Search

This study breaks new ground by assessing the influence of individual and collective errors in weather forecast variables and errors in model parameters on the prediction error of greenhouse power and gas demand. To achieve this a sample-based and a Polynomial Chaos based sensitivity analysis using higher order sensitivity indices is proposed. This is accompanied by a sensitivity analysis of the impact of reducing individual weather forecast errors on greenhouse energy demand prediction error. The findings of this study show that weather forecast errors have a far greater role in creating mean gas (27 W m − 2 ) and power (24 W m − 2 ) prediction uncertainty than parametric errors (5.7 W m − 2 and 4.6 W m − 2 ). In addition, and crucially, weather forecast and parameter errors were found to be independent factors. Reducing weather forecast error exhibited large diminishing returns with the reduction in prediction error. For instance, a scenario where the forecast error of all variables is reduced by 80–90 % resulted in only a 50 % decrease in gas and electrical power prediction error. The radiation forecast errors emerged as the primary contributor to power demand prediction errors, exhibiting the potential to reduce the power demand prediction error by approximately 60 %. Reductions of forecast errors in wind and outdoor air temperature were identified as the predominant contributors, offering a respective potential for a 17 % reduction in gas demand prediction error. • A gas demand and electrical power uncertainty analysis was performed. • This analysis considered both weather forecast and model parameter error. • Weather forecast error had a greater impact than parametric error. • An 80 % forecast error reduction creates a 50 % decrease in energy prediction error. [ABSTRACT FROM AUTHOR]

Greenhouse design is an optimisation problem that might be solved by a model-based greenhouse design method. A sensitivity analysis of a combined greenhouse climate-crop yield model of tomato was done to identify the parameters, i.e. greenhouse design parameters, outdoor climate and climate set-points, that most strongly influence greenhouse performance. The analysis was performed for a low-tech greenhouse in Almeria, Spain, and a high-tech greenhouse in Texas, USA. A single-variate sensitivity analysis showed that outdoor climate has the strongest impact on the performance of the greenhouse system, followed by greenhouse design parameters and greenhouse climate set-points. The high impact of the outdoor conditions stresses the need to select a proper location for the greenhouse. Concerning the design parameters, the analysis revealed different results for the two locations and greenhouses studied. This emphasises that a ‘custom made’ approach to greenhouse design should be followed exploiting local conditions. In both cases, structures with a higher PAR transmission and a NIR-selective whitewash should be used. Seasonal patterns in the model sensitivity of for instance PAR, NIR and FIR emission coefficients of the cover indicate that a greenhouse with adjustable cover parameters will be advantageous over a design with fixed greenhouse cover parameters, as is usually implemented. A multi-variate sensitivity analysis revealed strong joint effects of parameters on crop yield. A joint increase of the PAR transmission and temperature set-point for ventilation favoured the crop yield for both greenhouses, stressing a simultaneous approach to both design and control of greenhouse systems. [ABSTRACT FROM AUTHOR]

The paper presents results of research on an inverse kinematics algorithm that has been used in a functional model of a cucumber-harvesting robot consisting of a redundant P6R manipulator. Within a first generic approach, the inverse kinematics problem was reformulated as a non-linear programming problem and solved with a Genetic Algorithm (GA). Although solutions were easily obtained, the considerable calculation time needed to solve the problem prevented on-line implementation. To circumvent this problem, a second, less generic, approach was developed which consisted of a mixed numerical-analytic solution of the inverse kinematics problem exploiting the particular structure of the P6R manipulator. Using the latter approach, calculation time was considerably reduced. During the early stages of the cucumber-harvesting project, this inverse kinematics algorithm was used off-line to evaluate the ability of the robot to harvest cucumbers using 3D-information obtained from a cucumber crop in a real greenhouse. Thereafter, the algorithm was employed successfully in a functional model of the cucumber harvester to determine if cucumbers were hanging within the reachable workspace of the robot and to determine a collision-free harvest posture to be used for motion control of the manipulator during harvesting. The inverse kinematics algorithm is presented and demonstrated with some illustrative examples of cucumber harvesting, both off-line during the design phase as well as on-line during a field test. [Copyright &y& Elsevier]

The paper presents an autonomous robot for removing the leaves from cucumber plants grown in a high-wire cultivation system. Leaves at the lower end of the plants are removed because of their reduced vitality, their negligible contribution to canopy photosynthesis and their increased sensitivity to fungal diseases. Consuming 19% of the total labour input, leaf removal is considered by the growers and their staff as a tedious, repetitive and costly task. Automation alleviates their job and results in a significant cost reduction. The paper describes a functional model and preliminary results of a field test of a de-leafing robot. Despite the small number of repetitions during the field test, the favourable results confirmed the feasibility of the concept of the de-leafing robot. The vision system and manipulator control performed relatively well. However, particularly, the mechanics of the end-effector were failure prone and need re-engineering. With a cycle time of 140s per two leaves on average, the execution time of the de-leafing robot is approximately 35 times slower than manual leaf picking. For economic feasibility this cycle time should be reduced. [Copyright &y& Elsevier]

