34results about How to "Realize fault identification" patented technology

The invention discloses a method and device for online monitoring and fault identification of a mechanical state of a high-voltage circuit breaker, and relates to the technical field of electrical equipment testing. The method for online monitoring and fault identification of a mechanical state of a high-voltage circuit breaker uses four different sensors to synchronously monitor various parameters of an operating mechanism of the high-voltage circuit breaker during operation, and performs time-domain and frequency-domain analysis on the information obtained by the sensors in different mannersto form a feature hybrid discriminant matrix, thereby realizing on-line monitoring and fault identification of the state of the high-voltage circuit breaker, solving the problem that it is difficultto accurately determine the failure of the high-voltage circuit breaker during the operator's regular inspection only due to perceptual knowledge and experience, improving the inspection efficiency, and saving resources such as manpower and materials.

The invention discloses a railway wagon bogie side frame fracture fault image recognition method, and belongs to the technical field of railway wagon bogie safety. The invention aims to solve the problem of poor reliability due to the fact that the side frame fracture detection of the existing railway wagon bogie is carried out in a manual mode. The method comprises the steps of collecting an original gray image of a truck bogie side frame in operation; determining a side frame area of each grayscale image, preprocessing the side frame areas to obtain side frame area sample images, forming a sample image set from all the side frame area sample images, configuring mark information for each side frame area sample image to form a mark file, and forming a sample data set based on the sample image set and the mark file; training the convolutional neural network inception v2 and the convolutional neural network Faster rcnn, and obtaining a trained inception v2 model and a trained Faster rcnnmodel; and processing the to-be-detected image by using the trained inceptionv2 model and the Faster rcnn model to obtain a corresponding side frame state prediction result so as to realize fault identification. The method is used for bogie side frame fracture identification.

The invention relates to a high-pressure diaphragm pump check valve fault diagnosis method based on vibration signal analysis, and belongs to the field of mechanical fault diagnosis and signal processing. The high-pressure diaphragm pump check valve fault diagnosis method based on the vibration signal analysis comprises the following steps that firstly, VMD decomposition is carried out on a vibration signal of a check valve of a high-pressure diaphragm pump, and the number K of decomposition modes is determined through a center frequency to obtain K IMF components with physical significance; the MPE of the IMF components is then calculated to form a multi-scale complexity measure feature vector; and finally, a high-dimensional feature matrix is input into a classifier established by a support vector machine optimized based on a genetic algorithm to identify a working state of the check valve of the high-pressure diaphragm pump. According to the high-pressure diaphragm pump check valvefault diagnosis method based on the vibration signal analysis, the vibration signal of the check valve is denoised and decomposed into the IMF components without mode-mixing through a VMD algorithm, the multi-scale arrangement entropy of each IMF component is calculated to collect fault characteristic information distributed on multiple scales, the correct rate of fault identification of the checkvalve is improved, and higher practicability and engineering significance are achieved.

A device based on network big data storing, collecting and analyzing is disclosed. The device includes a host computer of the device and an installed operating system installed in the host computer, awireless network AP big data collection management software, a wireless network AP fault prediction and diagnosis software, a wireless network AP resource recovery and release software, a wireless network AP user location service software and a wireless network security management software. Also disclosed are a collection and storage method based on the device, a wireless network intelligent operation and maintenance method, a wireless network AP resource recovery and release method and a wireless network security judgment method. The invention has the advantages of strong self-determinationperformance, high stability, extremely high adaptability, safety and expansibility, can meet the needs of schools, government agencies, shopping malls, tourist attractions, improves the utilization oforiginal resources, and improves the level of smart city construction.

The invention provides a power distribution network weak feature fault identification method based on transfer learning, and relates to the technical field of power grid detection. The identificationmethod comprises the following steps: firstly, establishing a 20kV neutral point ungrounded AC power distribution network model, setting two data acquisition points at an outgoing line, and constructing a training sample set and a transfer learning sample set; and training a sparse auto-encoder by using the training sample set to realize high accuracy of fault identification, and finally carryingout transfer learning of the network model by using a small number of transfer sample sets so that the accuracy of the algorithm model can reach 98% under a new topological structure. According to theinvention, in power distribution networks of different topology types, high-resistance fault type identification and fault line selection can be realized, and interference signals of capacitor switching and load switching can be distinguished; and the method is simple in principle, high in reliability, small in training sample number and high in generalization ability, and can realize weak feature fault identification in different power distribution network topologies.

