Vol. 82, No. 4 Abstract

 

Development of Monitoring System Using Unmanned Aerial Vehicle and Artificial Intelligence for Potato Yield Prediction (Part 1)

Dai TANABE, Shigeru ICHIURA, Ayumi NAKATSUBO, Takashi KOBAYASHI, Mitsuhiko KATAHIRA

[Keywords: artificial intelligence (AI), deep learning, information and communication technology, normalized difference vegetation index (NDVI), potato, unmanned aerial vehicles (UAV)]

 

 We used UAV as a remote sensing device in potato cultivation, and assessed the relationship between image information and yield. The experiment was performed in a potato field with six fertilization conditions, aerial images were acquired, and growth and yield were investigated. Yield prediction models were constructed by multiple regression analysis and AI. The aerial image showed a difference in canopy. Plant height differed each of the treatment as the growth progressed. For the prediction of potato yield, multiple regression analysis based on NDVI and plant height data during the growing and flowering seasons had high accuracy. Yield prediction by AI, the accuracy is low due to imbalance of datasets and overfitting, so it is necessary to improve accuracy.

Utilization of Mechanical Parameters of the Petiole for a Simple Estimation of Tomato Leaf Area

Takashi FUKUSHIMA, Kiyoshi NAGASUGA, Yosuke INAGAKI, Kenji TAKISAWA

[Keywords: tomato leaf area, mechanical parameter, flexural rigidity, multiple regression analysis, estimation model]

 

 As the largest facility cultivation in Japan, tomato cultivation has been managed by considering the maximizing photosynthesis with the leaf area as the index. We investigated the relationship between the leaf area and the morphological and mechanical characteristics of the petiole of the leaf of two tomato varieties, based on the hypothesis that the mechanical parameters of the petiole as the supporting strength increase with increasing leaf weight/growth. Results showed a high correlation between the leaf area and the mechanical parameters of the petiole during the earlier growth stage, which supports our hypothesis, as well as the high accuracy of our leaf area estimation model using the flexural rigidity and the long diameter of the petiole.

Development of a Rapid Evaluation Method for Cane Deterioration Using Near-infrared Spectroscopy

Yoshinari IZUMIKAWA, Atsumi KINJYO, Kittipon APARATANA, Hiroo TAKARAGAWA, Shinya NAKAMURA, Munehiro TANAKA, Eizo TAIRA

[Keywords: sugarcane, deterioration, near-infrared spectroscopy, quality index, payment]

 

 Deteriorated sugarcane considerably decreases sugar production in sugar mills. This study investigated a rapid evaluation method to determine cane deterioration using near-infrared spectroscopy (NIRS). We examined the calibration models of sucrose, glucose, and fructose in deteriorated cane juice using NIRS. The ratio of sucrose to the total sugar components (pure sucrose ratio) was proposed as the quality index for cane deterioration because the conventional method could not analyze sugar content. Accuracy of the calibration models for each sugar component and pure sucrose ratio was adequate to detect cane deterioration. Moreover, deterioration of cane could be predicted using the NIRS on payment data. These results indicate that it is possible to identify deteriorated sugarcane using NIRS.

Calibration Equations for the Weight of Rough Rice Layers of Various Shapes Piled in a Grain Tank

Yasumaru HIRAI, Masakazu KASHINO, Ikuo UEKA, Muneshi MITSUOKA, Eiji INOUE, Takashi OKAYASU

[Keywords: combine harvester, load cell, yield, piled shapes of rough rice layers, center of gravity, smart agriculture, farming support system, rice]

 

 Japanese yield monitoring systems adopt a calibration equation using load cell values measured at the grain tank bottom. We evaluated the accuracy of the current calibration equation and investigated alternatives to improve the weight estimation accuracy of rough rice layers of various shapes piled in a grain tank. The root mean squared relative error of cross validation (RMSRECV) was 22 % for the current equation. The relative error was -41 % to 60 % when a 100 kg rough rice layer was piled, involving large variations of x coordinates in the center of gravity. The calibration equation using front and rear load cell values (RMSRECV: 4.4 %) was considered practical for reducing the estimation error caused by the piled shapes.

