Search

Density of microalgae is critical information for production of algae in a closed cultivation system since it can be used to optimally control their growth rate, biomass concentration and quality of the products. Given advancement in image processing techniques and thanks to low-cost camera sensors, image based methods are increasingly widely utilized to indirectly estimate the density. Advantages of the image based techniques include being less invasive and more nondestructive and biosecured. However, most of the existing techniques rely on averaging all pixels of a microalgae image, which may eliminate crucial information of their spatial correlation. Therefore, in this work we propose to exploit a convolutional neural network (CNN) to efficiently extract information from the microalgae images, which are then employed to regress the density. Interestingly, the proposed deep CNN regression model accepts the whole color image as its input while the density is calculated in the output. The proposed CNN regression architecture was validated in real-world experiments where the microalgal strain Chlorella vulgaris was cultured and their images were captured by our low-cost camera sensor system. The obtained results demonstrate that the averaged estimation accuracy of the proposed model is 0.55\% (+/- 0.68\%) while the R2 value between the density predictions and the ground truths is 0.9997, which is highly accurate and practical.

The correct and timely completion of the sensing and action loop is of utmost importance in safety critical autonomous systems. Crucial to the performance of this feedback control loop are the computation time and accuracy of the estimator which produces state estimates used by the controller. These state estimators often use computationally expensive perception algorithms like visual feature tracking. With on-board computers on autonomous robots being computationally limited, the computation time of such an estimation algorithm can at times be high enough to result in poor control performance. We develop a framework for codesign of anytime estimation and robust control algorithms, taking into account computation delays and estimation inaccuracies. This is achieved by constructing an anytime estimator from an off-the-shelf perception-based estimation algorithm and obtaining a trade-off curve for its computation time versus estimation error. This is used in the design of a robust predictive control algorithm that at run-time decides a contract, or operation mode, for the estimator in addition to controlling the dynamical system to meet its control objectives at a reduced computation energy cost. This codesign provides a mechanism through which the controller can use the tradeoff curve to reduce estimation delay at the cost of higher inaccuracy, while guaranteeing satisfaction of control objectives. Experiments on a hexrotor platform running a visual-based algorithm for state estimation show how our method results in up to a 10% improvement in control performance while simultaneously saving 5%–6% in computation energy as compared to a method that does not leverage the codesign.

The paper presents a novel approach by using multi- step predictions to address the adaptive sampling problem in a resources and obstacles constrained mobile robotic sensor network to efficiently monitor environmental spatial phenomena. It is first proposed to employ the Gaussian process (GP) to represent the spatial field, which can then be used to predict the field at unmeasured locations. The adaptive sampling problem aims to drive the mobile sensors to optimally navigate the environment where the sensors adaptively take measurements of the spatial phenomena at each sampling step. To this end, a conditional entropy based optimality criterion is proposed, which aims to minimize prediction uncertainty of the GP model. By predicting possible measurements the mobile sensors potentially take in a horizon of multiple sampling steps ahead and exploiting the chain rule of the conditional entropy, a multi-step predictions based adaptive sampling optimization problem is formulated. The objective of the optimization problem is to find the optimal sampling paths for the mobile sensors in multiple sampling steps ahead, which then provides their benefits in terms of better navigation, deployment and data collection with more informative sensor readings. However, the optimization problem is nonconvex, complex, constrained and mixed-integer. Therefore, it is proposed to employ the proximal alternating direction method of multipliers algorithm to efficiently solve the problem. More importantly, the solution obtained by the proposed approach is theoretically guaranteed to be converged to a stationary value. Effectiveness of the proposed algorithm was extensively validated by the real- world dataset, where the obtained results are highly promising.

