In order to effectively reduce the energy consumption, improve the efficiency of data collection in HWSNs, and prolong the lifetime of the overall network, the clustering method is one of the most effective methods in the data collection methods for HWSNs. The data collection strategy of HWSNs based on the clustering method mainly includes three stages: (1) selecting the appropriate cluster head, (2) forming between clusters, and (3) transferring data between clusters. Among them, the selection of the cluster heads in the first stage. The optimal number of cluster heads in the formation of clusters in the second stage is the core and key to the clustering data collection of HWSNs. In the stage of cluster head selection, a data collection strategy for HWSNs based on the clustering method is proposed. Sink establishes an extreme learning machine neural network model. The cluster member nodes select cluster heads based on the remaining energy of the sensor node, the number of the neighbor node, and the distance to the sink. The best cluster head node is selected through the adaptive learning of the online sequence extreme learning machine. Through comprehensive consideration of various factors to complete the clustering process, the gray wolf algorithm is used to optimize the number of clusters, balance the effect of clustering, and improve the efficiency of data collection while reducing energy consumption. An energy efficient and reliable clustering data collection strategy for HWSNs based on the online sequence extreme learning machine and the gray wolf optimization algorithm is proposed in this paper. The simulation results show that the proposed algorithm not only significantly improves the efficiency of the data collection and reduces energy consumption but also comprehensively improves the reliability of the network and prolongs the network’s lifetime.

In wireless sensor networks, each sensor node has a finite amount of energy to expend. The clustering method is an efficient way to deal with the imbalance in node energy consumption. A topology optimization technique for wireless sensor networks based on the Cauchy variation optimization crow search algorithm (CM-CSA) is suggested to address the issues of rapid energy consumption, short life cycles, and unstable topology in wireless sensor networks. At the same time, a clustering approach for wireless sensor networks based on the enhanced Cauchy mutation crow search algorithm is developed to address the issue of the crow algorithm’s sluggish convergence speed and ease of falling into the local optimum. It utilizes the Cauchy mutation to improve the population’s variety and prevent settling for the local optimum, as well as to broaden the range of variation and the capacity to carry out global searches. When the leader realizes he is being followed, the discriminative probability is introduced to improve the current person’s location update approach. According to the simulation findings, the suggested CM-CSA algorithm decreases the network’s average energy consumption by 66.7%, 50%, and 33.3% and enhances its connectivity performance by 52.9%, 37.6%, and 23.5% when compared to the PSO algorithm, AFSA method, and basic CSA algorithm.