Tsuneo Kawamura

A Mini- grain Yield Sensor Compensating for the Drift of its Own Output*

A very small (0.021 kg) grain yield sensor, comprising an octagonal ring load cell and an impact plate was developed for easily mounting onto grain combines used in East Asian countries. During the interval when the sensor is receiving no impact from the grain, the signals are averaged to determine the instantaneous zero-point; thus the sensor compensates for its own long-term drift. The sensor was mounted on a two-row jidatsu combine, and the weight of grain in the grain tank was estimated in the field experiment. The relative error in the validation without compensation was 19%, whereas that with the compensation decreased to 1.5%. The recommendable range of the threshold for detecting the impact was 0.5 to 4 times the root-mean squared average of the output without throughput of grain.

Impact-by-impact Sensing of Grain Flow on Jidatsu Combine

Abstract To enhance the accuracy of an impact-based yield sensor installed inside the grain tank of a combine, we measured the individual impacts of intermittent grain flow accelerated by an auger blade. Non-linear calibration was modeled to relate each impulse received by the sensor to the weight of grain released at a single rotation of the auger blade, taking into account the rotational speed of the auger. The parameters were optimized through pairs of signal recording and grain weight at harvest in situ . The relative error of calibration was less than 2%. The proposed instrumentation and signal processing showed robustness to the flow rate of grain, and the maximum relative error of validation was 3.5%.

In-situ non-linear calibration of grain-yield sensor — Optimization of parameters for flow rate of grain vs. force on the sensor —

Abstract In-situ calibration of a mini-yield sensor (mounted on a five-row grain combine) with a non-linear relation to the flow rate was examined. Instead of measuring or controlling the flow rate of grain for calibration purposes, 10 or 12 pairs of grain weights and signal recordings were collected directly in the fields; three such data sets (weights and signals) were obtained. Two parameters of the relation were optimized so as to minimize the standard error. The relative error of validation was 3% to 5% with data sets of a wide range of flow rates, whereas it was up to nearly 10% with a data set of low flow rates. The optimized parameters varied with each data set, but those yielding low errors were common in the response surface of error, regardless of the data sets.

Development of Fluidics for Driving and Steering Unit of Orchard Sprinkler Boat: Drag Coefficient of the Boat and Fluidics Thrust?

Abstract Small boat with no screw or rudder has been used for showering the canopies of orchards in Thailand. During the showering operation, the boat moves forward by a reaction force of two flat nozzles used for showering. This study discusses the drag coefficient of the boat and the feasibility of using fluidics as a driving and steering unit of the boat in place of conventional flat nozzles. A one-fifth scale model of the boat and a fluidics were built to determine its drag coefficient, the thrust imparted by the fluidics and boat traveling speed. The drag coefficient of the model was found to be inversely proportional to the Froude number and effective in predicting the traveling speed of a boat.

Microphone sensor for grain yield monitoring

A microphone sensor was developed as an alternative to a load cell for detecting flow rate of grain for economical installation, and its performance was evaluated on a 0.9 m wide Japanese-style (jidatsu) rice combine. The sensor consisted of a 60×40×3 mm steel plate and a ceramic earphone glued behind it to receive impacts of a portion of grain conveyed into the grain tank. In the field experiment, the combine harvested 1,570 kg of rice grain over seven runs, and the root-mean squared relative error of calibration was 3.5%.

Development and Improvement of Parallel Yield Sensors for Measuring Individual Weights of Onion Bulbs

