In the process industries,local linearization of a nonlinear process,which is essentially modeling of the process dynamics at a specific operating point with a simplified linear model,is commonly used for design of model-based controllers,even though many controlled processes have evident nonlinearity.Switches of operating conditions are common and would result in changes of process dynamics of the controlled process and therefore the nonlinearity,namely,divergence of process dynamics at different operating points,should be investigated,which would be useful for further controller design and control loop performance assessment.In this paper,nonlinearity quantification using the technique of sum-of-norms regularization is studied.By including a penalty term that is the sum of the l1 norm of the model parameter change,the main objective of the regression is to model the nonlinear process with approximately fewest linear models.By examining variations of model parameters,changes in process dynamics can be detected.A major advantage of this approach over existing methods is that more accurate change detection time can be guaranteed due to the sparse property.A nonlinearity index is also proposed to quantify the process nonlinearity.A simulated continuous stirred tank heater(CSTH)and an industrial case are studied to demonstrate the proposed method.