Research On Optimized Learning Approach Based On Chaotic Harris Hawks Algorithm

As a raw material commodity for mass production and consumption of agriculture,soybean price fluctuations have a significant impact on a country’s economic production activities and directly affect the national economy and people’s livelihood.At the same time,soybeans are also an important type of agricultural products in bulk commodities.In the international trade market,countries that have a large amount of bulk commodity trading resources have certain advantages in economic development.Therefore,changes in soybean prices have always attracted attention.As a major importer of soybeans in the world,our country’s soybean trade has always been affected by the market,politics,supply and demand,and the overall soybean industry has become increasingly risky.After entering the WTO,our country imported a large number of soybeans to meet domestic soybean consumption needs,and at the same time,the international price decline trend was obvious,which caused the price of domestic soybeans to be suppressed and increased pressure on the soybean planting industry.In order to avoid the risk of soybean spot price fluctuations,soybean futures provide traders with important reference and help methods such as price discovery and hedging.Therefore,the forecast of soybean futures prices is directly related to the asset allocation of investors,and is helpful to the production and consumption decision-making of related agricultural products companies,making the formulation and implementation of national trade and monetary policies more effective.Due to the complex influencing factors and mechanisms of the soybean futures market,soybean futures prices have significant nonlinearity and high volatility.This thesis takes soybean futures prices as the research object,and proposes a new optimized learning approach based on decomposition and optimization algorithms.The full thesis is mainly divided into the following five parts:The first part mainly introduces the research background and significance of this article,a review of relevant domestic and foreign literature and the main research content of the article.In the second part,the basic theories of related statistical prediction models and Variational Mode Decomposition are explained.The third part,based on the traditional Harris Hawks Optimization algorithm,discusses in detail the construction process of the improved Chaotic Harris Hawks Optimization algorithm,and applies the improved algorithm to the optimization of the initial weight and threshold of the Extreme Learning Machine.At the same time,the optimized learning framework and prediction steps based on decomposition and optimization algorithms are given.In the fourth part,the optimized learning approach based on decomposition and optimization algorithms proposed in this thesis and the other six commodity futures price prediction models are used to predict the futures prices of soybean and crude oil,and the statistical errors and statistical tests are compared to verify the proposed method.The fifth part is conclusions and prospects.It summarizes the research results of this thesis,points out the deficiencies of existing work and prospects for future work and research directions.The main contributions of this thesis are:(1)A new optimized learning method based on decomposition and optimization algorithms is proposed to improve the level prediction accuracy and direction prediction accuracy of commodity futures prices.(2)Aiming at the shortcomings of low convergence accuracy of Harris Hawks Optimization,an improved Chaotic Harris Hawks Optimization algorithm based on chaotic mapping is proposed.The empirical results show that the improved Chaotic Harris Hawks Optimization algorithm has a significant improvement in optimization accuracy and search ability.(3)The Chaotic Harris Hawks Optimization is used to optimize the initial weights and thresholds of the Extreme Learning Machine model,avoiding the randomness and blindness of the model parameter setting,and improving the prediction accuracy of the model.