Chemical Industrial Production Safety Status Forcast Method

In recent years,safety accidents in chemical industries are happening frequently,caused incalculable casualties and economic losses,so the Nation are paying more and more attention to the chemical industrial production status monitoring and early warning technology.Current forecast domestic research on chemical safety in production state is relatively slow,mainly predict a single device or variable based on single hardware technology.This paper analyzes multivariate statistical methods in the field of chemical production process monitoring algorithm,combines partial least squares algorithm(PLS model)with Auto-Regressive and moving average model,proposes a process prediction algorithm based on multivariate statistical algorithm to achieve chemical production safety status prediction,and this method is applied to the ammonia refrigeration systems.This paper focuses on the practical application of the prediction algorithm,works are expending as follows.1.Investigate the related production technology and production status of the application research unit,do preliminary data processing on data obtained.After that,do data cleansing,data preprocessing according to the actual situation,prepare training data and testing data for multivariable statistical models.2.Based on multivariate statistical algorithms,this paper builds two kinds of models--principal component analysis model(PCA)and partial least squares model(PLS),which are widely used.We also do process monitoring based on the Q statistics and T2 statistics,plot contribution figure if the status is abnormal and analyze the cause.Compare the two models on aspects of the process of monitoring results,contribution figure,prediction accuracy etc.,After the actual verification,PLS model is superior to PCA model.3.This paper introduce the idea of the time series prediction algorithms to chemical production safety status monitor model,proposes ARMA-PLS prediction algorithm by combining ARMA model with PLS model,achieve the prediction of chemical production security status.ARMA-PLS model,proposed through above research work,are more suitable for the chemical safety state prediction,has certain theoretical significance for realizing chemical production safety prediction and early warning technology.