Research On Energy Efficiency Analysis And Optimization Method For Ethylene Production Process

Energy is the driving force and material basis of national economic development,and its safe supply and efficient use are of great significance to the construction of modern cities,and its development and utilization are directly related to the high and low living standards of people.As the world’s largest energy producer and consumer,China is in the middle and late stages of industrialization,and its carbon emissions continue to rise compared to other countries,so the time to achieve the "double carbon" target is getting shorter and the task is getting more and more arduous.As an important chemical production process in China,the complex ethylene production process has a large production scale,long process and complex structure,and many factors affect its energy consumption,conversion and utilization.Therefore,how to improve the energy efficiency of ethylene production process and reduce energy consumption is one of the effective measures to achieve the goal of "double carbon".This paper mainly focuses on the statistical,panel and operational data reliability analysis of complex ethylene production process,studies the energy use and consumption of complex ethylene production process,combines the characteristics of complex ethylene process structure,state,process and taskrelated energy use mechanisms,and proposes a multi-level(industrytechnology-apparatus)energy efficiency analysis and intelligent optimization method for complex ethylene production process from the multi-level structure of process system.The method is used to analyze,predict and optimize the energy efficiency of complex chemical processes.At the same time,we studied the ethylene input-output data processing and optimization problems,including feature dimensionality reduction,elimination of redundancy and spatial correlation among data,and processing of dynamic time-series data,and constructed a neural network-based ethylene production prediction model and a multi-objective optimization model for ethylene cracking furnace based on membrane optimization algorithm to verify the effectiveness and universality of the proposed theory and method.The proposed theory and method are validated to provide theoretical basis and methodological guidance for energy saving and operation optimization of ethylene production process.The main research contents of this paper are as follows.(1)To address the problem that the industry-wide production energy efficiency evaluation in the ethylene production process is single and cannot find its effective quantitative improvement direction,an energy efficiency analysis and evaluation method based on Data Envelopment Analysis Cross Model(DEACM)fused with principal element feature extraction is proposed.Firstly,to address the problem of redundant information in the industry-wide chemical process data,the Interpretive Structural Modeling(ISM)method is used to analyze the relationship between different variables in combination with partial correlation coefficients.Then,the factors in each layer are combined by Analytic Hierarchy Process(AHP)to reduce the number of dimensions of factors affecting the energy efficiency of the industry-wide chemical process.Finally,the merged main factors are input into DEACM,so as to enhance the differentiation of industry-wide ethylene production process production energy efficiency analysis and evaluation,and then guide the energy efficiency optimization direction of non-effective ethylene production process production configuration based on the optimal allocation of industry-wide ethylene production process production inputs and generation.(2)To address the problem of low accuracy of industry-wide ethylene production unit production forecasting,a t-distributed Stochastic Neighbor Embedding(t SNE)and Extreme Learning Machine(ELM)model(t SNE-ELM)method for chemical process production prediction.First,the t SNE method is introduced to reduce the dimensionality of the raw data of chemical process units in the whole industry.Then,the processing results are input into the ELM network to build the chemical process production prediction model.Finally,the proposed t SNE-ELM prediction model is compared with Back Propagation(BP)neural network,Radial Basis Function(RBF)neural network and ELM method to verify the effectiveness of the proposed t SNE-ELM based prediction model,and through the prediction analysis of ethylene production in the actual ethylene production process,the proposed t SNE-ELM prediction model was compared with the ELM method to verify the effectiveness of the proposed t SNE-ELM prediction model.(3)In order to accurately predict the key variables in the production process within the ethylene production process plant,an ethylene production process yield prediction combining the Attention-Long Short-Term Memory(Attention-LSTM)network with the attention mechanism is proposed for the characteristics of dynamic and real-time data in the actual operation of the ethylene production process plant.model.First,the data are divided and preprocessed according to the time-series data of ethylene production process units.Then,the prediction modeling analysis of ethylene production capacity based on Attention-LSTM is performed,and the effectiveness of the proposed Attention-LSTM prediction model is verified by comparing with BP,RBF and LSTM prediction models.Finally,the results of ethylene production capacity prediction are used as a reference to optimize the feedstock input structure for the data sets with low production efficiency,and realize the energy efficiency prediction and optimization of ethylene production units under specific technologies.(4)For the purpose of ethylene cracking process with complex production process,diverse production objectives and expectation of optimized production,an evolutionary membrane algorithm(EMACM)containing competitive communication strategy is proposed,combining the flexibility and scalability of membrane optimization methods and the ability to enhance the search capability of optimization algorithms.competitive communication).The membrane algorithm with competitive communication can be effectively adjusted to optimize the algorithm,and the optimized solution set is obtained by the CCEMA(Constrained competitive exchange membrane algorithm)with constraints to solve the practical problem of multi-objective optimization of ethylene crackers.The improved algorithm is also applied to the ethylene cracker,and the optimization results show the effectiveness and rationality of the proposed multi-objective intelligent optimization algorithm for ethylene cracker based on membrane algorithm by optimizing the operating variables that have important influence on the ethylene production process cracker.