Theory and Application of Swarm Intelligence and Machine LearningView this Special Issue
Analysis on the Way and Potential of Economic Low-Carbon Development of China Based on Genetic Algorithm
How a developing China can meet the challenges of the post-Kyoto era in the process of rapid industrialization is a hot issue in current academic research. Facing the pressure of the international community to reduce emissions and the energy and resource constraints under the development trend of the heavy chemical industry, China can only turn the pressure into a driving force and seek a low-carbon development path. This paper proposes a prediction model for China’s low-carbon economic development based on the combined model of genetic algorithm (GA) and long short-term memory neural network (LSTM). The data are encoded with one-hot, embedding is used to reduce the dimension, and the genetic algorithm is used to obtain the optimal hyperparameters of the LSTM model to improve the accuracy of the model. The results show that the model accuracy remains above 90%.
Under the influence of the global warming environment and the market environment impacted by the financial crisis, all countries in the world are developing at a high speed in order to achieve a more ideal economic level as soon as possible, and in most cases, a low-carbon economy is based on the actual situation. Development strategy emerges based on economic growth. In the process of developing a low-carbon economy, my country has adopted many effective technologies and formulated many innovative mechanisms to better promote the development of my country’s low-carbon economy and effectively realize the traditional economic model. Turning to a low-carbon economic model, after fully combining my country’s current economic development, a low-carbon economic model is created that can meet my country’s characteristics [1–5].
Affected by the pattern of economic globalization, the emission reduction behavior of some developed countries will have a certain economic impact on most developing countries. Countries with advanced technology consume a lot of energy and emit a lot of energy. In order to reduce the amount of greenhouse gas emissions, industries with high emissions such as electricity, industry, and building materials industry will start to explore their own energy-saving potential and carry out effective technology. At the same time, it will also carry out overseas market development work, with the help of foreign direct investment to improve corporate profits, and then make up for the losses caused by domestic compliance [6–8]. Affected by global economization, most developing countries, including China, have transformed into world industrial bases after implementing the agreement, and China has thus obtained much-needed development funds and high-tech. Economic growth has developed rapidly since 2005, but the economic price paid at that time cannot be ignored. Our country is still in the stage of economic development. Once we need to deal with greenhouse gas emission reduction or fulfill the obligation to limit emissions, it is likely to be trapped by investment again.
As far as the characteristics of the technology economy are concerned, the development of a low-carbon economic model that is coordinated with the climate and environment can help our country realize the transformation of the economic growth model and, at the same time, can effectively protect the domestic environment. On the whole, the essence of a low-carbon economy is high energy efficiency. As well as the structural problems of clean energy, the core content is the innovation of energy technology and the innovation of the system. Analyzing this characteristic, it is basically consistent with the measures of energy conservation currently carried out in our country. Therefore, it can be considered that the development and application of low-carbon economic energy and new technologies in the international arena will greatly improve my country’s traditional development model of high energy consumption, high-carbon emissions, and low economic benefits. The goal of effectively promoting the construction of a well-off society can be achieved. Especially in the low-carbon economy, the content of technological innovation for the utilization of traditional energy has far-reaching value for my country [9–13]. From two perspectives, first, in order to achieve the ideal emission reduction target, most developed countries will promote energy technology cooperation; that is, government departments and enterprises will increase investment in research and development and increase the speed of technology development; thereby, promoting technological innovation is considered from a technical point of view, and the measures taken by developed countries have a relatively positive impact on developing countries. Second, in order to reduce costs, developed countries will transfer emission reduction projects to developing countries. The cost of emission reduction in developed countries is relatively high, so they will reduce costs by transferring them to developing countries. Developed countries provide certain financial assistance and technical support and assist developing countries to reduce emissions, so that developing countries can also achieve good sustainable development. For China, at the current stage and in the future development, it is necessary to adopt some active and effective policies, grasp the favorable opportunities it already has, and use the cooperation with developed countries to reduce greenhouse gas emission technologies and strong financial support.
This paper proposes a prediction model for China’s low-carbon economic development based on a combination of a genetic algorithm (GA) and long short-term memory (LSTM) model [14–17]. The weather, temperature, week, and holiday features are added to the original data of time and low-carbon economy to assist in prediction, which improves the prediction accuracy. The accuracy rate of the model remains above 90%, and the accuracy is also higher than the one used in the current scheduling system. Exponential smoothing model only needs to be trained once, and then, the model data can be directly loaded for prediction without retraining, which has a good prospect of practical application.
