Evaluation and analysis of irrigation water use efficiency based on an extreme learning machine model optimized by the spider monkey optimization algorithm

Abstract

To promote the efficient use of water resources and the sustainable development of the agricultural economy, the main evaluation index that links economic production and the effective use of water resources, irrigation water use efficiency (IWUE), is scientifically evaluated and analyzed. To obtain accurate evaluation results, an evaluation model for IWUE based on an extreme learning machine (ELM) model optimized by the spider monkey optimization (SMO) algorithm (i.e., the SMO-ELM model) was proposed. The model was applied to the evaluation and analysis of IWUE in 24 large and medium irrigation districts in Heilongjiang Province. On this basis, the main driving factors of IWUE differentiation in the irrigation districts were identified, and adaptive control strategies were explored. In addition, to thoroughly test the performance of the constructed SMO-ELM evaluation model, a comparative analysis was performed for a BP model, an ELM model, an ELM model with crow search algorithm optimization (CSA-ELM) and the SMO-ELM model. The results show that there are regional differences in the IWUE of the irrigation districts and that the IWUE of the districts in the eastern part of the study area is better than that of the districts in the central and western parts. The irrigation districts with relatively high evaluation grades (IV to V) are distributed mostly in the eastern part of the study area, and most of the irrigation districts with low evaluation grades (I to II) are located near the main branch of the Songhuajiang River. Differences in indexes, such as the soak field quota and proportion of rice sowing area, are the most important factors related to variations in IWUE grades among the irrigation districts, and targeted control strategies such as planting structure optimization are proposed. The rationality and superiority of the SMO-ELM model are verified by comparing the coefficient of determination (R2), mean square error (MSE), mean absolute percentage error (MAPE), and accuracy of classification (accuracy) of the BP, ELM, CSA-ELM and SMO-ELM models. The R2, MSE, MAPE, and accuracy of the SMO-ELM model are 0.9999, 4.33 × 10−10, 0.00039, and 99.8%, respectively, which are much better than those of the other models. The results of complexity analysis verify the stability of the IWUE evaluation results for the irrigation districts. This article provides a new model for the evaluation of IWUE and scientific and technological support for ensuring the safety of water resources and maintaining the sustainable green development of agriculture.