Research on an Intelligent Identification Method for Carbon Monitoring Big Data Anomaly Based on Improved Support Vector Machine

Conventional carbon monitoring big data anomaly identification and definition methods are mostly target-based division, and the scope of identification is limited, resulting in an increase in the error rate. To this end, a study on an intelligent identification method for carbon monitoring big data anomalies based on improved support vector machines is proposed. According to the current measurement needs, data preprocessing and abnormal feature extraction are first performed, and the limit values are broken through a two-way definition method to complete the fuzzy definition of failure outliers and outlier outliers. On this basis, an improved support vector machine carbon monitoring big data anomaly intelligent identification model is constructed, and a dynamic continuous monitoring method is used to realize anomaly intelligent identification processing. The test results show that: for five randomly selected test groups, compared with the deep belief network data anomaly identification method and the improved M-robust regression data online anomaly identification method, the improved support vector abnormal data identification method designed this time finally obtained. The misidentification rate is relatively low and can be controlled below 15%. This shows that with the assistance of improved support vectors, the designed anomaly identification method in carbon monitoring big data is more targeted and stable, and can perform better in different background environments, the overall efficiency and accuracy of this recognition form are also higher.