2. Theoretical Background
2.1. Long Short-Term Memory
LSTM is improved by the recurrent neural network (RNN). RNN will be affected by short-term memory. If a sequence is long enough, it is difficult to transmit information from earlier time steps to later time steps, so RNN is not suitable for long-term memory. LSTM solves the problem of RNN short-term memory very well. LSTMs have internal mechanisms called gates, and the gate structure learns what information to save or forget during training. Subsequently, it can pass relevant information along long-chain sequences for prediction. The structure of each cell unit of LSTM is shown in Figure 1.
A cell unit of LSTM contains three gate structures: forget gate, input gate, and output gate. At the same time, it contains two states, the cell state vector and the hidden layer state. The hidden layer state is the output of the final network, and the cell state will participate in the calculation of the hidden layer state. The expression of the LSTM layer is
In (1), the symbol represents the Hadamard product. is the size of a batch of training data. is the length of the time series in the time dimension. is the feature number included by the time series. , , , and are called input gates, forget gates, cell gates, and output gates at time t, respectively. and are called cell states and hidden states at time t.
2.2. Genetic Algorithm
Genetic algorithm originated from the computer simulation research of biological systems, and it is an efficient, parallel, global search method to seek the optimal solution of the problem by imitating the evolution mechanism of the biological world. The genetic algorithm includes coding, initializing population, evaluating fitness, judging termination, condition, selection, crossover, mutation, and decoding steps, and its process is shown in Figure 2.
Due to the strong scalability of the genetic algorithm, the number of cells in each layer of LSTM, the fully connected layer, the L2 regularization weight parameter, the dropout parameter, the number of iterations, and the batch_size can be used as the parameter set, and the RMSE can be used as the fitness function. Using the genetic algorithm iterations, theoptimal hyperparameters of the LSTM model were found. The L2 regularization parameter is a penalty item in the model learning process, and the purpose is to prevent the model from overfitting. Dropout means that during the training process of a deep learning network, the neural network unit is temporarily dropped from the network according to a certain probability. The dropout parameter is the probability of dropping. batch_size is the number of samples selected for one iteration of training. RMSE is the square root of the ratio of the square of the deviation between the predicted value and the true value to the number of observations n, and it is one of the indicators to evaluate the accuracy of the model.
2.3. One-Hot Encoding
The main purpose of embedding is to reduce the dimension of sparse feature vectors and reduce the dimension by calculating the weight matrix of the embedding layer. In addition, the embedding matrix assigns a fixed-length vector representation to each categorical variable, and this length can be set by yourself.
Let the maximum length of the text be n, and there are m different kinds of texts. Each text is represented by a vector Ti (i = 1, 2, ... , m) after one-hot encoding. Then, the dimensions of Ti are all 1n. The resulting text vector is
3. The Proposed Model
3.1. Feature Screening
Feature importance evaluation is performed using a random forest; that is, the feature importance score of each tree in the random forest is calculated and averaged. Usually, the Gini index can be used as an evaluation index to measure. The feature importance score is represented by VIM, and the Gini index is represented by GI. Suppose there are m features, namely, X1, X2,..., Xm. It is necessary to calculate the Gini index score VIMj(Gini) of each feature Xj, which is the average of the node splits of the jth feature in all decision trees in the random forest. The calculation formula of the Gini index GIm of this variable in node m is
Among them, K indicates that the random forest has K categories, mk indicates the proportion of category k in node m, and mk indicates the proportion of category k in node m. Feature importance scores are obtained by random forest for temperature, week, weather, and holidays. As shown in Figure 3, it can be seen that the importance scores of weather and temperature features are both low, but considering that severe weather has a greater impact on the low-carbon economy, only the temperature feature is deleted. From an objective point of view, the reason why temperature has little effect on sales volume is actually a geographical factor; that is, the low-carbon economic development predicted by this project is located in Kunming, and the temperature change is relatively small.
