Method and device for quickly identifying factors for analyzing target pollutant source

Abstract

The invention provides a method for quickly identifying factors for analyzing a target pollutant source. The method comprises the following steps: collecting published documents related to the targetpollutant source analysis in contents; extracting graphs and data related to sources corresponding to target pollutant source analysis factors from the published documents, and constructing a trainingset and a test set for training learning to obtain a preliminary training model; verifying the training model by adopting a cross validation method based on an artificial neural network to obtain a final factor recognition model; analyzing a to-be-detected sample, and calculating unrecognized factors; calculating the unrecognized factors by using a factor recognition model, and analyzing a pollutant source corresponding to the factors. According to the method for quickly identifying PMF factors, a source corresponding to each factor is automatically identified by an artificial neural networkdeep learning algorithm, so that the factor identification speed and the factor identification accuracy are greatly improved. Meanwhile, the subjectivity of manual selection is avoided.

Description

technical field [0001] The invention relates to the technical field of pollution source analysis, in particular to a method and device for rapid identification of PMF factors and source analysis. Background technique [0002] Source apportionment is an important condition for formulating scientific and reasonable relevant policies and regulations, and has become one of the core contents of environmental pollution prevention and control in various parts of our country. [0003] The positive matrix factorization model (Positive Matrix Factorization, PMF), PMF is a receptor source analysis model, based on the principle of factor analysis, but relying too much on the emission source component spectrum, the change of the contribution value of the source in different time series can be obtained, It is widely used in the research of atmospheric particulate matter source analysis, atmospheric VOCs, water pollution source analysis, and soil pollution source analysis. However, due to...

AI Technical Summary

Problems solved by technology

However, due to the great uncertainty in the currently used receptor model PMF method, that is, after the n factors calculated by the PMF, how to identify the sources corresponding to these factors, such as coal-burning sources, motor vehicle sources, and dust sources The determination of factors such as factors depends entirely on the experience and subjective judgment of the PMF operator; secondly, the PMF cannot determine the final number of factors, and it is also through the operator's repeated calculation of the combination of multiple factor numbers, relying on experience and subjective judgment to determine the number of factors. Determine how many factors should be selected in the end

The operator’s experience comes from reading previous literature and the general cognition of the chemical components contained in the pollution source, so the source apportionment results obtained by different operators vary from person to person, and there is no uniform standard and recognized correct result

This leads to great uncertainty in source apportionment results, which affects the effectiveness of emission reduction and governance decisions for users of final source apportionment results

Images