Intelligent Monitoring Of Grounding Faults In Power Grids Of Biochemical Energy

The grid structure has gradually changed from the original backbone network to the distributed grid.The main reason is that biomass energy,solar energy and other clean energy are injected into the grid,which makes the whole network more complicated and brings many potential risks to the security monitoring of the power grid.It is difficult to continue the fault detection by using the traditional methods,so this paper proposes a new method for grounding fault detection.Firstly,this paper proposes to reveal the change law of grounding faults by the means of fault feature images.By comparing the common time-frequency analysis methods,S-transform algorithm is used to analyze the transient features of existing grounding fault data.Different fault types can obtain different fault feature images after S-transform,from which the law of time-frequency-mode mutation of the fault signal when the fault occurs can be visually displayed.Secondly,aiming at the problem that the existing fault data set is not rich enough and the fault type is single,the power simulation software PSCAD is used to build the simulation model,set the grounding fault types under different conditions,and analyze the impact of different fault occurrence times,grounding resistance,fault initial phase angle and other factors on the fault features.Finally,this paper proposes to combine the convolutional neural network in deep learning with S-transform algorithm for the identification of grounding faults,and constructs deep model to reflect the mapping relationship between fault types and fault features to complete the classification.The model of this paper uses the fault feature images as input to learn the fault features.By observing the changes of various parameters in the training process,the network parameters are continuously adjusted to achieve the purpose of identifying the fault types and the optimal model of fault detection.In this paper,the single-phase grounding faults in the distributed grid is taken as the research object,which can accurately identify the high-resistance grounding,low-resistance grounding and normal conditions under load disturbance,with the accuracy rate of over 90%.Compared with other algorithms,the proposed method has high recognition accuracy,strong robustness and good engineering application value.