3.2. Model Establishment
The original data of this project are the time series data of time and sales, and the modeling process is shown in Figure 4. (a) Feature extraction: add temperature, weather, holidays, and week features. (b) Feature screening: the feature importance score is performed on the expanded features through the random forest, and the temperature feature is deleted. (c) Text encoding: one-hot encoding is performed on the two text-like features of weather and week. Then, use the embedding method for dimensionality reduction, reduce the dimensionality of the week from the 17-dimensional vector to 11, and reduce the weather from the 18-dimensional vector to 13. (d) LSTM model prediction: put the processed data into the LSTM model for prediction. (e) Genetic algorithm parameter adjustment: input the hyperparameters of the LSTM model as the parameter set into the genetic algorithm, use the RMSE as the objective function, and obtain the maximum value through the genetic algorithm. (f) Excellent hyperparameters: result evaluation: verify the accuracy of the model.
3.3. Analysis of Results
GA-LSTM and one-time exponential smoothing were used for prediction, respectively. The results obtained are shown in Table 1. It can be seen that the one-time exponential smoothing model has the problem of prediction lag, especially in the time period before and after the Spring Festival. The carbon economy fluctuates violently, the prediction lag of the first-order exponential smoothing model is obvious, and the prediction of the first-order exponential smoothing in the second half of the test set is basically a straight line. The prediction of GA-LSTM can better predict the peaks and troughs, and it can also have a good prediction effect on data with obvious nonlinear characteristics such as low-carbon economy.
At the same time, due to the large number of LSTM hyperparameters, the genetic algorithm is used to adjust the hyperparameters of LSTM. Compared with the empirically set LSTM model, the GA-LSTM model improves the accuracy by 3%, and the RMSE decreases by 40 L/d, which makes the LSTM model prediction more accurate. From the data analysis, the RMSE of GA-LSTM is much smaller than that of one-time exponential smoothing, and the accuracy is improved by 9%. At the same time, due to the addition of holiday features, GA-LSTM is also more sensitive to the severe fluctuation of a low-carbon economy before and after holidays which is able to make effective predictions.
4. Analysis of China’s Development Paths to Realize Low-Carbon Economy
4.1. Adhere to the Overall Interests of the Country, and Make a Correct Assessment of the Obligations and Responsibilities of Greenhouse Gas Emission Reduction within My Country
In the process of evaluating my country’s carbon emissions, it is necessary to fully combine the various influencing factors that my country will involve in the process of development. At the same time, effective and accurate assessment and analysis are carried out according to the actual development of modernized trade activities and the situation of investment introduction and transfer. In particular, it must be combined with the carbon emissions that may arise in the process of exporting high-carbon products. Combining with the specific developmental stage of our country, formulate a reasonable plan, put survival and development on the level that must be considered, compare carbon emissions with some western countries to find the unique characteristics of carbon emissions, and strictly control the export process. During the export process, it is necessary to strictly abide by the carbon emission regulations of the demand side for products, and at the same time improve the situation of paying the bill for emissions. In this way, it can significantly reduce the emission pressure of various greenhouse gases that my country needs to face, refuse to pay for emissions at will, and strictly prohibit developed countries from setting traps for my country’s economic development. In addition, each country is carrying out win-win cooperation for low-carbon economic development. In the process of cooperation, my country needs to adhere to its own development path and, on this basis, carry out detailed inspections in combination with various technologies and products, and at the same time, the product cost is repeatedly checked, so that the emission trading can maintain the ideal degree of fairness and openness and effectively avoid the developing countries becoming a platform for developed countries to carry out carbon emissions in the process of economic exchanges.
4.2. Improve the Concept Content and Development Form of Backward Economic Development
If our country wants to realize the effective development of low-carbon economy to the greatest extent, what we need to do is to actively enter into the action of emission reduction, have an in-depth understanding and analysis of the low-carbon economy, and effectively take appropriate measures and means to carry out comprehensive abatement treatment interventions. First of all, as a government department, we must first formulate a relatively complete basic model that can promote the all-round development of a low-carbon economy and give macroscopic adjustment and control in combination with the environment of economic development. At the same time, it is also necessary to build a comprehensive legal foundation and implement an effective policy support system to carry out the low-carbon transformation in stages. It cannot be rushed for a while, and stable development should be combined. Combined with the efforts and characteristics of multiple parties, there may be problems in the process of transformation. The problems that occur should be properly resolved and dealt with so that the mode of economic development can be properly transformed. Secondly, strengthen the emphasis on people to ensure that people can achieve solidarity and cooperation in the process of economic development transformation on the basis of efficient cooperation. In the work, the actual effect of low-carbon emission reduction is effectively guaranteed.
