Research On Adaptive Home Control System Based On Improved Neural Networks

Today,the information technology industry is growing faster than other industries.It presents us with a new form of industry-the Internet of Things.It can be seen that the development of the Internet of Things is becoming more and more diversified,and its problem-solving solutions is becoming more mature.It is time to the intelligent products to appear in our daily life.With the continuous improvement of technical standards,the performance of smart accessories has been greatly improved,and the manufacturing cost of intelligent devices has been getting lower and lower.This provides a strong driving force to integrate intelligent devices into life and changes traditional lifestyles.It can be said that smart home have more and more opportunities for development,but the challenges are not to be underestimated.Faced with an endless stream of intelligent control technology,how to find the optimal solution has become a hot research topic for many scholars.So the core issue is improving the intelligence of smart home.The main work and contribution of this thesis include.1 ?This research background is provided.Firstly,the application status of equipment development,platform construction and adaptive control technologies are highlighted.Secondly,it focused on the analysis of the user's needs and found a feasible research results.2?Detailed analysis of the smart home adaptive control system principles and features.For the purpose of self-adaptive optimization of the indoor environment,analyzed the influencing factors of the indoor environment in detail,then built a prediction model of indoor environment comfort using neural network.Finally,a joint control strategy was proposed based on the prediction results.An entire smart home solution was finally formed.3?The smart home interior environment is very complicated.There is a loss of data collected for the hardware system in this environment.In this paper,K-means is used to cluster and normalize the data so that the data set used for prediction has higher quality and accuracy.4?Aiming at the limitations and disadvantages of using a single BP neural network to build a predictive model,this is a problem that the accuracy of the predictive model is low,it caused by the initial weight.The prediction result easily falls into a local minimum.Through research deeply,it has found a solution to improve this prediction model.That is,the use of genetic algorithms to find the optimal initial weights,and combined with BP neural network model prediction.This method has achieved good prediction results in the actual simulation,improved the prediction accuracy.This thesis takes the smart home control system as the main line,and analyzes the composition of smart home system according to the progressive relationship.The purpose is to continuously explore and enrich the best way of smart home control systems and optimize a control system design scheme with reference value.