The existing tool fault diagnosis system has the problems of huge system, high cost, low precision and so on. It is urgent to develop a high precision and low cost tool fault diagnosis system. A tool fault diagnosis system which based on bionic strain sensors for CNC ( Computerized Numerical Control ) machine tools is proposed. Firstly, the bionic flexible crack array vibration sensitive component which has high-precision and inexpensive encapsulated into a rigid bionic strain sensor, making it suitable for collecting machine tool vibration signals. Then time-frequency features are extracted from the tool vibration signals and tool fault diagnosis model is built by support vector machine algorithm. Through the diagnosis of offline faults and online faults in the real-time machine-processing environment, the results show that the proposed tool fault diagnosis system has an accuracy of greater than 88% in machine-processing fault diagnosis. And the cost of system is reduced. This is a brand-new attempt to apply high-sensitivity and low-cost bionic flexible sensitive component to industrial fault diagnosis.

In order to drive OPA(Optical Phased Array) to realize beam scanning of all solid state lidar, two drive circuits for OPA phase control are designed and implemented, including current source drive and voltage source drive. The upper computer software at the PC ( Phase Control ) terminal is used to control the microcontroller connected to the current source or voltage source, and the current or voltage at each phase control terminal of OPA is adjusted to change the main maximum position of OPA far-field beam, so as to achieve the purpose of beam deflection and beam scanning. The driving principle of multi-channel, high power, high precision, wide range current source and voltage source is analyzed. Finally, the error and linearity test and OPA
phase control experiment are carried out. The experimental results prove the feasibility and correctness of the drive circuit. The proposed two OPA phase control circuits are simple in principle, strong in stability, convenient and practical, and have certain application value in high precision OPA phase control.

According to VMD(Variational Mode Decomposition) in processing the signal, the selection of mode number K and parameter α will affect the signal decomposition results, an optimized VMD(Variational Mode Decomposition) algorithm improving PSO(Particle Swarm Optimization) (PSO-VMD) was proposed. Based on the analysis of signal, high frequency noise is reconstructed by singular value denoising. Experiments on simulation signal and pipeline leakage signal, by selecting multiple K and combined with proposed the correlation method to select mode and reconstruct signal, compare SNR ( Signal to Noise Ratio ), correlation CC (Coefficient), SAE(Square Absolute Error), MSE(Mean Square Error) of reconstructed signal of different K and original signal, proved the reliability of optimizing VMD by improved PSO. The proposed method is compared with PSO-VMD combined with HD-SVD(Haussteff Distance and Singular Value), MI-SVD(Mutual Information and Singular Values), CC-WT(Correlation Coefficient and Wavelet Transform) and other methods.The proposed denoising method has the best effect.

The nodes of wireless sensor networks for oil and gas pipelines are linearly distributed according to the pipeline direction. It is unable to replace the batteries of the energy exhausted nodes in a large scale in time.Effective energy management has been a key factor restricting the performance of the oil and gas Internet of things. Network end-to-end delay determines the response time of network to pipeline accidents, so it is also a key parameter. The routing algorithm proposed in the paper uses energy acquisition technology to extend the service life of the network. The end-to-end delay of the network can be effectively reduced by selecting the node which is far away from the sender as the forwarding node. Simulation results show that the algorithm can effectively consider network lifetime and network delay, and improve network performance compared to other opportunistic routing protocols and protocols using energy harvesting technology.

Signal leakage detection of oil and gas pipelines is easily affected by noise, so de-noising becomes the key point. In order to improve the de-noising effect of oil and gas pipeline signals, a combined de-noising method based on savitzky-Golay smoothing filter, variational mode decomposition and frequency domain singular value de-noising method is proposed. Firstly, savitzky-Golay smoothing filter is used to reduce the noise of the leakage
signal in the time domain to remove the sharp pulse, high-frequency components and other noises and to improve the signal-to-noise ratio of the input signal. By calculating the Intrinsic Mode components (IMF: Intrinsic Mode Function) and the Manhattan distance between signals, distinguishing the signal components and noise, the noise component frequency-domain SVD (Singular Value De-noising), finally the filter components and signal are constructed. Simulation and practical experiments show that compared with single VMD method, VMD- wavelet transform and SG-VMD-SVD method, the signal to noise ratio is relatively higher, which verifies the superiority of de-noising effect and the feasibility of de-noising oil and gas pipeline leakage signal.

