Directing at various problems of the traditional Q-Learning algorithm, such as heavy repetition and disequilibrium of explorations, the reinforcement-exploration strategy was used to replace the decayed ε-greedy strategy in the traditional Q-Learning algorithm, and thus a novel self-adaptive reinforcement-exploration Q-Learning (SARE-Q) algorithm was proposed. First, the concept of behavior utility trace was introduced in the proposed algorithm, and the probability for each action to be chosen was adjusted according to the behavior utility trace, so as to improve the efficiency of exploration. Second, the attenuation process of exploration factor ε was designed into two phases, where the first phase centered on the exploration and the second one transited the focus from the exploration into utilization, and the exploration rate was dynamically adjusted according to the success rate. Finally, by establishing a list of state access times, the exploration factor of the current state is adaptively adjusted according to the number of times the state is accessed. The symmetric grid map environment was established via OpenAI Gym platform to carry out the symmetrical simulation experiments on the Q-Learning algorithm, self-adaptive Q-Learning (SA-Q) algorithm and SARE-Q algorithm. The experimental results show that the proposed algorithm has obvious advantages over the first two algorithms in the average number of turning times, average inside success rate, and number of times with the shortest planned route.

The traditional lateral flow immunoassay (LFIA) detection method suffers from issues such as unstable detection results and low quantitative accuracy. In this study, we propose a novel multi-test line lateral flow immunoassay quantitative detection method using smartphone-based SAA immunoassay strips. Following the utilization of image processing techniques to extract and analyze the pigments on the immunoassay strips, quantitative analysis of the detection results was conducted. Experimental setups with controlled lighting conditions in a dark box were designed to capture samples using smartphones with different specifications for analysis. The algorithm’s sensitivity and robustness were validated by introducing noise to the samples, and the detection performance on immunoassay strips using different algorithms was determined. The experimental results demonstrate that the proposed lateral flow immunoassay quantitative detection method based on image processing techniques achieves an accuracy rate of 94.23% on 260 samples, which is comparable to the traditional methods but with higher stability and lower algorithm complexity.

Considering the low indoor positioning accuracy and poor positioning stability of traditional machine-learning algorithms, an indoor-fingerprint-positioning algorithm based on weighted k-nearest neighbors (WKNN) and extreme gradient boosting (XGBoost) was proposed in this study. Firstly, the outliers in the dataset of established fingerprints were removed by Gaussian filtering to enhance the data reliability. Secondly, the sample set was divided into a training set and a test set, followed by modeling using the XGBoost algorithm with the received signal strength data at each access point (AP) in the training set as the feature, and the coordinates as the label. Meanwhile, such parameters as the learning rate in the XGBoost algorithm were dynamically adjusted via the genetic algorithm (GA), and the optimal value was searched based on a fitness function. Then, the nearest neighbor set searched by the WKNN algorithm was introduced into the XGBoost model, and the final predicted coordinates were acquired after weighted fusion. As indicated in the experimental results, the average positioning error of the proposed algorithm is 1.22 m, which is 20.26–45.58% lower than that of traditional indoor positioning algorithms. In addition, the cumulative distribution function (CDF) curve can converge faster, reflecting better positioning performance.

A photovoltaic power supply with a simple structure and high tracking efficiency is needed in self-powered, wireless sensor networks. First, a maximum power point tracking (MPPT) algorithm, including the load current maximization-perturbation and observation (LCM-P&O) methods, with a fixed step size, is proposed by integrating the traditional load current maximization (LCM) method and perturbation and observation (P&O) method. By sampling the changes of load current and photovoltaic cell input current once the disturbance is applied, the pulse width modulation (PWM) regulation mode, i.e., increasing or reducing, can be determined in the next process. Then, the above algorithm is improved by using the variable step size strategy. By comparing the difference between the absolute value of the observed current value and the theoretical current value at the maximum power point of the photovoltaic cell with the set threshold value, the variable step size for perturbation is determined. MATLAB simulation results show that the LCM-P&O method, with a variable step size, has faster convergence speed and higher tracking accuracy. Finally, the two MPPT algorithms are tested and analyzed under constant voltage source input and indoor fluorescent lamp illumination through an actual circuit, respectively. The experimental results show that the LCM-P&O method with variable step size has a higher tracking efficiency, about 90%–92%, and has higher stability and lower power consumption.

The colorimetric method, due to its rapid and low-cost characteristics, demonstrates a wide range of application prospects in on-site water quality testing. Current research on colorimetric detection using deep learning algorithms predominantly focuses on single-target classification. To address this limitation, we propose a multi-task water quality colorimetric detection method based on YOLOv8n, leveraging deep learning techniques to achieve a fully automated process of “image input and result output”. Initially, we constructed a dataset that encompasses colorimetric sensor data under varying lighting conditions to enhance model generalization. Subsequently, to effectively improve detection accuracy while reducing model parameters and computational load, we implemented several improvements to the deep learning algorithm, including the MGFF (Multi-Scale Grouped Feature Fusion) module, the LSKA-SPPF (Large Separable Kernel Attention-Spatial Pyramid Pooling-Fast) module, and the GNDCDH (Group Norm Detail Convolution Detection Head). Experimental results demonstrate that the optimized deep learning algorithm excels in precision (96.4%), recall (96.2%), and mAP50 (98.3), significantly outperforming other mainstream models. Furthermore, compared to YOLOv8n, the parameter count and computational load were reduced by 25.8% and 25.6%, respectively. Additionally, precision improved by 2.8%, recall increased by 3.5%, mAP50 enhanced by 2%, and mAP95 rose by 1.9%. These results affirm the substantial potential of our proposed method for rapid on-site water quality detection, offering new technological insights for future water quality monitoring.