This paper discusses the adaptive sampling problem in a nonholonomic mobile robotic sensor network for efficiently monitoring a spatial field. It is proposed to employ Gaussian process to model a spatial phenomenon and predict it at unmeasured positions, which enables the sampling optimization problem to be formulated by the use of the log determinant of a predicted covariance matrix at next sampling locations. The control, movement and nonholonomic dynamics constraints of the mobile sensors are also considered in the adaptive sampling optimization problem. In order to tackle the nonlinearity and nonconvexity of the objective function in the optimization problem we first exploit the linearized alternating direction method of multipliers (L-ADMM) method that can effectively simplify the objective function, though it is computationally expensive since a nonconvex problem needs to be solved exactly in each iteration. We then propose a novel approach called the successive convexified ADMM (SC-ADMM) that sequentially convexify the nonlinear dynamic constraints so that the original optimization problem can be split into convex subproblems. It is noted that both the L-ADMM algorithm and our SC-ADMM approach can solve the sampling optimization problem in either a centralized or a distributed manner. We validated the proposed approaches in 1000 experiments in a synthetic environment with a real-world dataset, where the obtained results suggest that both the L-ADMM and SC- ADMM techniques can provide good accuracy for the monitoring purpose. However, our proposed SC-ADMM approach computationally outperforms the L-ADMM counterpart, demonstrating its better practicality., Comment: submitted to IEEE Sensors Journal, revised version

Adaptive sampling in a resource-constrained mobile robotic sensor network for monitoring a spatial phenomenon is a fundamental but challenging problem. In applications where a Gaussian Process is employed to model a spatial field and then to predict the field at unobserved locations, the adaptive sampling problem can be formulated as minimizing the negative log determinant of a predicted covariance matrix, which is a non-convex and highly complex function. Consequently, this optimization problem is typically addressed in a grid-based discrete domain, although it is combinatorial NP-hard and only a near-optimal solution can be obtained. To overcome this challenge, we propose using a proximal alternating direction method of multipliers (Px-ADMM) technique to solve the adaptive sampling optimization problem in a continuous domain. Numerical simulations using a real-world dataset demonstrate that the proposed PxADMM- based method outperforms a commonly used grid-based greedy method in the final model accuracy.

In distributed optimization schemes that consist of a group of agents coordinated by a coordinator, the optimization algorithm often involves the agents solving private local proximal minimization subproblems and exchanging data frequently with the coordinator. Such schemes usually incur excessive communication cost, effecting the need for communication reduction in distributed optimization. Gaussian Processes (GPs) have been shown to be effective for learning the agents’ proximal operators and hence for reducing the communication of the Alternating Direction Method of Multipliers (ADMM). We combine this learning-based approach with an adaptive uniform quantization approach to achieve even higher communication reduction. Our approach exploits the probabilistic prediction of the GPs to adapt and refine the quantizers along the progress of the ADMM algorithm. Moreover, following a linear minimum mean square error estimation (LMMSE) approach, we improve the GP regression and hyperparameter tuning by taking into account the statistics of the resulting quantization errors. The proposed approach can achieve significant communication reduction for ADMM without sacrificing the convergence nor the optimality even with small numbers of quantization levels, as demonstrated in simulations of a distributed optimal power dispatch application.

Gaussian Process based Model Predictive Control (GP-MPC), where the system model is learned from data by a Gaussian Process (GP) statistical model, has gained increasing interests in the control community. This is due to the modeling flexibility of GPs and their ability to provide probabilistic estimates of prediction uncertainty, which can be used to assess and guarantee the performance or safety of a control system. However, GP-MPC optimization problems are typically non-convex and highly demanding, and scales poorly with the model size, causing unsatisfactory solving performance, even with state-of-the-art solvers. These drawbacks hinder the application of GP-MPC in large-scale systems and in real-time control. Our previous work has developed a new concept, linearized Gaussian Process (linGP), and a linGP-based Sequential Convex Programming (linGP-SCP) algorithm that can accelerate solving GP-MPC problems significantly. However, the algorithm ignored uncertainty propagation in its multi-step GP predictions, resulting in over-confident predictions and an over-optimistic controller. This paper completes linGP-SCP by incorporating uncertainty propagation into GP-MPC and overcoming the computational challenge of uncertainty propagation to maintain the scalability and good performance of the overall algorithm. The improved algorithm is validated in two numerical examples, including a real-time control example of swinging up a single-arm pendulum.

