Peter-Jules van Overloop

Delivery-Oriented Hierarchical Predictive Control of an Irrigation Canal: Event-Driven Versus Time-Driven Approaches

In this paper, we present the concept of a hierarchical predictive controller used for irrigation canals. The motivation behind this paper is the need in the field of irrigation to deliver water to farmers fast, but with minimal resources involved, as the communication links in the field are not dependable in practice. In response to such a control problem, we propose a hierarchical controller: the lower control layer is formed by decentralized proportional integral (PI) controllers and the higher control layer is constituted by a centralized predictive controller, the purpose of which is to control the inflow to the canal and, importantly, to coordinate the local controllers by modifying their setpoints. Having in mind the restrictions on the available communication infrastructure and the control equipment already present, the scheme is designed to be event driven, i.e., activated when there are either delivery requests or non-delivery-related events of any sort, requiring special care on top of the control provided by the PI controllers. We also study a time-driven formulation with an additional postprocessing step to avoid excessive negligible setpoint modifications. We compare the event-driven formulation and the time-driven formulation theoretically as well as by means of a simulation study for the West-M irrigation canal in Phoenix, Arizona, illustrating the findings of this paper. It is shown that the event-driven controller is able to provide a good balance between the control performance and the required update frequency of the control settings.

Multivariable model predictive control of water levels on a laboratory canal

Automatic control of irrigation canals can reduce the loss of water in considerable amounts, therefore it is generating ecologic and economic benefits. There have been many different types of automatic controllers developed, but only few of them had the opportunity of being tested on the field due to the long delay time and the inconveniences of interrupting the operation of the irrigation. Therefore, the automatic controllers developed for large irrigation canals should be tested before by means of numerical simulations and/or laboratory experiments. The Technical University of Catalonia possesses a laboratory irrigation canal with the length of 220m, with 3 motorized gates, and 11 level sensors that are connected to a SCADA system. This facility makes it possible to test controllers of any type, since all the instrumentation and real time operation runs within a flexible working environment running in Matlab-Simulink . The canal can be configured from one pool to three pools, which allows the development of multivariable control. A numerical model of the canal has been developed using the 1D hydrodynamic model SIC. With the help of this software it is possible to simulate the hydraulics of the canal and, due to the link between SIC and Matlab, also to test any controller developed previously in the Matlab environment. In this work a centralized multivariable model predictive controller for water levels is developed and validated by means of numerical simulation.

Incorporating transport over water in the multi-objective water management of the Lake IJssel area in The Netherlands

Safety and navigation are two important elements in the multi-objective water management of the Lake IJssel area in The Netherlands including Lake IJssel, Lake Marker and North Sea Canal. In order to maintain these important stakes and to make them explicit, transport over water is treated as individual agent that has to negotiate its objectives within the total operational water management. In this paper, dual decomposition is used in a distributed model predictive controller applied to the main control structures in this area. The controller is evaluated on a realistic scenario simulated on a detailed model of the water system.

Portable, automatic water level estimation using mobile phone cameras

Frequent and more accurate water level measurement will allow for a more efficient distribution of water, resulting in less water loss. Therefore in this paper we propose a novel method for accurate water level detection and measurement applied on images of staff gauges, retrieved from mobile device camera. In the first step, we propose fast segmentation of the staff gauge using a 2-class random forest classifier based on a feature vector of textons. To obtain bars and numbers we apply Gaussian Mixture Model segmentation followed by optical character recognition based on random forest classifier and bar detection using shape moments. Based on the recognized lines and numbers a quadratic function for the water level measurement to obtain metric values is introduced. Finally, we propose a novel step for the water level line detection. The water level function and the detected water line provide the value of the water level based on the units on the staff-gauge. The water level can then be uploaded to a central server to determine if water flow needs to increase or decrease. Testing with a real world images from Dutch canals show very accurate detection with many different staff-gauge locations despite complex challenges of viewpoints variations, low quality images as well as changing illumination conditions.

Hierarchical control of irrigation canals in the presence of disturbances: Framework and comparison

We study a control problem of delivering water to farmers through an irrigation canal and introduce a hierarchical controller with a Coordinator that by employing Model Predictive Control principles coordinates local canal reaches by modifying setpoints only when it is needed. Once the setpoints are set, the Coordinator does not interfere with the functioning of the local sites and the canal is fully controlled by local PI controllers located at each gate. Therefore the communication between the centralized controller and the local sites is kept minimal, which is motivated by the communication restrictions that are present in the field of irrigation. We consider three predictive control designs, namely a nominal controller, and two robust designs: a constraint tightening controller adapted to fit our application and a min-max controller. We present a numerical example to compare the performance obtained by the three controllers. It is found that for the given case study with a small disturbance realization, the nominal controller performs better than the robust controllers, the behaviors of which prove overly conservative.