Sumit Mukherjee

Building temperature control: A passivity-based approach

This paper focuses on the temperature control in a multi-zone building. The lumped heat transfer model based on thermal resistance and capacitance is used to analyze the system dynamics and control strategy. The resulting thermal network, including the zones, walls, and ambient environment, may be represented as an undirected graph. The thermal capacitances are the nodes in the graph, connected by thermal resistances as links. We assume the temperature measurements and temperature control elements (heating and cooling) are collocated. We show that the resulting input/output system is strictly passive, and any passive output feedback controller may be used to improve the transient and steady state performance without affecting the closed loop stability. The storage functions associated with passive systems may be used to construct a Lyapunov function, to demonstrate closed loop stability and motivates the construction of an adaptive feedforward control. A four-room example is included to illustrate the performance of the proposed passivity based control strategy.

Single-cell profiling of the developing mouse brain and spinal cord with split-pool barcoding

Identifying single-cell types in the mouse brain The recent development of single-cell genomic techniques allows us to profile gene expression at the single-cell level easily, although many of these methods have limited throughput. Rosenberg et al. describe a strategy called split-pool ligation-based transcriptome sequencing, or SPLiT-seq, which uses combinatorial barcoding to profile single-cell transcriptomes without requiring the physical isolation of each cell. The authors used their method to profile >100,000 single-cell transcriptomes from mouse brains and spinal cords at 2 and 11 days after birth. Comparisons with in situ hybridization data on RNA expression from Allen Institute atlases linked these transcriptomes with spatial mapping, from which developmental lineages could be identified. Science, this issue p. 176 Single-cell analyses with SPLiT-seq (split-pool ligation-based transcriptome sequencing) elucidate development of the mouse nervous system. To facilitate scalable profiling of single cells, we developed split-pool ligation-based transcriptome sequencing (SPLiT-seq), a single-cell RNA-seq (scRNA-seq) method that labels the cellular origin of RNA through combinatorial barcoding. SPLiT-seq is compatible with fixed cells or nuclei, allows efficient sample multiplexing, and requires no customized equipment. We used SPLiT-seq to analyze 156,049 single-nucleus transcriptomes from postnatal day 2 and 11 mouse brains and spinal cords. More than 100 cell types were identified, with gene expression patterns corresponding to cellular function, regional specificity, and stage of differentiation. Pseudotime analysis revealed transcriptional programs driving four developmental lineages, providing a snapshot of early postnatal development in the murine central nervous system. SPLiT-seq provides a path toward comprehensive single-cell transcriptomic analysis of other similarly complex multicellular systems.

Continuous order identification of PHWR models under step-back for the design of hyper-damped power tracking controller with enhanced reactor safety

In this paper, discrete time higher integer order linear transfer function models have been identified first for a 500 MWe Pressurized Heavy Water Reactor (PHWR) which has highly nonlinear dynamical nature. Linear discrete time models of the nonlinear nuclear reactor have been identified around eight different operating points (power reduction or step-back conditions) with least square estimator (LSE) and its four variants. From the synthetic frequency domain data of these identified discrete time models, fractional order (FO) models with sampled continuous order distribution are identified for the nuclear reactor. This enables design of continuous order Proportional–Integral–Derivative (PID) like compensators in the complex w-plane for global power tracking at a wide range of operating conditions. Modeling of the PHWR is attempted with various levels of discrete commensurate-orders and the achievable accuracies are also elucidated along with the hidden issues, regarding modeling and controller design. Credible simulation studies are presented to show the effectiveness of the proposed reactor modeling and power level controller design. The controller pushes the reactor poles in higher Riemann sheets and thus makes the closed loop system hyper-damped which ensures safer reactor operation at varying dc-gain while making the power tracking temporal response slightly sluggish; but ensuring greater safety margin.

Scaling single cell transcriptomics through split pool barcoding

Constructing an atlas of cell types in complex organisms will require a collective effort to characterize billions of individual cells. Single cell RNA sequencing (scRNA-seq) has emerged as the main tool for characterizing cellular diversity, but current methods use custom microfluidics or microwells to compartmentalize single cells, limiting scalability and widespread adoption. Here we present Split Pool Ligation-based Transcriptome sequencing (SPLiT-seq), a scRNA-seq method that labels the cellular origin of RNA through combinatorial indexing. SPLiT-seq is compatible with fixed cells, scales exponentially, uses only basic laboratory equipment, and costs one cent per cell. We used this approach to analyze 109,069 single cell transcriptomes from an entire postnatal day 5 mouse brain, providing the first global snapshot at this stage of development. We identified 13 main populations comprising different types of neurons, glia, immune cells, endothelia, as well as types in the blood-brain-barrier. Moreover, we resolve substructure within these clusters corresponding to cells at different stages of development. As sequencing capacity increases, SPLiT-seq will enable profiling of billions of cells in a single experiment.

