Energy storage power station intelligent monitoring method based on multilayer feedforward neural network
A technology of feedforward neural network and energy storage power station, applied in the field of intelligent monitoring of energy storage power station based on multi-layer feedforward neural network, can solve the problems of untimely accident discovery and insufficient monitoring of energy storage power station, and achieve short response time , high accuracy, and the effect of improving efficiency
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0074] M×N high-precision sensors are arranged in a rectangular distribution in the energy storage power station, and each sensor is at t 0 ~t y-1 A total of y concentration data are collected at the moment
[0075]
[0076] where x=CO, CO 2 or H 2 , Indicates that the sensor located at row i and column j is at t 0 The concentration of x gas collected at time.
[0077] Perform data denoising processing:
[0078] Firstly, the corresponding Hankel matrix is constructed based on the time series of the concentration signal, and then the singular values of the matrix are extracted as the characteristics of discharge. The Hankel matrix is a matrix in which the elements of each inverse diagonal line in the matrix are equal. When extracting, the elements located in the same position in the matrix of each page are constructed into a one-dimensional signal sequence.
[0079]
[0080] is the concentration time series collected by the sensor located in row i, column ...
Embodiment 2
[0120] Two neural networks are constructed to predict the number of rows and columns of leak locations in energy storage power plants. Taking a small energy storage power station with only 12 sensors as an example, the best model accuracy is achieved using a 12-30-30-1 architecture (number of neurons in each layer), as shown in Figure 5. The input layer uses concentration data from 12 sensors at specific time points as input data, and a neuron in the output layer represents the predicted result of row or column number.
[0121] The artificial neural network model can detect and locate the gas leak location after the training process. like Figure 5 The shown convergence of training error and accuracy shows that the minimum loss reaches 8.8 × 10 after 500 iterations -3 , with a maximum accuracy of 0.996.
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