At the Institute of Agricultural and Environmental Engineering (IMAG B.V.) an autonomous harvesting robot for cucumbers was developed and tested in a greenhouse in autumn 2001. Analysis of the harvest process had revealed that at a 2 ha Dutch production facility four robots are needed to replace the skilled human work force during the peak season. Then assuming a success rate of 100%, a harvest cycle might take at most 10 s per cucumber fruit. In this paper the results of the field test of the harvesting robot are reported and analysed in view of the performance criteria mentioned above. Cucumbers (Cucumis sativus cv. Korinda) were grown in a high-wire cultivation system. In four independent experiments the robot was tested. The average success rate was 74·4%. The majority of failures originated from inaccurate positioning of the end-effector at the stalk of the fruit. It was found to be a great advantage that the system was able to perform several harvest attempts on a single cucumber from different harvest positions of the robot. This improved the success rate considerably. A single successful harvest cycle took 65·2 s per cucumber. Since not all attempts were successful, a cycle time of 124 s per harvested cucumber was measured under practical circumstances. The test confirmed the ability to harvest more than one cucumber using a single set of images which reduced the cycle time of a successful harvest to 56·7 and 53·0 s if two or three cucumbers were harvested. To bridge the gap between the measured performance and the design specifications, future research focuses on improving the success rate, faster hardware and software for image processing and motion planning as well as the reduction of the motion time of the manipulator. [Copyright &y& Elsevier]

One of the most challenging aspects of the development, at the Institute of Agricultural and Environmental Engineering (IMAG B.V.), of an automatic harvesting machine for cucumbers was to achieve a fast and accurate eye–hand co-ordination during the picking operation. This paper presents a procedure developed for the cucumber harvesting robot to pursue this objective. The procedure contains two main components. First of all acquisition of sensory information about the working environment of the robot and, secondly, a program to generate collision-free manipulator motions to direct the end-effector to and from the cucumber. This paper elaborates on the latter. Collision-free manipulator motions were generated with a so-called path search algorithm. In this research the A*-search algorithm was used. With some numerical examples the search procedure is illustrated and analysed in view of application to cucumber harvesting. It is concluded that collision-free motions can be calculated for the seven degrees-of-freedom manipulator used in the cucumber picking device. The A*-search algorithm is easy to implement and robust. This algorithm either produces a solution or stops when a solution cannot be found. This favourable property, however, makes the algorithm prohibitively slow. The results showed that the algorithm does not include much intelligence in the search procedure. It is concluded that to meet the required 10 s for a single harvest cycle, further research is needed to find fast algorithms that produce solutions using as much information about the particular structure of the problem as possible and give a clear message if such a solution can not be found. [Copyright &y& Elsevier]

Optimal control systems are based on a performance measure to be optimised and a model description of the dynamic process to be controlled. When on-line implementation is considered, the performance of optimally controlled processes will depend on the accuracy of the model description used. Sensitivity analysis offers insight into the impact of uncertainty in the model parameters on the performance of the optimally controlled process. Additionally, sensitivity analysis may reveal the mechanisms underlying optimal process operation. This paper describes the methodology and results of a sensitivity analysis of an optimal control problem in greenhouse climate management. The methodology used, is based on variational arguments and requires a single solution of the optimal control problem, resulting in a computationally efficient technique. The example considered deals with economic optimal greenhouse climate management during the cultivation of a lettuce crop. The sensitivity analysis produced valuable insight into the performance sensitivity and operation of the controlled process. Both the model description of crop growth and production as well as the outside climate conditions have a strong impact on the performance. Humidity control plays a dominant role in economic optimal greenhouse climate management, emphasising the need for an accurate description of humidity effects on crop growth and production, either in terms of quantitative models or time-varying constraints on the humidity level in the greenhouse. Finally, the study revealed that the dynamic response times in the greenhouse climate are not limiting factors for economic optimal greenhouse climate control. [Copyright &y& Elsevier]

With the aim of developing a model-based method to design greenhouses for a broad range of climatic and economic conditions, a greenhouse climate model has been developed and validated. This model describes the effects of the outdoor climate and greenhouse design on the indoor greenhouse climate. For use in a greenhouse design method that focused on the optimisation of a set of design elements, the model should fulfil the following three requirements: 1) predict the temperature, vapour pressure and CO2 concentration of the greenhouse air, with sufficient accuracy for a wide variety of greenhouse designs under varying climate conditions, 2) include the commonly used greenhouse construction parameters and climate conditioning equipment, and 3) consist of a set of first order differential equations to ensure that it can be combined with a tomato yield model (of a similar structure) and to allow the use of ordinary differential equation solvers. The dynamic model was validated for four different greenhouse designs under three climatic conditions: a temperate marine climate, a Mediterranean climate and a semi-arid climate. For these conditions, the model accurately predicted the greenhouse climate for all four designs without modification of the model parameters (except for one case). In more than 78% of the cases, comparison of simulations and measurements of the indoor climate yielded a relative root mean square error of less than 10%. Given these results, the model is considered to be sufficiently accurate and sufficiently generic to be used for developing a model-based greenhouse design method. [ABSTRACT FROM AUTHOR]

