Visible-near-infrared spectrum (Vis-NIR) spectroscopy technology is one of the most important methods for non-destructive and rapid detection of soil total nitrogen (STN) content. In order to find a practical way to build STN content prediction model, three conventional machine learning methods and one deep learning approach are investigated and their predictive performances are compared and analyzed by using a public dataset called LUCAS Soil (19,019 samples). The three conventional machine learning methods include ordinary least square estimation (OLSE), random forest (RF), and extreme learning machine (ELM), while for the deep learning method, three different structures of convolutional neural network (CNN) incorporated Inception module are constructed and investigated. In order to clarify effectiveness of different pre-treatments on predicting STN content, the three conventional machine learning methods are combined with four pre-processing approaches (including baseline correction, smoothing, dimensional reduction, and feature selection) are investigated, compared, and analyzed. The results indicate that the baseline-corrected and smoothed ELM model reaches practical precision (coefficient of determination (R²) = 0.89, root mean square error of prediction (RMSEP) = 1.60 g/kg, and residual prediction deviation (RPD) = 2.34). While among three different structured CNN models, the one with more 1 × 1 convolutions preforms better (R² = 0.93; RMSEP = 0.95 g/kg; and RPD = 3.85 in optimal case). In addition, in order to evaluate the influence of data set characteristics on the model, the LUCAS data set was divided into different data subsets according to dataset size, organic carbon (OC) content and countries, and the results show that the deep learning method is more effective and practical than conventional machine learning methods and, on the premise of enough data samples, it can be used to build a robust STN content prediction model with high accuracy for the same type of soil with similar agricultural treatment.

To improve prediction accuracy and provide sufficient time to control decision-making, a decomposition-based multi-step forecasting model for rabbit house environmental variables is proposed. Traditional forecasting methods for rabbit house environmental parameters perform poorly because the coupling relationship between sequences is ignored. Using the STL algorithm, the proposed model first decomposes the non-stationary time series into trend, seasonal, and residual components and then predicts separately based on the characteristics of each component. LSTM and Informer are used to predict the trend and residual components, respectively. The aforementioned two predicted values are added together with the seasonal component to obtain the final predicted value. The most important environmental variables in a rabbit house are temperature, humidity, and carbon dioxide concentration. The experimental results show that the encoder and decoder input sequence lengths in the Informer model have a significant impact on the model’s performance. The rabbit house environment’s multivariate correlation time series can be effectively predicted in a multi-input and single-output mode. The temperature and humidity prediction improved significantly, but the carbon dioxide concentration did not. Because of the effective extraction of the coupling relationship among the correlated time series, the proposed model can perfectly perform multivariate multi-step prediction of non-stationary time series.

Agricultural robots are flexible to obtain ambient information across large areas of farmland. However, it needs to face two major challenges: data compression and filtering noise. To address these challenges, an encoder for ambient data compression, named Tiny-Encoder, was presented to compress and filter raw ambient information, which can be applied to agricultural robots. Tiny-Encoder is based on the operation of convolutions and pooling, and it has a small number of layers and filters. With the aim of evaluating the performance of Tiny-Encoder, different three types of ambient information (including temperature, humidity, and light) were selected to show the performance of compressing raw data and filtering noise. In the task of compressing raw data, Tiny-Encoder obtained higher accuracy (less than the maximum error of sensors ±0.5°C or ±3.5% RH) and more appropriate size (the largest size is 205 KB) than the other two auto-encoders based convolutional operations with different compressed features (including 20, 60, and 200 features). As for filtering noise, Tiny-Encoder has comparable performance with three conventional filtering approaches (including median filtering, Gaussian filtering, and Savitzky-Golay filtering). With large kernel size (i.e., 5), Tiny-Encoder has the best performance among these four filtering approaches: the coefficients of variation with the large kernel (i.e., 5) were 8.6189% (temperature), 10.2684% (humidity), 57.3576% (light), respectively. Overall, Tiny-Encoder can be used for ambient information compression applied to microcontrollers in agricultural information acquisition robots. Keywords: agricultural information, robot, ambient information, data compression, embedded machine learning methods DOI: 10.25165/j.ijabe.20221504.6911 Citation: Wang Y T, Li M Z, Ji R H, Wang M J, Zhang Y, Zheng L H. Novel encoder for ambient data compression applied to microcontrollers in agricultural robots. Int J Agric & Biol Eng, 2022; 15(4): 197–204.

