摘要：Recently, energy‐related CO2 emissions are considered as one of the most crucial issues and are promptly augmented due to further urbanization. In this paper, in order to model and calculate yearly CO2 emission, an artificial neural network is used. For the first time, the IWO‐SVM method has been applied in modeling energy‐related CO2 emissions. In this regard, consumption of different energy sources such as renewable energy, natural gas, coal, and oil, and GDP of the G8 countries in various years (from 1990 to 2016) are regarded as input in the present study. For the aim of evaluating the exact ability of the SVM and SVM‐IWO models, the performance of these models in three different modes is carried out on the basis of the number of data in the test and train sections. For this purpose, implementations are split into three categories ( a = 80% of the data for the train section and 20% for the test section; b = 70% of the data for the train section and 30% for the test section; and c = 60% of the data for the train section and 40% for the test section). Furthermore, five scenarios were selected on the basis of the number of input parameters and input parameters for achieving the best model. As indicated in the results in all scenarios, the correlation of the model with the hybrid invasive weed algorithm based on SVM is more favorable than that in the support vector machine model, due to better training of the SVM‐IWO model than the SVM model. Moreover, the technological orientations of the G8 countries to mitigate CO2 emissions are determined through patent analysis. While the patents have essential information, investigating the published patent by a country could be helpful for determination of technological orientations. Hence, all published patents by these countries are extracted and deeply analyzed. In the next step, to find out main technological approaches, all patents and their intents have been studied. Eventually, the technological life cycles and trends of each main technology are drawn.