The invention relates to a construction method of a motor bearing fault identification model, in particular to a construction method of a wind driven generator bearing fault identification model. Theinvention aims to provide a construction method of a wind driven generator bearing fault identification model based on deep learning by utilizing a deep learning method, and further realize fault identification and positioning of a wind driven generator bearing. The construction method can greatly improve the accuracy of fault prediction. Compared with other methods, for the construction method, the general applicability and generalization of the deep learning network model to wind driven generator bearing fault identification are greatly improved. According to the construction method, accurate identification of different faults of the existing wind driven generator bearings of various models is easy to realize. The construction method is realized by the following steps: 1, presetting bearing fault types and quantity; 2, acquiring and preprocessing an original signal; 3, creating and configuring a deep learning network; 4, training a network; and 5, verifying the network accuracy.

A method for protect grounding electrode line of UHVDC system include injecting high-frequency current signal at head end of grounding electrode line, injecting high-frequency current signal at head end of grounding electrode line, inject high-frequency current signal at head end of grounding electrode line, injecting high-frequency current signal at head end of grounding electrode line. 2, collecting electric quantity data required by fault criterion of grounding electrode line based on injected high-frequency current signal; 3, jud that symmetry of the ground pole circuit structure based onthe data collect in the step 2; 4, based on the symmetry judging result of the step 3, judging the fault of the earth pole line; 5, aft that grounding electrode line fail, the relay protection devicesends out a fault alarm signal to the monitoring device; The reliability of grounding line protection method is improved.

The invention discloses a current transformer iron core coil fault diagnosis method based on spectral analysis. The method comprises the following steps of classifying typical faults of a current transformer iron core coil; establishing a current transformer equivalent circuit model under the typical faults, and obtaining equivalent circuit parameters of a current transformer iron core coil in a healthy and fault state; constructing a frequency spectrum response analysis circuit, acquiring frequency spectrum characteristic response curves of the current transformer under health, different fault types and different fault degrees, acquiring frequency spectrum change characteristics of the transformer in fault, and establishing a frequency spectrum characteristic library reflecting the state of the current transformer; and acquiring the current frequency spectrum characteristic curve of the current transformer iron core coil, then analyzing the frequency spectrum characteristics, comparing with data in a frequency spectrum characteristic library for analysis, and determining the fault type and the fault degree of the transformer iron core coil. According to the method, accurate diagnosis of the fault reason and the fault degree of the transformer iron core coil is achieved, and the defect that the specific fault type and the fault degree cannot be judged through an existing common method is overcome.

The invention provides a transformer fault identification method based on a kernel function extreme learning machine, and the method comprises the steps: obtaining a vibration signal as an analysis sample under the operation condition of a transformer, carrying out the noise elimination of the vibration signal, obtaining each frequency energy characteristic value of the vibration signal based on wavelet packet decomposition and reconstruction, extracting each frequency energy characteristic value as a fingerprint vector, dividing the fingerprint vectors into two sample sets, namely a training set and a test set, establishing an optimized kernel function extreme learning machine network model, performing model training by utilizing a fingerprint vector training set, inputting a to-be-tested set into the model, analyzing, calculating and outputting a check result, obtaining a working state of the transformer, and achieving fault identification of the transformer. According to the method, the abnormal problem that the system is caught in dimensionality disaster is solved, the performance of a fault judgment analysis result is further optimized and improved, and automatic identification of the abnormal fault of the power transformer is achieved.

The invention relates to a deep learning coalbed methane screw pump well health index prediction method. The method comprises one or more of the following steps: selecting at least one of a plurality of original parameters collected from a coalbed methane screw pump well as a main control parameter; performing merging processing on the at least one master control parameter to construct a health index; dividing the health state of the coal bed gas screw pump well into at least two different stages according to the health index; extracting health index data of the coal bed gas screw pump well as sample data, and constructing a health index prediction model by adopting a long-short term memory neural network; and predicting the health state change of the coal bed gas screw pump well by using the health index prediction model.