4.3. Create Pilot Areas and Find Specific Ways to Develop a Low-Carbon Economy
Pilot areas can be selected based on the characteristics of different regions and different departments. For example, a low-carbon economy pilot area can be set up in the eastern and western parts of my country, and appropriate policy support can be used to attract high-end manufacturing industries and R&D industries to invest in them and timely transform or appropriately eliminate some high energy consumption. High-polluting industries enable the pilot area to gradually become a low-carbon economy demonstration area.
5. The Potential of Low-Carbon Economic Development in China
5.1. The Development of Science and Technology
Technology is one of the important factors for realizing low-carbon economic development. From a microlevel, the importance attached to science and technology by various enterprises in our country has increased significantly. These enterprises have improved their environmental protection technology level through investment in environmental protection technology and the addition of human resources. This change has laid a good material foundation for the realization of the low-carbon economic development goals of enterprises. From a macro perspective, the proportion of scientific and technological development in national policies has also increased accordingly. Taking the national science and technology plan as an example, new energy and high technology have become the main target of introduction in the plan. In addition, ecological environmental protection work and new energy application work also belong to the key content of the national science and technology plan. The national science and technology plan’s emphasis on new energy companies provides a strong guarantee for the improvement of new energy companies’ market competitiveness. In this context, various strategic measures of new energy enterprises can be well realized.
5.2. Management and Control System
In this regard, the potential of low-carbon economic development in my country mainly includes the following: first is the potential of laws and regulations. Laws and regulations have an important binding and normative role. Judging from the current laws and regulations formulation work in my country, the formulation of laws and regulations based on energy conservation and emission reduction indirectly promotes the realization of economic development goals. Second is the potential of the management system. My country’s major enterprises began to form a correct concept of energy conservation and emission reduction. Some enterprises have begun to apply the whole-process energy conservation and emission reduction management to their actual operations. Compared with other management modes, the application advantage of the whole-process management mode of energy saving and emission reduction is that its application in all links such as production and sales can effectively avoid the occurrence of resource waste. In addition, some enterprises have also formulated a sound assessment system and responsibility system for energy conservation and emission reduction. As far as the employees of these enterprises are concerned, they can obtain certain material and spiritual rewards through the reasonable development of energy conservation and emission reduction work, so the enthusiasm of these employees to participate in energy conservation and emission reduction work will increase accordingly. From another point of view, the differentiation of responsibilities and the application of punishment measures reduce the probability of employees wasting resources. Therefore, the establishment of an energy-saving and emission-reduction management system has effectively improved the sustainable development capacity of enterprises, and this change will have a positive effect on the realization of economic low-carbon development goals.
5.3. Potential for the Transformation of the Global Low-Carbon Economy
As far as the current situation is concerned, the low-carbon economy has become the mainstream development trend of the global economy. In this context, the potential for low-carbon development of my country’s economy is mainly manifested in the following aspects: first is to enhance international competitiveness. Developing international transactions and enhancing international competitiveness are one of the important contents of my country’s development strategy. From my country’s past development experience, it can be seen that in order to achieve this purpose, China has made efforts from multiple export channels. In this case, when the world is transforming into a low-carbon economy, my country’s economy will naturally take it as a development goal. The second is to enhance the international status. Compared with my country’s traditional extensive economic model, the advantage of a low-carbon economy is that it can obtain higher production quality through lower energy consumption, and the emissions in this process are relatively small. From an essential point of view, a low-carbon economy can be regarded as a high-quality activity that obtains the greatest return at the least cost. This advantage and the development goal of enhancing international status have induced the formation of my country’s economic low-carbon development potential.
5.4. Growth Potential of National Economy
The growth potential of the national economy to achieve the goal of low-carbon economic development is mainly reflected in the following aspects: first is the consumption of household appliances. The consumption process of this kind of consumer goods is based on the continuous consumption of energy, and in addition, it itself belongs to heavy chemical materials. Therefore, the consumption of household appliances can be used to reflect the low-carbon development potential of my country’s economy. As the main tool for people to meet people’s living needs, the consumption of household electrical appliances in urban residents in my country in recent years is mainly reflected in the categories of advanced and technological electrical appliances. The popularity of basic household appliances among rural residents is relatively low, so there is a certain potential for the realization of low-carbon economic development goals. Second is the car consumption. Judging from the current situation, there are relatively many cities in my country that have reached the threshold of 30 cars per thousand people. First-tier cities such as Beijing and Shenzhen exceeded this threshold by 10 times as early as 2001. These data also reflect the potential of my country’s low-carbon economic development.
