Abstract:

Throughout history, the quality of the air has always been regarded as one of the most significant environmental problems by the general public and society. When it comes to the examination of future air quality trends from a macro viewpoint, the use of machine learning algorithms for the prediction of the Air Quality Index (AQI) is beneficial. It is difficult to produce a satisfactory prediction result when using a single machine learning model to forecast air quality in a traditional manner. This is because there are many different Air Quality Index (AQI) fluctuation patterns. A genetic algorithm-based improved extreme learning machine (GA-KELM) prediction approach is upgraded in order to increase its ability to successfully solve this challenge. In the beginning, a kernel approach is presented in order to generate the kernel matrix, which is then used to substitute the output matrix of the hidden layer. A genetic algorithm is then used to optimize the number of hidden nodes and layers of the kernel limit learning machine in order to address the problem that is caused by the conventional limit learning machine. This problem is caused by the fact that the number of hidden nodes, as well as the random generation of thresholds and weights, lead to a decrease in the network's capacity for learning. The definition of the fitness function is accomplished by the use of the thresholds, the weights, and the root mean square error. The last step in the process involves using the least squares approach in order to determine the output weights of the model. via an iterative optimization process, genetic algorithms are able to locate the best solution in the search space and steadily enhance the performance of the model. This is accomplished via finding the optimum solution. Based on the collected basic data of long-term air quality forecast at a monitoring point in a city in China, the optimized kernel extreme learning machine is applied to predict air quality (SO2, NO2, PM10, CO, O3, PM2.5 concentration, and AQI). Comparative experiments based on CMAQ (Community Multiscale Air Quality), SVM (Support Vector Machines), and DBN-BP (Deep Belief Networks with Boundaries) are also conducted in order to verify the ability of GA-KELM to make accurate predictions. There is back-propagation. The findings indicate that the suggested model is capable of training more quickly and producing more accurate predictions.

Description:

.