Incomplete control of volunteer potato plants causes a high environmental load through increased crop protection chemical usage in potato cropping. A joint effort of industry, policy makers and science initiated a four year scientific project on detection and control of volunteer potato plants. A proof-of-principle machine for automated detection and control of volunteer potato plants in sugar beet fields has been tested in experimental fields. Machine vision-based detection at 100mm2 precision is combined with a micro-sprayer with five needles and a working width of 0.2m. The accuracy of the system was ±14mm in longitudinal direction and ±7.5mm in transverse direction. The main error source was the variability in micro-sprayer droplet velocity that caused longitudinal errors. However, 77% of volunteer plants with a size larger than 1200mm2 were successfully controlled at machine speeds up to 0.8ms−1. Within the crop row, glyphosate was applied on weed potato plants and this resulted in the unwanted death of up to 1.0% of sugar beet plants. [Copyright &y& Elsevier]

With the aim of developing a model-based method to design greenhouses for a broad range of climatic and economic conditions, a tomato yield model that describes the effects of greenhouse climate on yield was described and validated. A literature survey of temperature effects on tomato yield was performed and the main temperature effects were implemented in the model. Subsequently, the yield model was validated for four temperature regimes. Results demonstrated that the tomato yield was simulated accurately for both near-optimal and non-optimal temperature conditions in the Netherlands and southern Spain, respectively, with varying light and CO2-concentrations. In addition, the adverse effects of extremely low as well as high mean temperatures on yield and timing of first fruit harvest were simulated with fair accuracy. The simulated yield response to extreme diurnal temperature oscillations were in agreement with literature values. Given these results, the model is considered to be sufficiently accurate to be used for developing a model-based greenhouse design method. Therefore, the presented model will be integrated in a model-based design method with the aim to design the best greenhouse for local climate and economic conditions. [ABSTRACT FROM AUTHOR]

This paper describes a model to calculate the concentrations of (Z)-3-hexenol, α-pinene, α-terpinene, β-caryophyllene, and methyl salicylate in a greenhouse on the basis of their source and sink behaviour. The model was used to determine whether these volatile organic compounds (VOCs) can be used to indicate Botrytis cinerea infection in a large-scale tomato production greenhouse with a volume of 5 × 104 m3 containing 2.2 × 104 plants. Seven experiments were done to parameterise the model for these VOCs. Based on model predictions, the B. cinerea-induced increase in concentration of methyl salicylate is detectable in a large-scale tomato production greenhouse when: (a) windows are fully opened, and (b) the increase continues for at least 1 h, and (c) 5% of the plants are infected. The B. cinerea-induced increase in concentration of methyl salicylate is also detectable when: (a) windows are closed, and (b) the increase continues for at least 6 h, and (c) 5% of the plants are infected. The B. cinerea-induced increase in concentration of (Z)-3-hexenol is detectable under all conditions studied. However, it is expected that besides infected plants, many additional sources of (Z)-3-hexenol exist including plant debris and nearby field crops especially upon harvest or stress. The B. cinerea-induced increases in concentration of α-pinene, α-terpinene and β-caryophyllene are probably undetectable in a large-scale tomato production greenhouse. Therefore, it is recommended to focus on the detection of methyl salicylate to indicate B. cinerea infections in large-scale tomato production greenhouses. [Copyright &y& Elsevier]

A leaf picking device for cucumbers was designed and evaluated. The picking device is manually operated but can be used as a picking tool for a robot. The device consisted of a picking rotor composed from knives and brushes, a motor and a vacuum cleaner. The performance of removal, cutting, torque and shredding was investigated in the laboratory experiments. In the greenhouse experiments, the performance of picking and cutting was investigated. The results were as follows: (1) the highest removal success rate was achieved at a rotation speed of 1000min−1, a rotor configuration of ‘two knives and two brushes’ and an insertion speed of 50mms−1; (2) in this mechanical setting, the percentage of the summation of the smooth cut surface of the leaf stalk and the smooth cut surface with small skin was 90%; (3) the required torque was 0.09–0.96Nm and the average particle size of the shredded leaves was 7.3–21.2cm2; (4) the percentages of area of the dropped particles to the average area of leaves was 10–16%; (5) the average execution time per leaf was 1.1–2.3s. [Copyright &y& Elsevier]