Abstract: A dual-mode combined control method for a multi-inverter system based on a double split transformer is provided. For an extremely-weak grid, the method provides the dual-mode combined control method for a multi-inverter system based on a double split transformer. According to the method, the equivalent grid impedance at a point of common coupling (PCC) of one grid-connected inverter (GCI) in the multi-inverter system based on the double split transformer is obtained with a grid impedance identification algorithm, and the system sequentially operates in a full current source mode, a hybrid mode, and a full voltage source mode according to a gradually increasing equivalent grid impedance, thereby effectively improving the stability of the multi-inverter system based on the double split transformer during variation of the strength of the grid. The method ensures that the system can still operate stably in the extremely-weak grid.

Abstract: A stability control method for a virtual synchronous generator (VSG) in a strong grid based on an inductance-current differential feedback is provided. A grid-connected topological structure of a VSG using the control method includes a direct-current (DC)-side voltage source, a three-phase inverter, a three-phase grid impedance and a three-phase grid. By controlling the VSG and controlling the inductance-current differential feedback, the method suppresses the oscillation of the output power from the VSG in the strong grid and implements the stable operation of an inner-loop-free VSG in the strong grid, without adding the physical inductance, increasing the cost of the filter and additionally providing a grid-side current sensor.

Abstract: Methods, systems and computer program products implementing character-level deep neural networks for information extraction are disclosed. A system uses character-level information retrieved from a transaction record to classify the transaction as a whole and to tag individual sections of the transaction record by entity type. The system processes the transaction record using multiple and separate character-level models. The system can use a one-dimensional neural network for featurization fed into a fully connected network for classification for identifying the most common classes of a transaction record. The system can identify one or more entities, e.g., service provider names, from the transaction using an RNN. The RNN can include one or more LSTM models. The LSTM models can be BI-LSTM models.

Abstract: Methods, systems and computer program products implementing character-level deep neural networks for information extraction are disclosed. A system uses character-level information retrieved from a transaction record to classify the transaction as a whole and to tag individual sections of the transaction record by entity type. The system processes the transaction record using multiple and separate character-level models. The system can use a one-dimensional neural network for featurization fed into a fully connected network for classification for identifying the most common classes of a transaction record. The system can identify one or more entities, e.g., service provider names, from the transaction using an RNN. The RNN can include one or more LSTM models. The LSTM models can be BI-LSTM models.