Test-bed for pump as turbine transition process based on digital twinning

A test-bed for a pump as turbine transition process based on digital twinning, comprising:

• a pump as turbine (PAT) physical model, a PAT service system, a PAT twinning virtual system and a PAT twinning data system;
• the PAT physical model comprises a PAT module, a booster pump module, a control system module and an acquisition module;
• wherein the PAT module comprises a lower water level tank, a static pressure sensor, at least two valves including a first valve (valve 1) and a second valve (valve 2), a flowmeter, an upper water level tank and a motor;
• the booster pump module comprises the lower water level tank, the valves, a booster pump, the static pressure sensor, and the upper water level tank;
• the control system module comprises a frequency converter, an energy feedback unit and an electric cabinet;
• and
• the acquisition module comprises the static pressure sensor, a dynamic pressure sensor, a rotating speed torque sensor, an upper computer, and an acquisition card, wherein the PAT physical model is used for controlling the pump as turbine transition process and collecting data of the pump as turbine transition process;
• the PAT twinning data system is used for processing the data of the pump as turbine transition process to obtain a measurement database;
• the PAT twinning virtual system is used for carrying out numerical simulation on the pump as turbine transition process based on the measurement database to obtain a value obtained by numerical simulation of a PAT virtual system;
• and
• the PAT service system is used for processing the value obtained by the numerical simulation of the PAT virtual system, obtaining an optimal operation strategy and feeding the strategy back to the PAT physical model, and updating the measurement database at a same time until a target working condition is reached;
• wherein the PAT service system comprises:
• a machine learning module, an optimal strategy selection module and a judgment module;
• wherein the machine learning module is used for determining relationships among flow rate, outlet pressure, inlet pressure and valve opening through machine learning, and constructing a neural network model;
• the optimal strategy selection module is used for searching a corresponding strategy through the neural network model and obtaining the optimal operation strategy;
• and
• the judgment module is used for judging a matching relationship between the valve opening and adjustment time, calculating response time of the optimal strategy and judging a rationality of a strategy selection;
• an expression of a strategy optimization mode of the optimal strategy selection module is:
• Ymin=(valve 1 opening′−valve 1 opening)2+(valve 2 opening′−valve 2 opening)2;
• wherein the valve 1 opening′ is valve 1 target opening, and the valve 1 opening is valve 1 current opening;
• and the valve 2 opening′ is valve 2 target opening, the valve 2 opening is valve 2 current opening, and Ymin is an evaluation index;
• and
• the PAT service system is also used for evaluating a pressure pulsation value generated in the transition process of the numerical simulation, determining that an amplitude of a pressure pulsation is within a safe range, and after ensuring that a frequency of the pressure pulsation is inconsistent with a resonance frequency of a unit, transmitting a new valve opening command to the PAT physical model to operate the valve 1 and the valve 2 to adjust water flow between the upper and lower tanks, and initiate the motor excitation to generate stable voltage.