An elevator is a typical flexible lifting machine. In order to monitor the vibration of elevator structure, the vibration characteristics of an elevator with a traction ratio of 1:1 has been tested experimentally. Sensors were arranged on the platform frame, car roof, and hoist rope to test the vibrations of the elevator in both ascending and descending conditions. The transverse, longitudinal, and coupled transverse-longitudinal vibrations were compared and analyzed. Further, the short-time Fourier transform (STFT) method was used to examine the power spectral density (PSD) of the test results, and the main frequency distribution and influencing factors of the vibration of elevator components were investigated. The results revealed that the transverse and longitudinal vibrations of the platform frame were low-frequency vibrations, which was attributed to the elastic interaction between the platform frame and the car frame. The form and amplitude of longitudinal vibration of the car frame were basically consistent with those of the platform frame, but PSD of the transverse vibration had an obvious peak in the high-frequency region. The transverse and longitudinal vibration frequencies of the hoist rope were higher. Furthermore, the peak PSD value of transverse vibration of the hoist rope was 421 times larger than that of the car frame, so a small disturbance at the end of the rope could lead to a huge disturbance in the center of the rope. Overall, this study provides useful insights on designing an elevator monitoring sensor and relevant data processing.

Using fine-grained data analysis, non-invasive load monitoring (NILM) can reveal the detail of electricity customers’ habits, which is helpful in the improvement of refined management and better user experience. However, the possibility of electricity customers’ privacy leak is also gradually increasing, and the security of NILM data has become a priority problem to be solved. To protect the privacy disclosure of NILM data, this paper analyzes the NILM privacy leak problems and ways in which information leak occurs faced by NILM data. On the basis of the comprehensive survey of cryptographic algorithms to choose the most appropriate data security method for NILM, a hybrid cryptography scheme was proposed to protect the data security. In the scheme, symmetric algorithm AES (Advanced Encryption Standard) was used to encrypt data for high efficiency, and asymmetric algorithm RSA (Rivest-Shamir-Adleman) was used to encrypt AES key for identity authentication. The classical algorithm HMAC-SHA1 (Hash Message Authentication Codes-Secure Hash Algorithm 1) was further developed to guarantee the integrity of data. By transplanting the algorithm into STM32 MCU (STMicroelectronics 32 bit Micro Controller Unit) for performance test and using Visual studio 2017 + QT tools to develop the test interface, one optimal operation mode was selected for the scheme. At the same time, the effectiveness of the scheme was verified, and the scheme computing cost depended on the efficiency of encryption and decryption, or signature and verification of the RSA algorithm.

To improve the accuracy of braking distance measurement and reduce the impact of human factors on brake risk assessment of escalators, a non-contact braking distance measuring device based on infrared ranging principle is designed. Futher, a quantifiable risk assessment method of brake based on failure mode and effects analysis (FMEA) is propose in this paper. Characteristics and trends of distance changes during emergency braking process of the escalator are theoretically analyzed. To extract, process, and recognize the distance change signals, and finally to realize accurate measurement of the braking distance, a high-speed infrared distance measurement technology is used. Influencing factors of the escalator brake performance are studied. The risk analysis method based on FMEA is used to analyze its failure mode, failure mechanism, and consequences of the failure. A risk index assessment system is constructed and quantified. The test results show that the measurement error of the device designed is 1.6%, which is 1/6 of the traditional measurement method. The repeated measurement error of the device is 1.4%, which is 1/7 of the traditional measurement method. Application results show that the quantifiable risk assessment method quantifies and assigns the risk level according to the weight, which effectively avoids the inconsistent results caused by human bias. The evaluation results are more scientific and reliable.

In order to study the relationship between the braking distance and the load of escalator and realize the prediction of the rated load braking distance with a little load, the method of combining theoretical analysis and experimental research is used. First, the dynamic characteristics of the escalator during emergency braking are analyzed, and the prediction model of the braking distance of the escalator under different loads is derived based on the law of conservation of energy. Furthermore, the influence coefficients under different loads were determined through experimental studies, the model was revised, and the concept of equivalent no-load kinetic energy (ENKE) was proposed. The research shows that the braking distance of the escalator increases nonlinearly with the increase in load. When the no-load braking distance and the 25% rated load braking distance change greatly, the braking distance increases faster as the load increases; the escalators with large brake force have a small ENKE and are easy to stop. Otherwise, it is difficult to stop. The test results show that the comparison between the predicted value of the prediction model and the measured value has a maximum error of 2.7%, and the maximum error at rated load is only 2.0%, which fully meets the needs of engineering measurement. And the prediction method reduces test costs, enhances test security, and improves test coverage.

To address the low accuracy of defect detection in infrared images caused by missing details and blurred features, this paper proposes a joint method combining SwinIR and an improved YOLOv9. SwinIR performs super-resolution reconstruction and denoising using long-range dependency modeling and local window attention, enhancing high-frequency detail restoration. The refined images are then processed by YOLOv9 for object detection. Experiments on the Switchgear_detection.v1 dataset show that the method achieves PSNR of 38.27 dB and SSIM of 0.9653 at ×2 magnification, and 32.54 dB and 0.9194 at ×4, outperforming contrast algorithms. The approach improves both image quality and detection accuracy, offering a high-precision, low-latency solution suitable for mobile infrared monitoring in intelligent distribution rooms.