Building physics-based models of complex physical systems like buildings and chemical plants is extremely cost and time prohibitive for applications such as real-time optimal control, production planning and supply chain logistics. Machine learning algorithms can reduce this cost and time complexity, and are, consequently, more scalable for large-scale physical systems. However, there are many practical challenges that must be addressed before employing machine learning for closed-loop control. This paper proposes the use of Gaussian Processes (GP) for learning control-oriented models: (1) We develop methods for the optimal experiment design (OED) of functional tests to learn models of a physical system, subject to stringent operational constraints and limited availability of the system. Using a Bayesian approach with GP, our methods seek to select the most informative data for optimally updating an existing model. (2) We also show that black-box GP models can be used for receding horizon optimal control with probabilistic guarantees on constraint satisfaction through chance constraints. (3) We further propose an online method for continuously improving the GP model in closed-loop with a real-time controller. Our methods are demonstrated and validated in a case study of building energy control and Demand Response.

The complexity and diversity of future energy systems will require co-simulation solutions that enable the integration of tools from multiple domains for research and development. We introduce an open-source framework, OpenBuildNet, for distributed co-simulation of large-scale smart energy systems. Using a loose-coupling approach to co-simulate parallel processes, it can leverage and seamlessly integrate specialized simulation and computation tools in a common platform. Users can therefore benefit from the capabilities of state-of-the-art and widely used tools in each domain. OpenBuildNet is scalable and highly flexible as it uses a decentralized architecture, message-based communication, and peer-to-peer data exchange between subsystem nodes. It also provides a set of easy-to-use software tools tailored for researchers and engineers. This paper presents the architecture and tool suite of OpenBuildNet, and demonstrates its usefulness in a case study of controlling multiple buildings for demand response.

In this work, we study the optimal scheduling problem for embedded control systems. Given a number of numerical computation tasks that must be executed on an embedded computer, the goal is to find a periodic schedule that minimizes the error of the implementation with respect to the ideal system. In this paper, we pose two open problems : the decidability of the optimal scheduling problem for timed-triggered controllers (OSTTC) and the time complexity of the corresponding bounded version of the problem. We conjecture that the former is undecidable and that the latter is NP-Complete. In view of the previous conjectures, we present a Genetic Algorithm (GA) for the OSTTC problem. Initial experimental results of GA-OSTTC are very promising when compared to random search and exhaustive bounded search.

Energy-efficient control mechanisms are necessary to manage the ever increasing energy demand. Recently several tools for building energy consumption control have been proposed for small (e.g. homes) [8] and large (e.g. offices) buildings [3][6][1]. The mechanism each tool uses is different, e.g. HVAC control [3] and appliance rescheduling [8], but they share the goal of improving consumption of the buildings with respect to a given cost function. Some examples of cost functions are reduced energy consumption, reduced electricity bill, lower peak power, and increased ancillary service participation. The tools however do not capture the impacts of their control actions on the grid. These actions can lead to supply/demand imbalance and voltage/frequency deviation and thus, threaten grid stability. Utilities can take protective actions against those who cause instability by increasing electricity price or even momentarily disconnecting them from the grid. The effects of these protective actions can be so severe that the savings obtained by building management tools might disappear.

Uncertainty affects all aspects of building performance: from the identification of models, through the implementation of model-based control, to the operation of the deployed systems. We present IMpACT, a methodology and a toolbox for analysis of uncertainty propagation for building inverse modeling and controls. Given a plant model and data from the building, IMpACT automatically evaluates the effect of the uncertainty propagation from sensor data to model accuracy and control performance. We also present a statistical method to quantify the bias in the sensor measurement and to determine near optimal sensor placement and density for accurate signal measurements. In our previous work, we considered the end-to-end propagation of uncertainty in the form of fixed bias in the sensor data. In this paper, we extend the method to work with random errors in the sensor data, which is more realistic. Using a real building test-bed, we show how performing an uncertainty analysis can reveal trends about inverse model accuracy and control performance, which can be used to make informed decisions about sensor requirements and data accuracy.

