До питання керування виробництвом ентомофагів

Анотація

TO QUESTION OF MANAGEMENT OF ENTOMOPHAGES

V. Lysenko, I. Chernova

The current stage of development of biotechnologies, in particular, the production of guaranteed quality entomophages for biological plant protection requires the use of computer-integrated technologies. The management systems created on this basis should ensure high quality of produced products and minimization of consumed energy, which is a prerequisite for increasing the economic efficiency of such production in general. At the same time, an important feature of production is the uncertainty in the states of the biological component of the control object, which manifests itself in  different behavior under the influence of a combination of influence factors. Solving this issue is possible by creating a hybrid network based on the use of neural networks and fuzzy logic.

The aim of the research was the development of a hybrid network that can be used to form control influences in the production of entomophages on the quality criterion on an example of the production of entomophage Habrobracon hebetor using ANFIS - the editor of MATLAB.

The object of research is the technological process of production of entomophage Habrobracon hebetor. To create a hybrid network, the results of experimental studies conducted by scientists  of the Engineering Technology Institute "Biotechnics". Methods of research - analytical, expert, neuro-fuzzy inference.

The training of the sample (experimental data) for the 100 and 200 epochs has been conducted: the error during the training for 100 epochs is 2,2831; for 200 epochs - 0,85605. The testing of the training sample has been performed - the average test error is 0,85161.

The approximation is done of the dependence of the quality of the production of the entomophage Habrobracon hebetor on the number of infected caterpillars of the host insect - Ephestia kuehniella - on the parameters of technocenosis (temperature and relative humidity of air in the entomoculture zone) using a hybrid network based on a neuro-fuzzy conclusion. The average error of approximation between the results of the experiment and the use of the hybrid network is 7.3% (not exceeding 8-10%).

The surface of the fuzzy conclusion is obtained, the analysis of which allows to confirm the optimization of microclimate parameters in the range of air temperature 26-29 ° C and relative humidity of 60-80% (the number of infected caterpillars corresponds to regulatory requirements).

Using the ANFIS - the editor of MATLAB developed a self-learning hybrid network, which, based on the experimental studies of the dependence of the quality of the production of the entomophage Habrobracon hebetor on the number of infected caterpillars of the host insect - Ephestia kuehniella - on the parameters of technocenosis (temperature and relative humidity of air in the entomoculture zone) automatically created the knowledge base and the surface of the fuzzy logic conclusion

This allows, based on the information on the dependence of the biological indicators of the quality of entomocultures on the technocenous parameters, to form strategies for managing the production of entomophages, to provide a reduction of energy expenditure for decision making regarding the quality of the entomocultures taking into account the influence of a combination of factors.

The proposed approach is an effective tool for creating intelligent control systems for the production of entomological products of assured quality.