Modeling And Analysis Of Complex Chemical Industry Energy Efficiency Based On Extreme Learning Machine

The modeling and analysis of complex chemical energy efficiency need to consider the following aspects:First,the chemical process has the characteristics of complex mechanism,so the process data has high data dimension,and its measurement process usually has some error noise and external interference.Based on the above-mentioned problems,this paper proposes a method of chemical energy efficiency modeling based on Extreme learning machine(ELM),which can be used to perform parallel distribution processing and high degree of nonlinear approximation,and it is not easy to fall into local minima value,so ELM can be used without understanding the chemical process mechanism under the premise of high-dimensional chemical data with the model.The noise,interference and error in the chemical process measurement can be solved by data preprocessing using the Index Decomposition Analysis(IDA).IDA is a method to analyze energy efficiency,it can not only comprehensive statistics of the energy efficiency of each cycle,but also intuitively get the physical production indicators and energy use between the correlation,so it is widely used in energy efficiency analysis of complex chemical industry.In this paper,topology evolution based on extreme learning machine(Extreme Learning Machine,ELM)modeling method is proposed.After the feature extraction of the complex chemical industry data,using neural network topology expansion(Neuro Evolution of Augmenting Algorithm Topologies,NEAT)to determine the extreme learning machine structure,the complex chemical process is modeled by the neural network.Combining the input and output data which is calculated by the model the proposed of the corresponding plants,and the index of IDA,we can model and analyze complex chemical industry.The research contents of this paper are as follows:1?The number of hidden node of ELM usually be set depend on the experience,and the impact of different number of hidden nodes is big for the calculating results.In this paper,the NEAT method is used to evolve the structure of ELM by evolving the number of hidden nodes.The experimental results verify the effectiveness of the proposed method.2?The data of complex chemical industry has the characteristics of high dimension,large noise and redundant data.In order to carry on the analysis,this paper proposes the IDA-NEAT-ELM modeling method,by using the index decomposition method extraction in chemical process to get feature extraction,reduce the data dimension,remove the noise in the data and reduce the redundant data in the data.The neural network structure and weight are trained by the extracted decomposition index,so as to achieve the purpose of simulating complex chemical process from data driven.The method proposed in this paper is applied to the actual production data to prove the validity of the method.3?Using the modeling method proposed in this paper,combined with the index data calculated in the modeling process,the analysis of the production of the complex device is analyzed.4?A B/S based software prototype system is designed to model and analyze energy efficiency.The data of chemical industry plants are used as the research object to verify the validity and feasibility of the proposed method.