Considering that there are many monitoring points in process industrial production system and aiming at the correlation among monitoring points, a method is proposed to comprehensively evaluate the state of multiple variables in the system, based on detrended cross correlation analysis (DCCA) method and network structure entropy method (DCCA-NSEn). We use the DCCA method to calculate the correlation between the multivariate variables, and construct a weighted network model which reflects the multivariable coupling relationship. Time window sliding in the monitoring sequence to obtain the dynamic evolution model of system′s coupling relationship network. The NSEn method is used to calculate the network structure entropy of the coupled network model in each time period. Finally the state of the complex electromechanical system is evaluated according to the network structure′s entropy changes over time. This paper presents the real production data of a compressor unit to verify the DCCA-NSEn method, then the multivariate analysis of the same group of production data is conducted by the coupling detrended fluctuation analysis (CDFA) method. The results of the two methods are compared. The results show that compared with the DCCA method, this method has the advantages of multivariate simultaneous monitoring and evaluation. Compared with the CDFA method, which is also a multivariate analysis method, the DCCA-NSEn method has the advantages of stable evaluation effect and obvious effect on the abnormal state detection of the system.