Research On Small Scale Greenhouse Temperature Control Based On Improved Heuristic Algorithm And Neural Network

The use of greenhouses in modern agricultural production,which has the benefits of safety,convenience,and environmental protection while also protecting crops from the elements and other natural factors,has grown in importance as a means of producing high-demand crops,and its influence in a variety of areas,including agricultural production and economic development,is expanding every year.Since demand for greenhouses is growing annually along with the development of greenhouse-related technology.However,because greenhouse crops must be very expensive and dense,losses and effects brought on by sudden temperature fluctuations can be catastrophic.It is now absolutely important to use intelligent temperature management in greenhouses.The main research of this article is as follows:(1)A proportional integral derivative(PID)control scheme based on adversarial learning grey wolf optimization(ALGWO)and back propagation neural network(BPNN)is designed.Firstly,a new nonlinear stochastic strategy was designed for updating the convergence factor,which in turn enhanced the unpredictable behavior of gray wolves.Secondly,a random wandering Levy flight was used to update the position of alpha gray wolves to achieve better exploration ability of the alpha wolves.Then a differential evolutionary search strategy centered on alpha gray wolves was applied to beta and delta gray wolf position shifts for enhancing the king’s ability to leadership of managers in the pack;Finally,in the omega gray wolf,a decision layer gray wolf antagonistic movement strategy was used to improve the overall diversity of the gray wolf population through the antagonistic behavior of the pack manager and the wolf king.In addition,a PID control model based on BP neural network was developed and the effect of choosing different activation functions in the output layer on the control effect was evaluated.The model can adjust the PID gain parameters according to the reference values obtained from sampling,the actual output values of the controlled system and the error between them.The simulation experimental results show that the adversarial learning-based gray wolf optimization technique(ALGWO)has superior performance among eleven benchmark functions that are commonly used for testing.Meanwhile,the PID control schemes based on BPNN and ALGWO are applied to different controlled systems,and the results verify that the chosen activation function causes some accuracy impact,but the sigmoid function performs better overall.In addition,the adaptation degree of the proposed PID control schemes can all achieve better control results.(2)A PID control scheme based on an improved war strategy algorithm was proposed to control the temperature of small greenhouses.Firstly,a nonlinearly varying parameter is designed to replace the fixed-value parameter R,which better balances the exploitation and exploration of the war strategy;secondly,since the existence of only one commander and around the local optimum often leads to a decrease in the search accuracy,this situation is avoided by increasing the number of commanders;then the opposing learning method is used to place the injured soldiers;finally,the improved war strategy algorithm with PID control technology is used in the temperature control of greenhouse.The consequences of simulation studies using eleven test functions demonstrate that,when compared to previous algorithms,the upgraded warfare strategy algorithm has a greater search impact when solving various benchmark functions and has higher stability in PID parameter search.The enhanced war strategy algorithm’s PID control scheme also works better for the application of temperature control in tiny greenhouses.