| With the development of the intelligent era,the current operation condition monitoring of electrical equipment combines new technology and new method to expand its development.Based on the visible and infrared images of electrical equipment,this paper proposed a neural network structure suitable for the type recognition of electrical equipment,and realized the intelligent recognition of electrical equipment.On this basis,we proposed a set of early warning evaluation scheme for the running state of the electrical equipment combined with the big data technology based on the data information of the running state parameters and environment of the electrical equipment.Inspired by the different temperature field distribution characteristics of infrared images of different types of electrical equipment,this paper combined with visible images with rich contour features and used the scale-invariant feature transform(SIFT)algorithm to carry out image registration and fusion processing.The fused images have both advantages.Deconvolution feature extraction algorithm is used to extract the features of the fusion image to obtain the feature mapping matrix exclusive to the electrical equipment.The successful acquisition of the fusion image and feature mapping matrix provides a necessary prerequisite for the intelligent recognition of the subsequent electrical equipment.Different from traditional random initialization convolution convolution kernels of neural networks method of study,this paper proposed an improved neural network identification model.Deconvolution feature extraction technology in feature mapping matrix of electrical equipment for fusion image are extracted and through migration study,this will be the characteristics of the mapping matrix as the initial convolution kernel matrix of convolution neural network model.Moreover,the improved network model is used to supervise and train the visible and infrared images of the equipment.Through repeated iterative learning,the accuracy of intelligent identification of electrical equipment is improved,and then the automatic identification of the equipment is realized.On the basis of intelligent identification of electrical equipment types,the temperature data of infrared images of equipment are collected.Combined with the condition monitoring quantity of equipment,the random matrix theory is used to integrate the specific monitoring quantity into a multi-dimensional matrix.The change rule of the eigenvalue ring formed by the matrix eigenvalue in the complex plane was statistically analyzed.The corresponding relationship between the equipment running state and the change rule of the eigenvalue ring was found out.The parameter threshold was given to judge the equipment running state.Finally,it showed the early warning suggestions for electrical equipment.Through the above research,intelligent identification and fault early warning of equipment can be realized,and the traditional manual monitoring method can be transformed into intelligent monitoring.The rationality and effectiveness of the proposed method are verified by an example analysis of the measured data,which provides a new idea for the research of electrical equipment identification and fault early warning. |