Hybrid energy systems, which consist of a load powered by a source and a form of energy storage, find applications in many systems, e.g., the electric grid and electric vehicles. A key problem for hybrid energy systems is the reduction of peak power consumption to ensure cost-efficient operation as peak power draws require additional resources and adversely affect the system reliability and lifetime. Furthermore, in some cases such as electric vehicles, the load dynamics are fast, not perfectly known in advance and the on-board computation power is often limited, making the implementation of traditional optimal control difficult. The authors aim to develop a control scheme to reduce the peak power drawn from the source for hybrid energy systems with limited computation power and limited load forecasts. The authors propose a scheme with two control levels and provide a sufficient condition for control of the different energy storage/generation components to meet the instantaneous load while satisfying a peak power threshold. The scheme provides performance comparable to Model Predictive Control, while requiring less computation power and only coarse-grained load predictions. For a case study, the authors implement the scheme for a battery-supercapacitor-powered electric vehicle with real world drive cycles to demonstrate the low execution time and effective reduction of the battery power (hence temperature), which is crucial to the lifetime of the battery.

A fundamental problem in the design of closed-loop Cyber-Physical Systems (CPS) is in accurately capturing the dynamics of the underlying physical system. To provide optimal control for such closed-loop systems, model-based controls require accurate physical plant models. It is hard to analytically establish (a) how data quality from sensors affects model accuracy, and consequently, (b) the effect of model accuracy on the operational cost of model-based controllers. We present the Model-IQ toolbox which, given a plant model and real input data, automatically evaluates the effect of this uncertainty propagation from sensor data to model accuracy to controller performance. We apply the Model-IQ uncertainty analysis for model-based controls in buildings to demonstrate the cost-benefit of adding temporary sensors to capture a building model. We show how sensor placement and density bias training data. For the real building considered, a bias of 1% degrades model accuracy by 20%. Model-IQ's automated process lowers the cost of sensor deployment, model training and evaluation of advanced controls for small and medium sized buildings. Such end-to-end analysis of uncertainty propagation has the potential to lower the cost for CPS with closed-loop model based control. We demonstrate this with real building data in the Department of Energy's HUB.

This paper looks at the problem of peak power demand reduction for intermittent operation of radiant systems in buildings. Uncoordinated operation of the circulation pumps of a multi-zone hydronic radiant system can cause temporally correlated electricity demand surges when multiple pumps are activated simultaneously. Under a demand-based electricity pricing policy, this uncoordinated behavior can result in high electricity costs and expensive system operation. We have previously presented Green Scheduling with the periodic scheduling approach for reducing the peak power demand of electric radiant heating systems while maintaining indoor thermal comfort. This paper develops an event-based state feedback scheduling strategy that, unlike periodic scheduling, directly takes into account the disturbances and is thus more suitable for building systems. The effectiveness of the new strategy is demonstrated through simulation in MATLAB.

In large building systems, such as a university campus, the air-conditioning systems are commonly served by chiller plants, which contribute a large fraction of the total electricity consumption of the campuses. The power consumption of a chiller is highly affected by its Coefficient of Performance (COP), which is optimal when the chiller is operated at or near full load. For a chiller plant, its overall COP can be optimized by utilizing a Thermal Energy Storage (TES) and switching its operation between COP-optimal charging and discharging modes. However, uncoordinated mode switchings of chiller plants may cause temporally-correlated high electricity demand when multiple plants are charging their TES concurrently. In this paper, a GS approach, proposed in our previous work, is used to schedule the chiller plants to reduce their peak aggregate power demand while ensuring safe operation of the TES. We present a scheduling algorithm based on backward reach set computation of the TES dynamics. The proposed algorithm is demonstrated in a numerical simulation in Mat lab to be effective for reducing the peak power demand and the overall electricity cost.