The health and productivity of rabbits are highly sensitive to the environmental conditions within the rabbit house, particularly to fluctuations and deviations in temperature, relative humidity, and carbon dioxide (CO₂) concentration. However, owing to the thermal inertia and residual evaporation effects inherent in ventilation and cooling systems, environmental changes often exhibit delayed responses, rendering real-time control inadequate. Accurate prediction of key environmental parameters is indispensable for formulating effective environmental control strategies, as it enables consideration of their future dynamics and thereby enhances the rationality of regulation in rabbit farming. Existing prediction models often exhibit unsatisfactory accuracy and weak generalization, which restricts the incorporation of prediction into effective environmental control strategies. To address these limitations, summer indoor and outdoor environmental data were collected from rabbit houses in Nanping, Fujian; Jiyuan, Henan; and Qingyang, Gansu, China—three climatically distinct regions—forming three datasets. Based on these datasets, a multi-parameter time-series prediction model, Patch and Cross-Attention Enhanced Transformer for Rabbit House Prediction (PatchCrossFormer-RHP), is introduced, integrating patching and attention mechanisms. The model partitions the sequences of rabbit house temperature, relative humidity, and CO₂ concentration into patches and incorporates auxiliary parameters, such as indoor air velocity and outdoor temperature and humidity, to enhance feature representation. Furthermore, it applies cross-attention with differentiated encoding to disentangle multi-parameter relationships and improve predictive performance. This study used the Fujian dataset as the primary benchmark. On this dataset, PatchCrossFormer-RHP achieved root mean square error (RMSE) values of 0.290 °C, 1.554%, and 38.837 ppm for rabbit house temperature, humidity, and CO₂ concentration, respectively, with corresponding R² values of 0.963, 0.956, and 0.838, consistently outperforming RNN, GRU, and LSTM. Transfer experiments with single- and multi-source pretraining followed by fine-tuning on Fujian demonstrated that strong cross-regional generalization can be achieved with only limited target-domain data.

Based on the staged continuous tabu search (SCTS) algorithm, a modified staged continuous tabu search (MSCTS) algorithm is proposed in this paper to improve the convergence, speed and robustness of tabu search (TS) algorithm. The improvements focus on the selection method of the neighborhoods in MSCTS algorithm. The generation of neighborhoods is guided by the multidimensional normal distribution function. In multidimensional normal distribution function, the mean value is the current optimal solution and the standard deviation is produced by the difference vector of the objective function at the current optimal solution. The range setting of neighborhood is different at different stage. 9 typical functions are used to test the performance of MSCTS and SCTS algorithm respectively. There are 5 indexes to evaluate the performance of both algorithms. The tests results show that MSCTS algorithm is good at dealing with the multivariate optimization problems. The calculated optimum solution for multi-variable function by MSCTS algorithm is about 5 to 17 times as near to the theoretical optimum solution as that of SCTS algorithm. As to the same test function, the calculation speed of MSCTS algorithm is about 3 to 17 times as many as that of SCTS algorithm. At the same time, the application of MSCTS algorithm is more extensive.

Pavement cracks are the main manifestations of the asphalt pavement distress for high-grade highway. Pavement cracks will not only affect the road appearance and the comfort of driving, but also easily lead to structural damage to pavement, hence shorten the life of road. The type of cracks varies according to the cause of distress. To some extent, the type of cracks shows the extent of pavement damage. The identification of the type of cracks plays an important role in exploring the causes of distress, detecting the extent of pavement damage, and finding a way to prevent the distress. In this paper, based on support vector machine, an identification algorithm of detecting the horizontal, longitudinal and net cracks of asphalt pavement is proposed. According to the geometry of different cracks, we use the number of pixels, the aspect ratio, the level projection, and the vertical projection of the connected area of the cracks, to design a classifier based on the support vector machine; and to identify the type of cracks. The experimental results show that the rates of correctly identifying different types of cracks by using the identification algorithm based on the support vector machine are as follows: 96.3% for the horizontal cracks, 97.2% for the longitudinal cracks, and 94.3% for the net cracks.