Neural network evaluation of basil (Ocimum basilicum L.) response to organic fertilizer additives under controlled chemo and light culture conditions

The demand for natural medicinal plant raw materials (MRM) in the world is growing from year to year. Therefore, industrial phytotrons that grow MRM in protected soil conditions have begun to develop. Modern phytotrons use automated technologies that support a stable microclimate, optimal lighting mode, water and microelement nutrition, as well as substrates that replace the soil. The Automated technologies for growing plants use BigData information obtained from various sensors located in the phytotron. One of the effective tools for statistical analysis of phytotronic BigData is trained neural networks. They are able to detect and visualize the informal dependence of the volume and quality of plant products on the physicochemical characteristics of nutrient and soil substrates, as well as on the lighting regime. In this work, for the first time, an original neural network algorithm for calculating the fractal index of profiles of chemical elements in plants is used to identify the relationship between their physiological and biochemical indicators and growing conditions. The aim of the work was to study the influence of special fertilizing agrotechnology on the physiological and biochemical characteristics of plants of the basil of the green-leaved variety Lemon (Enza Zaden, Holland) and the purple-lavender variety Rosie (Enza Zaden, Holland). Plants were grown under controlled chemo and light culture conditions. Depleted neutralized peat with perlite additives was used as a soil substrate. Plants were grown under controlled chemo and light culture conditions. Depleted neutralized peat with perlite additives was used as a soil substrate. The specificity of the fertilizer technology was the use of organic additives: humic and fulic acids isolated from sapropel raw materials, and an alkaline extract from dried and sieved excrement of Hermetia illucens fly larvae. Plants in the test versions were sprayed on the leaves with solutions of the test extracts once a week. Plants in the control version were watered with filtered water-wire water. At the end of the test, chlorophyll was measured in the leaves (SPAD 502 Plus, Minolta Camera Co, Ltd, Japan). The elemental composition of the biomass was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) (Agilent 5900. Agilent Technologies, США). Measured basil chemical element profiles were processed using the EuclidNN neural network. A distinctive feature of the EuclidNN neural network is the ability to search for the many-to-one conversion of the plant's chemical element profile to the CSI (Cognitive Salience Index), which characterizes the fractality of the chemical element profile. The computational algorithm of the neural network is not initially known. The algorithm is searched in the EuclidNN neural network training mode. According to the results of the study, it was found that only fulvous fertilizer additive was more effective on the green leaf variety, and fulvous fertilizer additive and zoocompost extract additive were more effective on purple-lavender. The humic fertilizer additive on both basil varieties demonstrated poor efficacy. There was a high correlation with the biomass and height of the plant in these variants (r = 0.85-0.96, p function show_eabstract() { $('#eabstract1').hide(); $('#eabstract2').show(); $('#eabstract_expand').hide(); }