Search

Comunicación presentada en IEEE 15th International Conference on Control and Automation (ICCA), 16-19 July 2019, Edinburgh, United Kingdom,. In this paper, we have developed an Economical Model Predictive Control (EMPC) for a Wastewater Treatment Plant (WWTP) with the use of a standard semidefinite programming solver. In this case, the objective has been to keep the ammonium concentration in the effluent under limits manipulating the air insufflation pumps at the biological reactor and an internal recycle valve. The minimized cost function consists of the product of the energy consumed by the air insufflator and the cost of the electricity, taking into account the variations of the tariffs over the day. We have simulated the behaviour of the WWTP using the Benchmark Model Simulation n° 1 (BSMI), and we have developed a linear prediction model in order to apply the EMPC method.

In this work we present a free software tool for the design and simulation of PID controllers with event based sampling. The tool allows to easily obtain the PID that minimizes the disturbance IAE while maintains the control action jumps under a desired bound, for a given robustness. The application calculates the phase and gain margins with respect the existence of limit cycles for SSOD sampling and RQ sampling, taking into account the describing function approximation. Therefore, the application can be used to design a PID that guarantees the absence of oscillations due to limit cycles with certain margins. A time response simulation window allows to simulate the response of the event based controlled system to setpoint and disturbance changes, choosing between SSOD and RQ with hysteresis. Performance data are automatically calculated and shown. The application is useful to analyze the implications of using PI and PID controllers with different event based sampling strategies, especially the existence of limit cycles and the compromise between control action jumps and performance.

Ponencia presentada al 3rd Conference on Control and Fault-Tolerant Systems (SysTol), Barcelona, Spain, Sept. 7-9, 2016 This work addresses the fault diagnosis problem for LTI systems under the presence of Gaussian noises through model-based proportional-integral observers with predefined gains. We propose an integrated design of residual generators and evaluators which takes into account the trade-off between physically meaningful parameters such as the false alarm rate, the minimum isolable faults and the integral squared error of the residuals under step faults. Dynamical fault isolation is also taken into account. In order to solve this design problem, we present two different approaches: one based on the steady-state Kalman filter and another based on convex optimization techniques. This work has been supported by grants FPU14/01592 from MECD and PI15734 from Universitat Jaume I, and by projects P11B2015-42 from Universitat Jaume I de Castellón o and MICINN project number TEC2015- 69155-R.

This paper presents a study about the timing behavior of software designed under the IEC 61499 standard. In particular, the effect of different event management strategies on the execution time of applications is analyzed. The study is focused on the closed loop control applications in which the controller algorithm is implemented according to the standard. In particular, the PID control algorithm is considered since it is a well known algorithm that is widely used in industry. The execution time of software implementing both single and multiple PID controllers are statistically characterized by means of an extensive experimental study. The results demonstrate that the timing behavior of the PID algorithms is strongly influenced by how the events are managed.

This article presents an accurate study about the validity of the tuning rule for symmetric-send-on-delta (SSOD) based PID controllers which was presented by the authors in a previous work. First we obtain the conditions under which the rule is valid and then we study its fulfillment for an extensive batch of 134 models. The results demonstrated that most of the models used in practical applications fulfill the validity conditions, and consequently the rule can be useful for tuning SSOD based PID for a wide range of systems.

In this paper we study the behaviour of event based PID controllers implemented with two different events based sampling strategies: Symmetric Send on Delta (SSOD) and Regular Quantisation (RQ). The effect of the derivative filter coefficient of the PID in the overall system response is also analysed. The study is carried out in an electro-mechanical experimental platform with CD brushless motors and a networked control system.

In this work, we consider a state estimation problem for linear uncertain discrete-time systems over a network with dropouts. The uncertainty of the plant model is described by a norm-bounded time-varying parameter that affects the system matrix. We design a jump filter that minimizes an upper bound of the trace of the state estimation error covariance for all possible parameter uncertainties and whose gains are selected from a precalculated finite set depending on the possible measurement reception scenarios. The presented procedure allows us to relate the complexity of the jump filter to the achievable estimation performance.

In this paper we analyse the conditions that produce oscillations in event based PID controllers. A Symmetric Send-On-Delta scheme (SSOD) is assumed, and the describing function is the tool used to address the problem. Once the conditions for oscillations are established, a new robustness to oscillations performance measure is introduced. Then a very simple and intuitive rule of thumb is given about how to tune PID controllers to avoid the limit cycles. The rule presented here entails with the classical concept of phase margin, providing the key to design SSOD based PID by using continuous PID tuning methods. The proposed ideas are tested in simulations with a batch of models widely used in PID design methods testing.

In this work, the observer design for systems operating over communication networks with previously unknown packet dropout rate (PDR) is addressed. It is assumed that the PDR is time-varying and that it can be estimated online by means of the acknowledgement on new data arrival. The observer gains depend on the estimated PDR and they are designed minimizing the H∞ norm from disturbances and measurement noises to estimation error over all the possible PDRs. The observer gains are rational (including as a particular case a polynomial) functions of the estimated PDR of the desired order. An optimization over polynomials has been carried out in order to find a solution for the proposed filter.