Risk Reduction Based on Accident Factor Analysis of Combine Harvesters and Consideration of Differences in Risk Factors Based on Mechanical Properties

Masami MATSUI, Yuya AOYAGI, Takuya TANAHASHI, Jun-ichi TAKEDA, Takashi FUKUSHIMA

[Keywords: combine harvester, accident, factor analysis, risk assessment, farm safety]

 

 Operators of combine harvesters have suffered serious accidents owing to the machine’s structure and working characteristics. A clarification of the risks and the relations between mechanical properties and accident types can contribute to developing effective countermeasures. In this study, the risks of accident factors were calculated, and the relationship between mechanical properties and accident types was examined. The risks of death and serious injury are expected to decrease by approximately 26.6 % by addressing four factors: blind spots, road inclination, unsafe and poor operability, soft field and slippery road. Moreover, three properties of combine harvesters (machine size, operator’s seat location, and multiple rotating elements) cause the machine to fall or operator to become entangled.

Evaluation of Solar Panel Regulations for Agrivoltaic Systems in Paddy Fields

Ryoichi GOHARA, Nobuhide TAKAHASHI, Shigeharu KATO, Tomoaki HORIE

[Keywords: paddy rice cultivation, Koshihikari, organic agriculture, agrivoltaic system, automatic solar panel control, uniaxial automatic solar tracking]

 

 Owing to global climate change and the aging population of farmers, potential reductions in crop production and a decline of agriculture are of increasing concern in Japan. Establishment of efficient agrivoltaic systems is considered important for achieving a low-carbon society and for recruiting farmers. However, the relationship between crop productivity and power generating efficiency remains to be investigated. In this study, three different solar panel regions were set up in a paddy field and the effect of the sunlight conditions on the cultivated rice was evaluated. No significant correlation was observed between the rice productivity and sunlight conditions tested. This suggested more scope for increasing the panel-to-area ratio and the number of panels on the same frames in the agrivoltaic system analyzed.

Accuracy of Predicted Value and Particle Size Composition Calibration Models Using Paddy Soil Spectra

Masakazu KODAIRA, Sakae SHIBUSAWA

[Keywords: clay, silt, sand, calibration model, prediction accuracy, outlier, relative quantitative value]

 

 When the particle size composition of agricultural soil can be determined by a rapid and simple method, it is possible to make site-specific management decisions and soil improvement according to soil variation, which is expected to improve yield, quality, and farmwork efficiency.
 In this paper, we estimated calibration model performance for clay, sand, and silt (CLSSL), using soil diffuse reflectance spectra collected from paddy fields, and investigated the accuracy of predicted values. The accuracy of the calibration model for CLSSL was R2Cal>0.90. The predicted value accuracy for 185 unknown samples was 0.37 to 0.48 for R2. The predicted CLSSL values followed the distribution trend line of measured values, and the average value error was 0.60 to 1.41 %. Therefore, it was concluded that the variation in the relative quantitative value could be visualized in the form of a soil map. In field measurement of spectra, it is necessary to establish a method to appropriately exclude outliers from predicted values.

Development of a High-Performance Levee Mower

Eiji KURIHARA, Kouhei YAMASHITA, Jun NISHIKAWA, Kazunobu HAYASHI, Tsutomu TODA, Hiroki TAMURA, Taichi KOBAYASHI, Kenji HIYOSHI

[Keywords: mower, electromotive, levee, slope, remote control operation, automatic run]

 

 In this study, we developed a high-performance levee mower. The developed levee mower can be equipped with a stable running device and can automatically run along a levee. The developed mower is an electromotive-type device with a remote-control operation. The device is adaptable to both levees and slopes. From the field test results, the work efficiency of the developed levee mower was 151.0 min/10 a when used on levees and 99.1 min/10 a when used on slopes. By operating the remote control unit, the heavy workload was reduced, and the noise, exhaust gas emission, and vibration were less pronounced when compared with those of mowers available in the market. Therefore, the mowing surroundings of the operator improved significantly.

Weeding Capability of Robotic Lawn Mower in Agriculture

Genta KANAI, Namiko YOSHINO

[Keywords: robotic lawn mower, weeding machine, weed, orchard, abandoned field]

Effect of Preheating Treatment on Guanylic Acid, Glutamic Acid and Flavor of Dried Tomato Powder

Takahiro ORIKASA, Chinami NAKAMURA, Shoji KOIDE, Uzuki MATSUSHIMA, Masumi OKADA

[Keywords: dried tomato powder, hot air drying, sensory evaluation, guanylic acid, glutamic acid]

Application of Snapper Header to Japanese Domestic Combine

Genta KANAI, Yoshiya SHINOTO, Yoshimichi YAMASHITA

[Keywords: combine, snapper header, corn, head loss, threshing loss]