Due to the limitation of chemical process technology and management system, the traditional treatment mode of heavy metal waste liquid pollution has many problems, such as high power consumption in the process, high wear of consuming steel, high management cost, and secondary pollution caused by complex solid waste residue in data detection. Team based on the front end of deep learning module and backend soft measurement model of collaborative structures, design and improvement of the learning algorithm combined with nested integration depth control series new front end treatment module, the successful implementation of heavy metal ions in industrial wastewater concentration of accurate prediction, and according to the predicted results and the analysis of liquid ingredients, targeted treatment after intelligent classification.