According to the invention, considering that unknown faults which are not considered in the operation process of a wind turbine generator are also likely to occur in addition to known faults, a wind generating set variable pitch bearing fault identification method based on an improved hidden Markov model is disclosed, comprises offline modeling and online identification, in the offline modeling step, a threshold statistic is defined based on the hidden Markov model, and the threshold statistic is used for identifying unknown faults. According to the method, a new threshold statistic is introduced, so that the probability that other unknown faults are mistakenly recognized as the variable-pitch bearing faults can be greatly reduced when online recognition is carried out on the variable-pitch bearing faults of the wind generating set. Therefore, compared with other existing methods, due to the fact that the time sequence of the data and the occurrence of unknown faults are fully considered, the method has higher accuracy for identifying the variable pitch bearing faults of the wind generating set.

The invention relates to a power distribution cabinet cable joint looseness early-warning method based on real-time data acquisition. The method comprises steps that a cable joint looseness early-warning comprehensive judgment expression is built through comprehensive analysis of cable joint looseness accident data and the corresponding relation between online monitoring data and joint fault data in a high-voltage power distribution cabinet; firstly, electrical parameters and external environment parameters influencing the stability of a cable joint are collected, calculated and stored to form a complete cable joint loosening accident data characteristic model; and then the corresponding relationship between the online monitoring data and the joint fault data in the power distribution cabinet is analyzed, the loosening state of the cable joint is judged more accurately, and fault identification is realized.

The invention provides a pilot protection method and device and a storage medium. The method comprises the following steps: acquiring time domain signal data of a target element at a preset sampling frequency; performing fusion processing on the time domain signal data of the plurality of first sampling periods to obtain first time domain signal combined data; based on a machine learning model, judging whether the target element has a fault according to the first time domain signal combination data; when it is judged that the target element breaks down according to the first time domain signal combination data, whether the target element breaks down in a second sampling period or not is judged according to the second time domain signal combination data based on a machine learning model, and the second sampling period is a sampling period after it is judged that the target element breaks down; and when it is judged that the target element has the same type of faults in multiple continuous second sampling periods, controlling the pilot protection system to execute a protection action on the target element. A machine learning model is utilized to realize fault identification of power system elements, so that the protection performance of a smart power grid is remarkably improved.

The invention belongs to the technical field of rolling bearing fault recognition, and discloses a low-speed heavy-duty bearing fault recognition method and system, a medium, equipment and a terminal, and the method comprises the steps: carrying out the filtering decomposition of a signal, solving the feature quantities of the first three component signals obtained through the decomposition, and constructing a feature value matrix; carrying out dimensionality reduction on the characteristic value matrix by adopting a distance evaluation technology, and screening out significant characteristics; and inputting the significant features into a BP neural network for training and testing, thereby realizing fault identification of the low-speed heavy-duty bearing. Aiming at the problems of broadband, non-stability and strong noise of an original signal, the method focuses on analyzing the step of filtering decomposition of the signal, solves the characteristic quantities of the first three component signals obtained by decomposition and constructs a characteristic value matrix, adopts a distance evaluation technology method to carry out dimension reduction on the characteristic value matrix, screens out significant characteristics, and finally obtains the characteristic value matrix. And inputting the significant features into a BP neural network for training and testing, thereby realizing accurate identification of the fault type of the low-speed heavy-duty bearing.

The invention provides a motor train unit traction converter performance detection method and device and terminal equipment. The method is applied to the field of data processing, and comprises the steps of obtaining target detection data of a motor train unit traction converter, wherein the target detection data comprises a target motor train unit speed and target performance influence information of multiple groups of traction converters corresponding to the target motor train unit speed, determining a target performance index value of a traction converter of the motor train unit based on the target speed of the motor train unit and a preset performance detection model, determining theoretical performance index values of the traction converters of the motor train unit based on the target performance influence information of the multiple groups of traction converters, and detecting the performance of the traction converter of the motor train unit according to the target performance index value and the theoretical performance index value. According to the motor train unit traction converter performance detection method and device and the terminal equipment provided by the invention, the working state of the traction converter can be evaluated more accurately.

The invention discloses a 500kV high-voltage cable fault automatic identification system, which is characterized by comprising a plurality of detection terminals, a cable fire-fighting terminal and anidentification controller, wherein the identification controller identifies the type of a fault occurring in a high-voltage cable based on data received from the detection terminals and the cable fire-fighting terminal and sends a processing instruction based on the type of the fault. The 500kV high-voltage cable fault automatic identification system disclosed by the invention can realize automatic detection and identification of high-voltage cable line faults, and can identify a fault in real time and output a processing instruction, so as to reduce the processing delay. System breakdown caused by a fault of a high-voltage cable is avoided to the greatest extent, and the working reliability of the power system is greatly improved.