In this paper we look at the problem of peak power reduction for buildings with electric radiant floor heating systems. Uncoordinated operation of a multi-zone radiant floor heating system can result in temporally correlated electricity demand surges or peaks in the building's electricity consumption. As peak power prices are 200–400 times that of the nominal rate, this uncoordinated activity can result in high electricity costs and expensive system operation. We have previously presented green scheduling as an approach for reducing the aggregate peak power consumption in buildings while ensuring that indoor thermal comfort is always maintained. This paper extends the theoretical results for general affine dynamical systems and applies them to electric radiant floor heating systems. The potential of the proposed method in reducing the peak power demand is demonstrated for a small-scale system through simulation in EnergyPlus and for a large-scale system through simulation in Matlab.

While building simulation software tools are excellent at carrying out accurate and realistic building simulations they only provide very basic control methods. On the other hand, control engineers and researchers have explored advanced control strategies for energy-efficient operation of a building, but more often than not, such methods are based on simplified physical models instead. We present MLE+, a tool for energy-efficient building automation design, co-simulation and analysis. The tool leverages the high-fidelity plant simulation capabilities of building modeling software, EnergyPlus, and the scientific computation and design capabilities of Matlab/Simulink for controller design. MLE+ facilitates integrated building simulation and controller formulation with integrated support for system identification, control design, optimization, simulation analysis and communication between software applications and building equipment. In this demo, we will show how MLE+ can be used for systematic design of controllers for energy-efficient building model in EnergyPlus. We will also show the capability of MLE+ to control Heating Ventilation and Air Conditioning (HVAC) devices in buildings through the BACnet protocol.

We present MLE+, a tool for energy-efficient building automation design, co-simulation and analysis. The tool leverages the high-fidelity building simulation capabilities of EnergyPlus and the scientific computation and design capabilities of Matlab for controller design. MLE+ facilitates integrated building simulation and controller formulation with integrated support for system identification, control design, optimization, simulation analysis and communication between software applications and building equipment. It provides streamlined workflows, a graphical front-end, and debugging support to help control engineers eliminate design and programming errors and take informed decisions early in the design stage, leading to fewer iterations in the building automation development cycle. We show through an example and two case studies how MLE+ can be used for designing energy-efficient control algorithms for both simulated buildings in EnergyPlus and real building equipment via BACnet.

In large energy systems, peak demand might cause severe issues such as service disruption and high cost of energy production and distribution. Under the widely adopted peak-demand pricing policy, electricity customers are charged a very high price for their maximum demand to discourage their energy usage in peak load conditions. In buildings, peak demand is often the result of temporally correlated energy demand surges caused by uncoordinated operation of subsystems such as heating, ventilating, air conditioning and refrigeration (HVAC&R) systems and lighting systems. We have previously presented green scheduling as an approach to schedule the building control systems within a constrained peak demand envelope while ensuring that custom climate conditions are facilitated. This paper provides a sufficient schedulability condition for the peak constraint to be realizable for a large and practical class of system dynamics that can capture certain nonlinear dynamics, inter-dependencies, and constrained disturbances. We also present a method for synthesizing periodic schedules for the system. The proposed method is demonstrated in a simulation example to be scalable and effective for a large-scale system.

Building systems such as heating, air quality control and refrigeration operate independently of each other and frequently result in temporally correlated energy demand surges. As peak power prices are 200–400 times that of the nominal rate, this uncoordinated activity is both expensive and operationally inefficient. We present an approach to fine-grained coordination of energy demand by scheduling the control systems within a constrained peak while ensuring custom climate environments are facilitated. The peak constraint is minimized for energy efficiency, while we provide feasibility conditions for the constraint to be realizable by a scheduling policy for the control systems. The physical systems are then coordinated by the scheduling controller so as both the peak constraint and the climate/safety constraint are satisfied. We also introduce a simple scheduling approach called lazy scheduling. The proposed control and scheduling strategy is implemented in simulation examples from small to large scales, which show that it can achieve significant peak demand reduction while being efficient and scalable.