Research On Interface Technology Of Swarm Intelligent Building Electrical Equipment

The traditional centralized system of building uses a layered architecture,and each subsystem forms a structure,which cause some difficulties in network debugging,upgrading,and interconnection of electromechanical equipment.As a result,these above-mentioned problems hinder the realization of cross-system functions seriously.Compared with the traditional centralized architecture of building,swarm intelligence is a new type of operation and management system of building.Its distributed architecture eliminates the central monitoring host and makes the entire system flatter.As an important part of the building,the interconnection of electromechanical equipment is the key to realize intelligent control of building.In this paper,the electrical equipment of building is upgraded by embedded in Computing Process Node(CPN).The results are that the external interface form of the electrical equipment is standardized and an improved load prediction algorithm is embedded in the electrical equipment,which upgrades traditional device to the swarm intelligent device.Regarding the standardized description of electrical equipment in swarm intelligent building,the compilation method of electrical equipment's standard information model is researched in this paper.From the perspective of control requirements,electrical equipment is divided into seven categories,and the distribution box is selected as a typical representative to establish its standardized information model.Then the interface characteristic of the distribution box and CPN is researched,which provides a theoretical basis for the information exchange between the distribution box and CPN.In order to analyze the energy consumption level of the equipment,it is necessary to integrate the load prediction function in the electrical equipment.Recurrent Neural Network(RNN),Long Short-Term Memory(LSTM)and Attention Long Short-Term Memory(Attention-LSTM)are studied separately in the paper,and the PyCharm editor is used to model the neural network above.The result shows that Attention-LSTM has higher accuracy than the other two networks and verifies the effectiveness of the proposed algorithm.In addition,by analyzing the design process of hardware and software,a protocol conversion device based on the STM32F407 chip is developed in this paper to realize the interface connection between the distribution box and the CPN.Then the way of deployment of neural networks in the STM32 chip is given.Finally,the information exchange is realized between the distribution box and CPN based on the standard information model and Attention-LSTM load prediction model.