In order to detect the abnormal working conditions such as overpressure and leakage, that may occur in pipelines and installations in the process of natural gas regional production, the current industrial control and alarm systems cannot accurately reflect the real state of the equipment, and the single-parameter early warning has a higher rate of error judgement, which is insufficient in practicality. A collaborative prediction and warning method for process parameters related to upstream and downstream stations in a natural gas production area was tested. Aiming at the characteristics of natural gas region with many stations, complex production process and diverse monitoring data, firstly, the parameters of each station were downgraded to extract the key process parameters of each station. Then, the key parameters are evaluated and grouped by correlation, and a multivariate nonlinear lasso regression prediction model was established with the highly correlated parameters in the same group as the independent variables. At the same time, a long and short-term memory prediction model was established for the key parameters, and a comparison analysis of the prediction results was performed to determine the dynamic prediction and early warning of natural gas production. Comparative analysis of the prediction results of the two models was used to determine the dynamic thresholds for coordinated early warning of regional production. The results show that the method can not only effectively reduce the misjudgment of single-value anomalies, but also locate the anomalous stations and points, which is of high practical value.