Prediction of dissolved oxygen mass concentration in aquaculture based on SARIMA-VMD-LSSVM
In order to make full use of the data characteristics of dissolved oxygen mass concentration and further im-prove the accuracy of dissolved oxygen mass concentration prediction in aquaculture,a dissolved oxygen mass concentration prediction model combining"linear and nonlinear"and"decomposition-prediction-integration"was proposed.Firstly,the seasonal auto regressive integrated moving average(SARIMA)model was used to linearly fit the dissolved oxygen mass concentration time series,and the residual sequence was decomposed using variational mode decomposition(VMD).Then,each residual component was substituted into the least square support vector machine(LSSVM)model optimized by the im-proved gray wolf algorithm(IGWO)to obtain the prediction results of the nonlinear component.Finally,the linear and non-linear prediction results were integrated to obtain the final dissolved oxygen mass concentration prediction value.Experimen-tal results showed that compared with SARIMA,LSSVM,and VMD-LSSVM models,the prediction accuracy of SARIMA-VMD-LSSVM model was significantly improved.The root mean square error(RMSE)was 0.078 7,and the mean absolute percentage error(MAPE)was 0.022 6,in-dicating that the combined model could effectively extract the multi-scale features of the time series of dissolved oxy-gen mass concentration,and achieve more accurate predic-tion.
aquaculturedissolved oxygenvariational mode decompositioncombinatorial forecasting methodsimproved gray wolf algorithm
万方数据
维普
