The inherent intermittency and uncertainty of wind power have brought challenges in accurate wind power output forecasting, which also cause tricky problems in the integration of wind power to the grid. In this paper, a hybrid deep learning model bidirectional long short term memory-convolutional neural network (BiLSTM-CNN) is proposed for short-term wind power forecasting. First, the grey correlation analysis is utilized to select the inputs for forecasting model; Then, the proposed hybrid model extracts multi-dimension features of inputs to predict the wind power from the temporal-spatial perspective, where the Bi-LSTM model is utilized to mine the bidirectional temporal characteristics while the convolution and pooling operations of CNN are utilized to extract the spatial characteristics from multiple input time series. Lastly, a case study is conducted to verify the superiority of the proposed model. Other deep learning models (Bi-LSTM, LSTM, CNN, LSTM-CNN, CNN-BiLSTM, CNN-LSTM) are also simulated to conduct comparison from three aspects. The results show that the BiLSTM-CNN model has the best accuracy with the lowest RMSE of 2.5492, MSE of 6.4984, MAE of 1.7344 and highest R2 of 0.9929. CNN has the fastest speed with an average computational time of 0.0741s. The hybrid model that mines the spatial feature based on the extracted temporal feature has a better performance than the model mines the temporal feature based on the extracted spatial feature.
From the optical fiber scattering light field distribution model a two_dimensional optical rough surface reflection_reception model was established,and a two_dimensional fiber optic sensor was designed with the partial space optical scattering of the rough surface tauen into consideration by introducing the gradation spaceconcept used by Huffman for image analysis,during the experiments,vertical milling standard samples from 3 to 7 were used as checked objects for two_dimensional roughness,and data was obtained from the measured curve peaks of the checked surface ,to establish the relationship between average disaffection value of roughness and standard deviation.
The projection pursuit regression theory is applied to analysis the reliability of vehicle hydraulic brake system and build the projection pursuit regression model. This model on training sample fitting effect is good and shows extremely strong adaptability. We predict the reliability of certain type hydraulic brake system by this model which provides scientific basis for research on reliability of hydraulic brake system.
A heterogeneous computing platform is an computing system that composed of different types of computing units. By fully using the computing ability of different types of computing units, the heterogeneous computing platform can achieve better performance and power efficiency than traditional homogeneous computing platform. In this paper, we firstly summarize and analyze the key factors that affect the performance of a heterogeneous computing platform. Next, we conduct a specific survey about these key factors from both software and hardware aspects and introduce some research results and key technologies at present. And then, we introduce the heterogeneous computing framework and make a comparison between OpenCL and HSA which are now both promising. At last, we analyze the future directions of the heterogeneous computing platform.
To gain and sustain competitive advantage, a firm may orient its strategy to differentiating its products and services or reducing its costs. Most prior studies measure strategy using data from executive surveys, proprietary databases, or panelist evaluation. These research methods are costly to implement and, consequently, restrict the scope of research. We develop a textual measure of strategy using the annual 10-K filings of U.S. public firms. We validate our measure by showing that its properties and its associations with operating activities are consistent with those suggested in the prior literature. We document that our measure of strategy is associated with firm performance as expected. Overall, our approach can be useful for many potential studies in management and other business disciplines.
Software patterns will provide mature soluitons to common software problems. They are abstraction of software domain knowledge and their use realizes the reuse and sharing of software domain knowledge. Based on the introduction and analysis of the primary situation of pattern research in the software domain,this paper reveals the necessity and importance of pattern reseach.
With the development of the power grid and the prompting of the unattended substation,the requirement of the substation automation system debugging is increasing.The current debugging methods are listed in this paper,with comparison among them.Related suggestion and improvement are also presented in detail,with the following work on the substation automation system debugging pointed out in the end.
Because of overheat and over wear, the distortion and invalidation will be produced easily in combining friction discs of wet clutch of heavy vehicle transmission system. So the paper studies the sliding friction force of a pair of steel and friction disc, calculates the heat flow density, obtains the heat exchange function with axial symmetry, and establishes the finite element model of temperature field. And then the transient heat field simulation is analyzed, and temperature field distribution curves of combining pairs are obtained. Thus the connection is found between time and temperature along radial and axial direction, and temperature field characteristics are obtained with touch press, relative speed, and sliding time.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)