Mass data processing method for ocean industry and ecology association DSGE analysis decision

Abstract

The invention provides a mass data processing method for marine industry and ecology association DSGE analysis decision, and belongs to the technical field of mass data processing of a DSGE model.The mass data processing method includes the steps that multi-source heterogeneous data streams are collected, time-space alignment preprocessing is executed, threshold co-integration inspection is executed on preprocessed data to recognize nonlinear threshold points, a state transition matrix is established, and the state transition matrix is established; inputting the data into a heterogeneous computing scheduler for resource allocation judgment and executing grid division, completing parameter calibration by adopting a multi-target decomposition framework in combination with a particle swarm algorithm and a simulated annealing algorithm, establishing a sliding window mechanism, and executing incremental updating by adopting a recursive least square algorithm; and low-rank approximate decomposition is performed on the high-dimensional weight matrix, and the optimized low-rank sub-matrix is used as the weight parameter of the DSGE model for decision analysis, so that the technical problem that the parameter calibration and decision optimization of the DSGE model are difficult to efficiently support under the condition of real-time processing of mass heterogeneous data is solved.

Images