CN-122018322-A - Multi-machine regulation and control method and device for electric drive compressed air station

CN122018322ACN 122018322 ACN122018322 ACN 122018322ACN-122018322-A

Abstract

The invention provides a multi-machine regulation and control method and device for an electric drive compressed air station, which relate to the technical field of compressor regulation and control, and are characterized in that a digital twin model is built, a plurality of optimization targets are fused, an upper energy management decision maker and a middle strategy generator are trained in a layered and off-line mode, the trained upper energy management decision maker, middle strategy generator and bottom execution controller are further utilized to jointly control the electric drive compressed air station, dependence on manual experience scheduling can be greatly reduced, self-adaptive capacity is improved, gas consumption prediction information and electricity price prediction information are fully utilized, unit gas consumption and comprehensive operation cost, equipment service life and carbon emission intensity can be considered, the rationality of start-stop and load distribution of each air compressor is improved, the stability and robustness of operation of the whole electric drive compressed air station are improved, and synchronous optimization of energy conservation, health and reliability of the electric drive compressed air station is realized.

Inventors

YU MENG

Assignees

重庆尧猛节能环保科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260327

Claims (8)

1. A multi-machine regulation method of an electrically driven compressed air station, comprising: The method comprises the steps of constructing a digital twin model of an electric drive compressed air station, wherein the digital twin model comprises a cascade coupling structure of a data driving model of each air compressor unit in the electric drive compressed air station and a pipe network terminal pressure response model; Establishing an offline training environment based on the digital twin model, converting a multi-machine regulation task of the electrically driven compressed air station into a plurality of optimization targets, applying a deep reinforcement learning algorithm based on each optimization target, performing offline training on an upper energy management decision maker in the offline training environment, and performing offline training on a middle strategy generator based on a supervised learning algorithm and a simulated learning algorithm; Based on the trained upper-layer energy management decision maker, applying the current state information, the gas consumption prediction information and the electricity price prediction information of the electric drive compressed air station to decide an air compressor unit operation strategy and target interval information of the electric drive compressed air station; based on the trained middle-layer strategy generator, applying the gas consumption prediction information and the target interval information to perform constraint optimization on the operation strategy of the air compressor unit, generating a control strategy, and based on a bottom-layer execution controller, applying the control strategy to control the electric-drive compressed air station; the construction step of the data driving model of each air compressor unit comprises the following steps: for any air compressor unit, acquiring the running state quantity of the air compressor in the any air compressor unit, the inlet and outlet pressure, the inlet and outlet temperature, the frequency parameter, the current running mode code and the historical statistical characteristics of the air compressor in the any air compressor unit, and determining the actual flow and the actual power of the any air compressor unit; Inputting the running state quantity, the inlet and outlet pressure, the inlet and outlet temperature, the frequency parameter, the current running mode code and the historical statistical characteristics into a data driving model of any air compressor unit to obtain predicted flow and predicted power output by the data driving model of any air compressor unit; Training the data driving model of any air compressor unit based on the actual flow, the actual power, the predicted flow and the predicted power, and completing the construction process of the data driving model of any air compressor unit after the training process is finished; The input of the pipe network end pressure response model comprises predicted flow rate and air compressor unit distribution structure information of each air compressor unit, air storage tank pressure of the electric drive compressed air station and flow rate of a representative pipe network flow measuring point of the electric drive compressed air station; the output of the pipe network end pressure response model comprises the predicted pressure of each key node of the electrically driven compressed air station.
2. The method of claim 1, wherein applying a deep reinforcement learning algorithm based on each of the optimization objectives, in the offline training environment, performs offline training on an upper energy management decision maker, comprises: Under the offline training environment, constructing a reward function based on each optimization target, and constructing a state space based on the current state information, the gas consumption prediction information and the electricity price prediction information and an action space based on the air compressor unit operation strategy, wherein the reward function comprises comprehensive energy consumption, operation cost, constraint violation degree and start-stop regulation information of the electric drive compressed air station; And based on the reward function, the state space and the action space, applying a deep reinforcement learning algorithm to perform offline training on the upper-layer energy management decision-maker.
3. The method for multi-machine regulation of an electrically driven compressed air station according to claim 1, wherein the step of constructing a data driving model of any one air compressor unit further comprises: acquiring theoretical power and pressure flow physical constraint of any air compressor unit; And updating the input or training loss of the data driving model of any air compressor unit by taking the theoretical power and the pressure flow physical constraint as mechanism prior.
4. A multi-machine regulating method of an electrically driven compressed air station according to any one of claims 1 to 3, wherein the predicting step of the air consumption prediction information includes: acquiring historical gas consumption information, production plan characteristics, time characteristics and environmental characteristics of the electric drive compressed air station; And inputting the historical gas consumption information, the production plan characteristics, the time characteristics and the environmental characteristics into a gas consumption prediction model to obtain the gas consumption prediction information output by the gas consumption prediction model.
5. A multi-machine regulating method of an electrically driven compressed air station according to any one of claims 1 to 3, wherein said floor-based execution controller applies said control strategy to control said electrically driven compressed air station, followed by: calculating the instant efficiency of each air compressor unit, and calculating the relative efficiency deviation according to the instant efficiency; Calculating health scores of the air compressor units based on the relative efficiency deviation, and counting the health scores of the air compressor units in a preset time period; And if the health score of the target air compressor unit in the preset time period is lower than the preset score, the priority of the target air compressor unit is reduced in the decision process of the upper energy management decision maker after training.
6. A multi-machine regulation method of an electrically driven compressed air station according to any one of claims 1 to 3, wherein the off-line training of the intermediate layer strategy generator based on a supervised learning algorithm and a simulated learning algorithm comprises: Acquiring an air compressor unit operation strategy sample and a target interval information sample which are obtained by the decision of the upper energy management decision maker after training, and acquiring a gas consumption change trend sample, an expert control strategy of each air compressor unit and a working mode label; Inputting the air compressor unit operation strategy sample, the target interval information sample and the gas consumption change trend sample into the middle-layer strategy generator to obtain a control strategy sample output by the middle-layer strategy generator; Based on a bottom layer execution controller, the control strategy sample is applied to control the electric drive compressed air station, and the unit compressed air energy consumption and the comprehensive operation cost of the electric drive compressed air station are detected; And performing offline training on the middle-layer strategy generator based on the expert control strategy, the control strategy sample, the unit compressed air energy consumption under different working mode labels and the comprehensive operation cost.
7. A multi-machine regulating device for an electrically driven compressed air station, comprising: The system comprises a model construction module, a model generation module and a control module, wherein the model construction module is used for constructing a digital twin model of an electric drive compressed air station, and the digital twin model comprises a data drive model of each air compressor unit in the electric drive compressed air station and a cascade coupling structure of a pipe network terminal pressure response model; The system comprises an off-line training module, an off-line strategy generator, an off-line training module, a power-driven compressed air station and a power-driven compressed air station, wherein the off-line training module is used for building an off-line training environment based on the digital twin model, converting a multi-machine regulation task of the power-driven compressed air station into a plurality of optimization targets, applying a deep reinforcement learning algorithm based on each optimization target, performing off-line training on an upper energy management decision-making device under the off-line training environment, and performing off-line training on a middle strategy generator based on a supervised learning algorithm and a simulated learning algorithm; The upper layer decision-making module is used for deciding an air compressor unit operation strategy and target interval information of the electric drive compressed air station based on the trained upper layer energy management decision-making device by applying the current state information, the gas consumption prediction information and the electricity price prediction information of the electric drive compressed air station; the control module is used for carrying out constraint optimization on the operation strategy of the air compressor unit based on the trained middle-layer strategy generator, applying the gas consumption prediction information and the target interval information, generating a control strategy, and controlling the electric drive compressed air station based on a bottom-layer execution controller by applying the control strategy; the construction step of the data driving model of each air compressor unit comprises the following steps: for any air compressor unit, acquiring the running state quantity of the air compressor in the any air compressor unit, the inlet and outlet pressure, the inlet and outlet temperature, the frequency parameter, the current running mode code and the historical statistical characteristics of the air compressor in the any air compressor unit, and determining the actual flow and the actual power of the any air compressor unit; Inputting the running state quantity, the inlet and outlet pressure, the inlet and outlet temperature, the frequency parameter, the current running mode code and the historical statistical characteristics into a data driving model of any air compressor unit to obtain predicted flow and predicted power output by the data driving model of any air compressor unit; Training the data driving model of any air compressor unit based on the actual flow, the actual power, the predicted flow and the predicted power, and completing the construction process of the data driving model of any air compressor unit after the training process is finished; The input of the pipe network end pressure response model comprises predicted flow rate and air compressor unit distribution structure information of each air compressor unit, air storage tank pressure of the electric drive compressed air station and flow rate of a representative pipe network flow measuring point of the electric drive compressed air station; the output of the pipe network end pressure response model comprises the predicted pressure of each key node of the electrically driven compressed air station.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements a multi-machine regulating method of an electrically driven compressed air station according to any one of claims 1-6 when executing the computer program.