Building temperature control with adaptive feedforward

A common approach to the modeling of temperature evolution in a multi-zone building is to use thermal resistance and capacitance to model zone and wall dynamics. The resulting thermal network may be represented as an undirected graph. The thermal capacitances are the nodes in the graph, connected by thermal resistances as links. The temperature measurements and temperature control elements (heating and cooling) in this lumped model are collocated. As a result, the input/output system is strictly passive and any passive output feedback controller may be used to improve the transient and steady state performance without affecting the closed loop stability. The storage functions associated with passive systems may be used to construct a Lyapunov function, to demonstrate closed loop stability and motivate the construction of an adaptive feedforward control to compensate for the variation of the ambient temperature and zone heat loads (due to changing occupancy). The approach lends itself naturally to an inner-outer loop control architecture where the inner loop is designed for stability, while the outer loop balances between temperature specification and power consumption. Energy efficiency consideration may be added by adjusting the target zone temperature based on user preference and energy usage. The initial analysis uses zone heating/cooling as input, but the approach may be extended to more general model where the zonal mass flow rate is the control variable. A four-room example with realistic ambient temperature variation is included to illustrate the performance of the proposed passivity based control strategy.

Identification of the core temperature in a fractional order noisy environment for thermal feedback in nuclear reactors

In this paper, the core temperature of a 540 MWe Pressurized Heavy Water Reactor (PHWR) has been estimated from the measured temperature of the coolant using system identification techniques. The formulation is done considering fractional order (FO) dynamics for the sensor noise (i.e. 1/ƒα noise) and the conventional white noise in measurement. Performance study with different least square based linear estimators and nonlinear AutoRegressive eXogenous (ARX) and nonlinear Hammerstein-Wiener class of estimators for system identification has been done to show their relative merits to handle fractional order noise dynamics for estimating reactor core temperature.

Identification of nonlinear systems from the knowledge around different operating conditions: A feed-forward multi-layer ANN based approach

The paper investigates nonlinear system identification using system output data at various linearized operating points. A feed-forward multi-layer Artificial Neural Network (ANN) based approach is used for this purpose and tested for two target applications i.e. nuclear reactor power level monitoring and an AC servo position control system. Various configurations of ANN using different activation functions, number of hidden layers and neurons in each layer are trained and tested to find out the best configuration. The training is carried out multiple times to check for consistency and the mean and standard deviation of the root mean square errors (RMSE) are reported for each configuration.

Evaluation of energy saving potential using Stochastic Model Predictive Control for stand alone Air Conditioning units a study in Indian scenario

Reduction in energy consumption has become imperative in the modern day. The building segment is responsible for almost 40% consumption. These installations are typically located at the distribution level. There are losses associated with the transmission system, such as T&D Losses and pilferage losses. Hence a single unit of energy saved at distribution level would amount to greater savings at the generation level. Developing nations rely largely on standalone Air Conditioning units for office and domestic use since these facilities rarely designed to accommodate centralized Heating ventilation and Air Conditioning (HVAC) systems. In this paper a novel approach to control all these air conditioning units using a centralized controller based on Stochastic Model Predictive Control (SMPC) has been presented. The SMPC takes into account the predicted weather to reduce energy consumption while maintaining the comfort level of the occupants. A sample office space has been modeled and performance of the algorithm has been studied for weather conditions of large cities of India. With centralized SMPC the system has significantly outperformed the existing SAC with localized controller.

Comparative studies on decentralized multiloop PID controller design using evolutionary algorithms

Decentralized PID controllers have been designed in this paper for simultaneous tracking of individual process variables in multivariable systems under step reference input. The controller design framework takes into account the minimization of a weighted sum of Integral of Time multiplied Squared Error (ITSE) and Integral of Squared Controller Output (ISCO) so as to balance the overall tracking errors for the process variables and required variation in the corresponding manipulated variables. Decentralized PID gains are tuned using three popular Evolutionary Algorithms (EAs) viz. Genetic Algorithm (GA), Evolutionary Strategy (ES) and Cultural Algorithm (CA). Credible simulation comparisons have been reported for four benchmark 2×2 multivariable processes.

Scalable preprocessing for sparse scRNA-seq data exploiting prior knowledge

Motivation Single cell RNA‐seq (scRNA‐seq) data contains a wealth of information which has to be inferred computationally from the observed sequencing reads. As the ability to sequence more cells improves rapidly, existing computational tools suffer from three problems. (i) The decreased reads‐per‐cell implies a highly sparse sample of the true cellular transcriptome. (ii) Many tools simply cannot handle the size of the resulting datasets. (iii) Prior biological knowledge such as bulk RNA‐seq information of certain cell types or qualitative marker information is not taken into account. Here we present UNCURL, a preprocessing framework based on non‐negative matrix factorization for scRNA‐seq data, that is able to handle varying sampling distributions, scales to very large cell numbers and can incorporate prior knowledge. Results We find that preprocessing using UNCURL consistently improves performance of commonly used scRNA‐seq tools for clustering, visualization and lineage estimation, both in the absence and presence of prior knowledge. Finally we demonstrate that UNCURL is extremely scalable and parallelizable, and runs faster than other methods on a scRNA‐seq dataset containing 1.3 million cells. Availability and implementation Source code is available at https://github.com/yjzhang/uncurl_python.