In order to improve the sensitivity of liquid refractive index measurement and make the measurement process more convenient, U-shaped POF(Plastic Optical Fiber) probes for RI(Refractive Index) sensing with high sensitivity is prepared by wet etching. The principle of the multi-mode POF macro-bending RI sensor is analyzed. The light field transmitted in the POF is approximated by the ray method, and the expression of mode loss caused by macro-bending in the multi-mode POF is obtained. The cladding and part of the core are removed through a mixed solution of acetone and methanol. The surface morphology of the optical fiber after corrosion with different etchant concentrations is observed, and the best etchant concentration is 80% with acetone. The corroded optical fiber is produced to macro-bending structure by thermal setting method, and the sensor performance is optimized by changing the fiber core diameter and bending radius. The sensitivity of 618% / RIU is obtained in the large index range of 1. 30 to 1. 45 at wavelength of 655 nm. It is proved that the sensitivity of the sensor can be improved through the macro-bending processing, and the U-shaped probes make the
measurement more convenient, which can be widely used in refractive index sensing.

To obtain the current attitude of the satellite in attitude control of CubeSat, a digital sun sensor for CubeSat is designed. Through the linear CCD(Charge Coupled Device) and the corresponding slit-type optical path design, the double-angle measurement of the point light source relative to the normal line of the installation plane is achieved, and the design index of the field of view angle of 40° and the theoretical measurement accuracy of 0. 01° is achieved. The sun sensor uses commercial off-the-shelf products and technologies (COTS: Commercial Off-The-Shelf), and has the characteristics of small size, low power consumption, simple design, high stability, and easy integration. As a COTS resource that satisfies the geometric dimensions of the CubeSat circuit board, it has potential application value in the future development of the CubeSat and plays a vital role in the further research of other CubeSats.

Grassland resources are the firm foundation for development of the country's animal husbandry.However, unreasonable development and utilization over the years has caused nearly 90% of the available grassland to experience varying degrees of degradation. Therefore, accurate and timely monitoring of grassland growth and mastering the quality parameters that reflect the growth of grassland are important for grassland management. In recent years, with the continuous development of remote sensing technology, many high-resolution satellites have added red-edge spectrum bands. Because it reflects more clearly on vegetation affected by plant diseases and insect pests, and is less affected by background, it is widely used in vegetation quality

research. Based on the B4 red edge band in Jilin No. 1 remote sensing image, the significant expression characteristics of vegetation are analyzed and studied, and the GQI (Grassland Quality Index) is constructed.Compared with the single vegetation index, this index can more comprehensively measure the advantages and disadvantages of different indicators of grassland quality. The vegetation index with high correlation with grassland quality is calculated. The principal component analysis is carried out to obtain the scoring formula of GQI, and the grassland quality index results are calculated. The results show that the 80% of grassland in Xianghai Nature Reserve is of medium quality and the range of grassland with good growth status accounts for

about 15% , which is more consistent with the actual situation.

In order to improve the accuracy and efficiency of leakage detection for long-distance pipeline, the modified fuzzy C-means algorithm is applied. Particle swarm optimization algorithm is introduced to optimize the troditional fuzzy C-means algorithm, which is used to represent the gradient descent so as to improve the efficiency and accuracy of fuzzy C-means algorithm. Then the proposed fuzzy C-means algorithm is used to analyze the same group of pipeline leakage experimental data compared with troditional fuzzy C-means algorithm and 3-layer BP (Back Propagation) neural network. The result proves that the proposed fuzzy C-means algorithm has a better property than the other two algorithms, so it is feasible to apply the PSO-based Fuzzy C-Means model in pipeline leakage detection.