Abstract Measuring the weights of individual tubers, bulbs, and fruits on the harvester may bring progress to Precision Agriculture and enhance their market value. In order to measure the individual weights on an onion picker in situ, we developed two impact-based parallel yield sensors and carried out the experiments in different cropping years. Both sensors (the indirect sensor and the direct sensor) consisted of plural load cells and sponge rubber cushions. The indirect sensor was set near the end of the conveyor of the picker and a bounce plate consisting of an acrylic plate and polyurethane cushion was set between the indirect sensor and the conveyor. On the other hand, the direct sensor was set directly under the end of the conveyor. Calibration were first carried out off the field to relate the impulses received by the sensor to the individual weights of the bulbs, resulted in the standard error of 26.3g and 16.1g, the root-mean squared relative error of 8.8% and 12.2% for the indirect sensor and the direct sensor, respectively. As the validation, field measurements were carried out. The bulbs were harvested in 12 and 26 containers (20kg each) for the indirect sensor and the direct sensor, respectively. After the harvest, we weighed all the bulbs in each container manually and performed the Kolmogorov-Smirnov tests of the distributions of the actual and the estimated weights at the significance level of 5 %. Using the indirect sensor, the similarity between the actual and estimated weight distributions was accepted for 9 of 12 containers. The difference in the estimated number of bulbs was about ±3 per container, caused mainly by the same bulb colliding twice or by no bulb contacting the sensor at all. On the other hand, using the direct sensor, the similarity of the distributions was accepted for 23 of 26 containers. The overestimation of the number of bulbs was a maximum of 40 bulbs per container containing about 100 bulbs.

Development of simplified grain-yield sensors for general-purpose combine harvesters

Abstract Measurement of yield variability in crop fields is important for implementation of Precision Agriculture. Yet, the sensors for such purpose are expensive and difficult in the maintenance especially in East Asian countries. We therefore developed and installed two simplified yield sensors: a force strain gauge sensor and an acoustic microphone sensor to estimate wheat grain weight. Impact waves from each sensor were extracted and calibrated with linear and non-linear models to actual grain weights. The output of the acoustic sensor was large enough without amplification and was not as affected by the vibration of the combine as that of the force sensor. Contrary to such advantageous expectation, however, the calibration of the force sensor showed the standard error of 27.9kg, and the relative error of 5.8%, whereas for the acoustic sensor, the standard error was 49.6kg and the relative error was 11.4%.

Development of Driving and Steering Unit of Orchard Sprinkler Boat Using Fluidics: Switching Characteristics, Bow Thrust of the Model Fluidics and Boat Turning Radius

Abstract A model fluidics (54 mm × 28 mm) was studied for its applicability as a steering unit to a sprinkler boat. A water jet was deflected at an inlet pressure of more than 2.0 kPa, where the velocity of the water jet is was more than 2.2 m/s. The bow thrust of the fluidics when attached onto the model boat (36 cm × 16 cm) was 0.02 to 0.12 N and quadratic to the increasing velocity of the water jet ranging between 2.2 to 5.4 m/s. Investigation of boat turning revealed that the turning trajectory transcribed a steady turning circle with a turning radius of 0.6 to 1.1 m at a peripheral velocity ranging between 0.15 and 0.28 m/s.

A Mini Grain-Yield Sensor and In-Situ Non-Linear Calibration – Impact-by-Impact Sensing to Compensate for it Own Drift and to Preserve Non-Linearity for Enhanced Accuracy –

We have developed an impact-based mini- grain yield sensor (6 cm × 4 cm) to achieve both handiness and accuracy especially for small-scale combines. For the accuracy, signal processing of the sensor was improved such that: 1) the output of the sensor was synchronized to the rotation of the grain auger whose upper edge is equipped with a blade for releasing the grain, 2) actual impacts of the grain were distinguished especially at a low flow rate of grain, and 3) zero-point of the sensor was compensated for automatically in the signal processing. For the handiness of calibration, 1) a non-linear model was proposed to relate individual impulse to the individual grain weight released at single rotation of the blade, and 2) the parameters were optimized simply using pairs of the signal recordings and the grain weights obtained in situ upon the harvest. In the field experiment, the sensor was installed on a 4-row Japanese-style rice (jidatsu) combine, and a total of 24 runs of wide flow rates of grain (0.17 – 0.84 kg/s) were harvested in two days of one week apart. The runs were divided in sequence into two datasets and calibration and full cross-validation were carried out. Relative errors of calibration were 2.4% and 2.7%, whereas those of validation were 2.2% and 2.5%. The method of signal processing showed robustness to extreme variation in flow rate of grain suited for practical applications.