Search

In short-term load forecasting, deep learning models with recurrent units are widely used. For Long Short-term Memory (LSTM) models, in terms of the input features at one particular time-step, there is a weight series for each input feature. Hence, an input feature with a larger value of weight can be regarded as more important to the final forecasting result. Since the weights of the features are time-invariant, the model considers the input features as equally important at each time step, thus lack of the capability of reflecting the importance fluctuation through the time span. To tackle this issue, this paper proposes a forecasting scheme, which integrates mutual information (MI) into bi-directional long short-term memory (BILSTM) network. The MI method is used to extract the importance value of input features at different time-steps, and constitute the fluctuation matrix of the importance values for input features, which is used as coefficients to correct the original input feature. The corrected features are substituted into the BILSTM network, which further improves the forecasting accuracy. Numerical experiments have been carried out based on real-world load data. The results demonstrate the superior performance of the proposed scheme compared with a series of state-of-the-art methods and thus indicate the importance and effectiveness of capturing the fluctuation of the importance value.

As a critical part of modern industry, assembly line plays an important role in mass-producers because of its great advantages in production efficiency and construction cost. With the increasing demand for customized production, the fixed processing flow of assembly line is becoming more and more difficult to adapt to the changes. The introduction of automated guided vehicles (AGVs) as the material handling equipment improve the flexibility of assembly line, which also derives a complex AGV scheduling problem. In the recent years, because of the limitation in place, no-buffer assembly line appears in some plant, which derives a new problem, processing sequence conflicts (PSC). This paper builds a Petri net model to deal with the AGV scheduling problem and processing sequence conflicts in the no-buffer assembly line. A genetic algorithm based look-ahead scheduling algorithm (LASA) with processing sequence conflict-free is developed to minimize the total processing makespan in a no-buffer assembly line by introducing prospective tasks. Chain task strategy is integrated to further improve the performance of LASA in the case of AGVs deficiency. Two groups of simulation experiments are carried out to test the performance of the proposed algorithms in different scenarios. The results show the proposed algorithm has advantages in processing makespan, AGVs’ efficiency and machines’ efficiency. A robot experiment is carried out in a real no-buffer assembly line. The processing makespan of experiment is close to the theoretical lower bound.