Research Support; Self; Boehringer Ingelheim Pharmaceuticals, Inc., Dexcom, Inc., Lilly Diabetes, Medtronic, Novo Nordisk A/S, Sanofi. R. Pop-Busui: Advisory Panel; Self; Boehringer Ingelheim Pharmaceuticals, Inc. Consultant; Self; Bayer Healthcare Pharmaceuticals Inc., Novo Nordisk Inc. Research Support; Self; AstraZeneca. Other Relationship; Self; American Diabetes Association. J. Sherr: Advisory Panel; Self; Bigfoot Biomedical, Cecelia Health. Consultant; Self; Eli Lilly and Company, Lexicon Pharmaceuticals, Inc., Medtronic, Sanofi, T1D Exchange. D.I. Shulman: Advisory Panel; Self; Medtronic. K.K. Singh: None. R.H. Slover: None. J. Thrasher: Advisory Panel; Self; Lexicon Pharmaceuticals, Inc., Novo Nordisk Inc., Sanofi. Board Member; Self; Medtronic. Consultant; Self; Medtronic. Research Support; Self; Boehringer Ingelheim Pharmaceuticals, Inc., Bristol-Myers Squibb, Eli Lilly and Company, Lexicon Pharmaceuticals, Inc., Medtronic, Mylan, Novo Nordisk Inc., Sanofi. Speaker’s Bureau; Self; AstraZeneca, Janssen Pharmaceuticals, Inc., Medtronic, Sanofi. X. Chen: Employee; Self; Medtronic. B. Grosman: Employee; Self; Medtronic. S.W. Lee: Employee; Self; Medtronic. L.J. Lintereur: Employee; Self; Medtronic. M. Liu: Employee; Self; Medtronic. N. Parikh: None. A.S. Rhinehart: Employee; Self; Glytec, Medtronic. Stock/Shareholder; Self; Dexcom, Inc., Glytec, Lexicon Pharmaceuticals, Inc., MannKind Corporation, Tandem Diabetes Care. F. Peng: None. A. Roy: Employee; Self; Medtronic. J. Shin: Employee; Self; Medtronic. D. Wu: None. R. Vigersky: Employee; Self; Medtronic.

A physics-informed Run-to-Run control framework for semiconductor manufacturing

For decades, Run-to-Run (R2R) controllers have been widely implemented in semiconductor manufacturing. They operate over key process parameters on the basis of the metrological measurements acquired from the process and their deviations from the target setpoints. Conventionally, R2R controllers have been implemented independently of the actual equipment condition, which is obviously affecting the process stability and performance. Therefore, both equipment signals and process states shall be considered to make the R2R controllers more robust to the equipment condition drifts. In this paper, we propose a novel physics-informed framework to integrate the real-time equipment condition, based on the Fault Detection and Classification (FDC) data, into the R2R controllers. By utilizing Dynamic Bayesian Networks (DBN), the implicit relationship structure between metrology measurements, FDC indicators, and R2R regulators can be learned and reviewed explicitly. The structure shall be further reviewed to valid with the existing relationships and expert knowledge. Infeasible causalities on the structure will be constrained via setting up the blacklist at the structure learning stage. The proposed framework consists of the offline modeling stage, which incorporates the process, equipment variables, and the expert knowledge in the structure learning, and the online control stage, which constructs the Structured R2R controller (SRC) based on the relationship structure. As a result, the model is consistent by design with empirically known relationships and fundamental physical laws. The proposed SRC not only optimizes the operation with respect to the target control values but also considers the equipment and process states simultaneously. The effectiveness of SRC and the derivative control strategy are validated through a real dataset of a Chemical-Mechanical Polishing (CMP) process, and two simulated studies.