A workload evaluation method for electric power design documents

Abstract

Under the background of the generation of big data, the number of power design documents has increased sharply. It is inefficient to use traditional manual methods to count the workload, and there is a lack of effect of automatic and accurate evaluation. Traditional data processing methods fail to mine the structural spatial features in design documents, which is difficult to adapt to the spatial transformation of design elements, resulting in inaccurate feature information. For the calculation method, the traditional method uses simple weighted calculation, which is difficult to describe the complex nonlinear relationship between eigenvalues and workload. In this paper, a calculation method of power design document workload based on neural network is proposed. After obtaining the common graphic information in the power design document with known workload, the graphic data is simplified and normalized to the first eigenvector; Taking the largest B feature coefficients of each type of element as the feature vector of this type of element, and combining the feature vectors of all types of elements into the second feature vector; The third feature vector is obtained by the offset angle of any two entities relative to the main direction of any one of the two entities. The trained deep neural network model is used to obtain the workload of power design documents to be evaluated. According to the characteristics of power design documents, the evaluation method can effectively represent the structural spatial relationship and related invariance features of graphic elements in design documents. It relies on extracting the number of graphic elements in design documents, the spatial structure tightness of graphic elements and the direction statistics of graphic elements as the local and global feature representation information, the deep neural network structure model is introduced as the learning and training model to mine the nonlinear structural relationship between features. Through modeling, the workload of power design documents is automatically processed and analyzed to improve the accuracy of workload evaluation and calculation. The establishment of the model is suitable for all kinds of graphic design documents, which is universal and practical.