In this work, a new strategy to minimize the use of the network in a closed loop system over networks is addressed, leading to a co-design procedure of the observer, controller and the policy for message sending. The sensor nodes implement a send-on-delta approach, sending new data only when there is a considerable deviation from the last sent measurement. This strategy is useful to minimize the network usage necessities from wireless sensors in order to reduce the energy consumption and to maximize its useful life. The inferential controller node implements a gain scheduling approach that takes into account the availability of new received data to estimate the state and compute the control action. The performance of the closed loop system is analyzed through ℌ ∞ norm, and this norm is used to design the gains of the observer and the controller, as well as the output variations that induce the sensors to send new outputs to the estimator node, while guaranteeing a given level of performance on the RMS norm of the controlled state. The design approach is based on an optimization procedure with linear and bilinear matrix inequalities constraints that is solved iteratively.

In this paper, a new event based PI control algorithm with adaptive thresholds is presented. The main advantage over the fixed threshold algorithms is that it removes the dependence between the magnitude of the input (set point or disturbance) and the number of events, to prevent the generation of unnecessary events when the magnitude of the input increases. Clear guidelines and methodologies for calculating the parameters of the new controller in a systematic way are proposed to obtain a good compromise between quality of control and number of events generated. The advantages of the proposal are illustrated by simulation using transfer functions that describe most of the dynamics present in industrial applications, including systems with large delays and non-minimum phase.

In this work, new analysis and prediction strategies for the power control in wind generators that use a doubly fed induction generator are addressed. The wind generator model and control strategies most widely used are first presented, including strategies used to punctually increase the power generation in order to respond to power demands derived from power system frequency deviations. A new on-line prediction strategy is addressed based on a wind observer plus the use of the different norms that the dynamic model presents from the inputs (the wind or the power demand) to the outputs (rotor speed, accelerations, and power generation). For that purpose, the wind observer and the computational derivation of the different input-output performances are developed via optimization over polynomials techniques. The developed analysis and predicting strategies are applied to a wind generator model and controller proposed in the literature, showing the effectiveness of the presented approach.

In this work, the parameter identification for systems with scarce measurements is addressed. A linear plant is assumed and its output is assumed to be available only at sporadic instants of time and affected by noise measurement. The identification is carried out estimating the missing outputs in order to construct the regression vector needed by the parameter estimation algorithm and using the available output information not only to update the estimated parameter vector, but also to update the regression vector in order to fasten the convergence of the algorithm. The problem is addressed with an adaptive extended Kalman filter that estimates and correct both the parameters and the regression vector, allowing to improve the convergence speed of the algorithm with respect to other existing ones on the literature as it is shown with several examples.

In this paper, a simple method to design PID controllers in the frequency domain is proposed. The method is based on the use of a single tuning factor, defined as the quotient between the final cross over frequency and the zero of the controller. Classical methods described in the basic control books propose to fix the above parameter in a predetermined value (usually 10), independently of the process model. Several examples show that this fixed value is not the best choice in general, in the sense that the behavior can be greatly improved if an adequate tuning of that parameter is done. A tuning procedure for the parameter is also proposed, based on the maximization of a controller gain. The procedure is valid for PI, PD and PID controller design, and is tested with a set of well known benchmark systems for PI and PID design and compared with other (more complex) approaches.

In this article we make use of dimensional analysis in order to obtain a model for control design purposes. The industrial system to be controlled is the ceramic tile glazing process. The dimensional analysis allows to go beyond first principles modeling and to consider modeling errors. The model obtained is used for feedforward control. Simulation examples show that it is possible to control the deposited glaze mass by just measuring the glaze temperature. However the measurement of the glaze flow results to be a critical variable if high accuracy on the deposited mass is pursued.

In this work, an inferential scheme for networked control systems is adressed. An LTI plant with disturbances and measurement noise is assumed to be controlled by a controller that communicates with the sensor and the actuator through a network. This network is assumed to have variable accessibility constraints due to a shared use of it, network-induced delay and packet dropout. In this work a control algorithm is proposed such that the available (and scarce) outputs that arrive delayed on time from its sampling instant are used to make a prediction of the system evolution. That prediction is then used to calculate the control actions that should be applied in the actuator. The algorithm takes into account the possibility of control action drop due to network constraints in order to calculate the control action. This networked control scheme is analyzed in depth and the separation principle between predictor and controller is demonstrated. Both the predictor and controller designs are addressed taking into account the disturbances, the noise measurement, the scarce availability of process output samples and the scarce capability to update the control actions due to the network constraints. Both designs are based on the probability of network availability to transmit the measurements and the input desirable updates on time. The time-varying sampling periods for the input and output of the plant due to the network constraints have been specified and bounded as a function of the probability of success on transmitting the output samples and input updates on a specified time. For both designs H ∞ performance has been established and LMI design techniques have been used to achieve a numerical solution.

In this paper, the analysis of the closed-loop dynamics of a control system with scarce random sampling is addressed. The control strategy is based on the use of a conventional single-rate controller combined with a model based output estimator that estimates the unmeasured intersampling outputs at the control (fast) rate. The proposed estimator takes into account the past measured outputs, as a difference with inferential control schemes. The separation principle between the estimator dynamics and the closedloop one is demonstrated. The design of the control system is simplified, since the controller and the estimator can be designed separately. Some examples illustrate how an adequate estimator design improves the closed-loop performance with respect to the inferential control scheme (open loop estimator), solving the problem of instability that appears when the open loop system is unstable.

In this paper, the problem of predicting the output of a system at regular periods from irregularly sampled signals is addressed. A model based predictor that takes into account the possibly delayed measurement signals is used. Robustness of the predictor to the time-delays and data availability as well as disturbance and noise attenuation is dealt with via Hinfin performance. A design strategy is proposed based on the available disturbances information