A given model of yield forecasting using an artificial neural network connects the wheat crop with the amount of productive moisture in the soil, soil fertility, weather, and factors in the presence of pests, diseases, and weeds. The difficulty of creating a yield forecast system is in the correct choice of predictors that have the greatest impact on yield. To build the model, moisture in the 100 cm layer of the soil, the content of nitrogen, phosphorus, humus, and soil acidity in the soil were used as input parameters. The amount of precipitation over 4 months, the average air temperature for the same period, as well as the presence of diseases, pests, and weeds were also taken into consideration. Data on 13 districts of the North Kazakhstan region in the period from 2008 to 2017 were used. The output parameter was the yield of spring wheat over the same time period. The relative importance of input variables in relation to the output variable was used to determine the weight values of input variables. An artificial neural network of error backpropagation was used as a method. The advantage of this method is that the quality of the forecast increases with a large amount of training data, as well as the ability to model nonlinear relationships between different data sources. After training the artificial neural network and obtaining predictive data, good results were achieved for predicting wheat yields (p=0.52, mean absolute error in percentage (MAPE)=12.02 %, root mean square error (RMSE)=3.368). Thus, it is assumed that the developed model for forecasting wheat yields based on data can be easily adapted for other crops and places and will allow the adoption of the right strategies to ensure food security

No abstract

This work reports a study into the possibility of using the GoogleNet neural network in the optoelectronic channel of the Data Fusion system. The search for the most accurate algorithms for detecting and recognizing unmanned aerial vehicles (UAVs) in Data Fusion systems has been carried out. The data processing scheme was selected (merging SVF state vectors and merging MF measurements), as well as the sensors and recognition models on each channel of the system. The Data Fusion model based on the Kalman Filter was chosen, integrating radar and optoelectronic channels. Mini-radars LPI-FMCW were used as a radar channel. Evaluation of the effectiveness of the selected Data Fusion channel model in UAV detection is based on the recognition accuracy. The main study is aimed at determining the possibility of using the GoogleNet neural network in the optoelectronic channel for UAV recognition under conditions of different range classes. The neural network for the recognition of drones was developed using transfer training technology. For training, validation, and testing of the GoogleNet neural network, a database has been built, and a special application has been developed in the MATLAB environment. The capabilities of the developed neural network were studied for 5 variants of the distance to the object. The detection objects were the Inspire 2, DJI Phantom 4 Pro, DJI F450, DU 1911 UAVs, not included in the training database. The UAV recognition accuracy by the neural network was 98.13 % at a distance of up to 5 m, 94.65 % at a distance of up to 20 m, 92.47 % at a distance of up to 50 m, 90.28 % at a distance of up to 100 m, and 88.76 % at a distance of up to 200 m. The average speed of UAV recognition by this method was 0.81 s.

No abstract

No abstract