Wireless sensor network (WSN) nodes are devices with limited power, and rational utilization of node energy and prolonging the network lifetime are the main objectives of the WSN’s routing protocol. However, irrational considerations of heterogeneity of node energy will lead to an energy imbalance between nodes in heterogeneous WSNs (HWSNs). Therefore, in this paper, a routing protocol for HWSNs based on the modified grey wolf optimizer (HMGWO) is proposed. First, the protocol selects the appropriate initial clusters by defining different fitness functions for heterogeneous energy nodes; the nodes’ fitness values are then calculated and treated as initial weights in the GWO. At the same time, the weights are dynamically updated according to the distance between the wolves and their prey and coefficient vectors to improve the GWO’s optimization ability and ensure the selection of the optimal cluster heads (CHs). The experimental results indicate that the network lifecycle of the HMGWO protocol improves by 55.7%, 31.9%, 46.3%, and 27.0%, respectively, compared with the stable election protocol (SEP), distributed energy-efficient clustering algorithm (DEEC), modified SEP (M-SEP), and fitness-value-based improved GWO (FIGWO) protocols. In terms of the power consumption and network throughput, the HMGWO is also superior to other protocols.

Real-time dynamic monitoring of orchard grape leaf diseases can greatly improve the efficiency of disease control and is of great significance to the healthy and stable development of the grape industry. Traditional manual disease-monitoring methods are inefficient, labor-intensive, and ineffective. Therefore, an efficient method is urgently needed for real-time dynamic monitoring of orchard grape diseases. The classical deep learning network can achieve high accuracy in recognizing grape leaf diseases; however, the large amount of model parameters requires huge computing resources, and it is difficult to deploy to actual application scenarios. To solve the above problems, a cross-channel interactive attention mechanism-based lightweight model (ECA-SNet) is proposed. First, based on 6,867 collected images of five common leaf diseases of measles, black rot, downy mildew, leaf blight, powdery mildew, and healthy leaves, image augmentation techniques are used to construct the training, validation, and test set. Then, with ShuffleNet-v2 as the backbone, an efficient channel attention strategy is introduced to strengthen the ability of the model for extracting fine-grained lesion features. Ultimately, the efficient lightweight model ECA-SNet is obtained by further simplifying the network layer structure. The model parameters amount of ECA-SNet 0.5× is only 24.6% of ShuffleNet-v2 1.0×, but the recognition accuracy is increased by 3.66 percentage points to 98.86%, and FLOPs are only 37.4 M, which means the performance is significantly better than other commonly used lightweight methods. Although the similarity of fine-grained features of different diseases image is relatively high, the average F1-score of the proposed lightweight model can still reach 0.988, which means the model has strong stability and anti-interference ability. The results show that the lightweight attention mechanism model proposed in this paper can efficiently use image fine-grained information to diagnose orchard grape leaf diseases at a low computing cost.

Dynamic cooperation clustering (DCC) becomes a main enabler for cell-free massive MIMO systems since it can improve the energy efficiency and reduce the complexity of signal processing significantly. However, DCC formation for all served users simultaneously is a very complicated mixed binary nonlinear programming problem, a single agent has limited capability to search the optimal schemes. In this paper, we propose a whale swarm reinforcement learning (WSRL) based DCC method. Exploiting multiple searching agent imitated by a group of whales, whale swarm optimization (WOA) algorithm searches the optimal DCC scheme simultaneously and learn the searching experience from each other. Moreover, the reinforcement learning is integrated to select the most efficient hunting action for each whale, which can accelerate the convergence and avoid the local trap. Simulation results demonstrate that the proposed method has better searching ability and higher convergence speed than the existing works.

To investigate the feasibility of using near-infrared (NIR) spectral technology to detect the soluble solids content (SSC) of Malus micromalus Makino, rapid and non-destructive prediction models of SSC were studied using least-square support vector regression (LS-SVR), partial least squares regression (PLSR), and the error back propagation artificial neural network (BP-ANN). First, 110 samples of NIR diffuse reflectance spectra in the wavelength range of 400.41–1083.89 nm were obtained, and then were divided into the calibration set and prediction set by sample set partitioning based on the joint x–y distance (SPXY) algorithm. Second, we compared the prediction performance of the PLSR model after preprocessing by nine spectral preprocessing methods, and applied data dimension reduction methods (random frog, the successive projections algorithm (SPA), and principal component analysis) for variable selection. Finally, the effect of applying full spectrum and characteristic spectrum modeling on SSC prediction accuracy was compared and analyzed. The comparison studies confirmed that the optimal fusion model of SPA-LS-SVR had the best performance (RC = 0.9629, RP = 0.9029, RMSEC = 0.199, RMSEP = 0.271). The experimental results could provide a reference for future development of the internal component analysis system for Malus micromalus Makino based on NIR spectroscopy and its classification system using SSC as the classification standard.