A gas station sales prediction model based on the GA-LSTM combination model is proposed, and the weather, temperature, oil price, week, and holiday features are added to the original data of time plus sales to assist the prediction, and the prediction accuracy is improved. The accuracy rate of the model remains at 90% above, the accuracy is also higher than the one-time exponential smoothing model used in the current refined oil scheduling system, and the model only needs to be trained once, and then, the model data can be directly loaded for prediction without retraining, which has a good prospect of practical application. The prediction flexibility of the model is strong, and the data of the first m days can be used to predict the data of the next n days to meet the needs of different predictions. The model is not only suitable for gas stations and sales forecasts but also for other sales and forecasts that are affected by a variety of factors. Further exploration is to add attention mechanisms and mine features that potentially affect sales. The model is configured in the refined oil scheduling system of the enterprise, so that the prediction model can be applied in the enterprise system.
On the whole, facing the impact of the external environment and the negative impact of the financial crisis all over the world, my country’s economy will inevitably embark on the development path of the low-carbon economy. However, in order to effectively and comprehensively develop and implement my country’s low-carbon economic development model, it is necessary to base itself on my country’s overall national interests and conduct a comprehensive analysis of the form of economic development under the traditional model on the development path of a low-carbon economy. In the process of developing a low-carbon economy, China has the advantages of Chinese characteristics, but it is necessary to actively adopt some scientific and effective means to effectively improve the level of low-carbon economic development.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Conflicts of Interest
The authors declare that they have no conflicts of interest regarding this work.
This work was supported by a Study on the Path of Huizhou Merchants’ Family Finance Affecting Modern Corporate Governance Structure (AHSKY2019D033).
G. Zhuang, “The path and potential analysis of China's economic low-carbon development,” Pacific Science, vol. 11, pp. 79–87, 2005.View at: Google Scholar
Z. Xiao-dan, Y. Cheng, and C. Xin, “Transformation of China's economic development mode - start a low-carbon era,” Energy & Environment, no. 1, pp. 2-3, 2011.View at: Google Scholar
W. Gao, “Ways and potential of developing low-carbon economy,” Environmental Protection, no. 14, pp. 38-39, 2010.View at: Google Scholar
W. Hou, “Low-carbon mode: the transformation mode of China's sustainable economic development,” Generation economics and Trade, no. 16, p. 66, 2011.View at: Google Scholar
S. Zheng and Y. Wang, “Development strategy of petroleum industry based on low carbon economy,” SouthwestJournal of The University of Petroleum: Social Science Edition, no. 4, pp. 12–17, 2011.View at: Google Scholar
G. Li, Research on China's Low-Carbon Economy Based on Technological Progress, Nanjing University of Aeronautics and Astronautics, Nanjing China, 2011.
Z. J. Sha, Research on the Development of Energy Conservation and Emission Reduction in China under the Background of Low-Carbon Economy, Southwestern University of Finance and Economics, Chengdu, China, 2011.
L. Li, Analysis and Suggestions on the Development Potential of Low Carbon Economy in China, Shandong University of Economics, Jinan, China, 2011.
fengliang Fan, “China's economic transformation and upgrading from the perspective of low carbon,” Inner Mongolia united front theory research, vol. 05, no. 03, pp. 30–33, 2016.View at: Google Scholar
feng Shi, “Research on the development and policy of low-carbon economy in China,” Macroeconomic management, vol. 02, no. 02, pp. 70–78, 2016.View at: Google Scholar
J. Mizutani and S. Fukuda, “Issues on modal shift of freight from road to rail in Japan:Review of rail track ownership, investment and access charges after the National Railway restructuring,” Research in Transportation Business & Management, vol. 35, Article ID 100484, 2020.View at: Publisher Site | Google Scholar
A. A. Betarelli, E. P. Domingues, and G. J. D. Hewings, “Transport policy, rail freight sector and market structure: the economic effects in Brazil,” Transportation Research Part A: Policy and Practice, vol. 135, pp. 1–23, 2020.View at: Google Scholar
A. Sagheer and M. Kotb, “Time series forecasting of petroleum production using deep LSTM recurrent networks,” Neurocomputing, vol. 323, Article ID 05224, 2018.View at: Google Scholar