SAC(Soft Actor Critic) has the advantages of good robustness, strong exploration ability, and good agent generalization ability. It also has the disadvantage of lower training speed and unstable training process, to solve this problem, we propose the PER(Prioritized Experience Replay)-SAC algorithm by introducing priority experience sampling into the SAC algorithm. Thus the network prioritizes training samples with large error in estimation function and poor strategy performance, which improves the stability and convergence speed of the Agent training process. The experimental results show that the PER-SAC algorithm is better than SAC algorithm in training speed and stability.

For quantitative analysis of the permafrost stability analysis of subsidence monitoring network, to reduce the construction cost of measuring points, in the northeast region, the actual project data analysis is performed. The results show that during December to February, as the temperature gradually reduced to less than 0 ℃ , frozen soil depth, settlement monitoring points increases gradually, February and march, frozen soil depth to achieve the minimum, monitoring the cumulative settlement up to 1 cm. Temperature has a direct impact on the frozen soil depth, and has a great impact on the formation of frozen soil depth. When the temperature drops, the frozen soil depth deepens and the monitoring point is lifted. With the increase of
temperature, the frozen soil depth decreases, and the monitoring point starts to decline and finally tends to be stable.

Ensemble learning is an effective method to classify multivariate time series. However, ensemble learning requires higher performance of the base classifier, and the combination of base classifier algorithms has a greater impact on the classification effect. This paper proposes a multivariate D-S(Dempster/ Shafer) evidence weighted ensemble classification method based on shapelets. First, learning the base classifier Shapelets on the univariate time series, the classification accuracy of the base classifier is determined as its weight in the multi-classifier. Shapelets are sub-sequences of time series. There is no dependency between different variable Shapelets, and the classification accuracy of a single Shapelets is high, and a “good but different” base classifier can be obtained. Then, a weighted probability assignment algorithm is proposed to increase the weight of the base classifier with high classification accuracy and reduce the weight of the base classifier with low classification accuracy. Two combination strategies are added to eliminate evidence conflicts and improve efficiency. Compared to some state-of-the-art algorithms on the standard dataset, our algorithm can obtain a better classification result.

It is generally lack of semantics, relatively scattered and system imperfect in terms of the demand of geological knowledge. In order to solve the organization of geological knowledge standards, based on the domestic and foreign standards within the geological domain and the study of knowledge graph, and on the basis of the knowledge of stratigraphy and petroleum geology, two standardized models of geological knowledge map had been constructed. By taking stratum and petroleum geology as application background, knowledge mapping technology is used to verify them afterwards. It turns out that the proposed standardization can provide standard support for the standardization of geological knowledge organization.

The laboratory safety in universities is the basic guarantee for the continuous development of school education and the growth of students. At present, there are some problems in the safety management of laboratories in colleges and universities, such as the inflexible way of education, the limitation of time and space, and the inconvenience of supervision. In order to improve the safety level of laboratories, we propose to develop a laboratory safety education and examination system based on wechat applet, which can effectively assist the safety management of laboratories. we start with the system architecture and system function construction, and analyzes the importance and necessity of the six functions of the system including knowledge learning, online practice, online examination, consultation questions and answers, wrong question review and user login. Taking the typical laboratory safety education examination content of domestic universities as an example, the paper expounds the construction of system content and knowledge question base, database design connection and the realization process of each module of the system in detail. The system works well after going online, and has
strong operability in practical applications, which greatly improves the enthusiasm and learning efficiency of the experimenters. And it effectively enhances the safety awareness of the experimenters, and has certain innovation and better practical value.

In order to solve the problems of the safety and reliability of the redundant design of the high-speed EMU network system, this research takes the factors that need to be considered in the redundant design of the “Fuxing” EMU network control system as a blueprint, and proposes the high-speed EMU network redundancy design principles, measures and methods of the control system in hardware and software. The batch application of EMUs verifies that the redundant design of the network control system is effectivein ensuring the safety, reliability and availability of the EMUs.