Description

Multi-machine regulation and control method and device for electric drive compressed air station Technical Field The invention relates to the technical field of compressor regulation and control, in particular to a multi-machine regulation and control method and device for an electric drive compressed air station. Background The industrial electrically-driven compressed air station comprises a plurality of air compressors driven by a motor, wherein the types of the air compressors can comprise a screw machine and a centrifugal machine, and the screw machine can comprise a power frequency screw machine and a variable frequency screw machine. The existing regulation and control of each air compressor in the electric drive compressed air station has the following problems that 1) a scheduling rule depends on manual experience and lacks self-adaptive capability, 2) economic information such as time-of-use electricity price, electricity charge demand and outsourcing maintenance cost is not fully utilized, 3) starting and stopping of each air compressor are unreasonable in load distribution, not only the energy consumption of unit gas is high, but also the service life and reliability of the air compressor are influenced by frequent starting and stopping and high-pressure ratio operation of the air compressor, and 4) the carbon emission intensity is still high on the premise of ensuring the pressure and flow of a workshop. Disclosure of Invention The invention provides a multi-machine regulation and control method and device for an electrically-driven compressed air station, which are used for solving the defects in the related art. The invention provides a multi-machine regulation and control method of an electrically-driven compressed air station, which comprises the following steps: The method comprises the steps of constructing a digital twin model of an electric drive compressed air station, wherein the digital twin model comprises a cascade coupling structure of a data driving model of each air compressor unit in the electric drive compressed air station and a pipe network terminal pressure response model; Establishing an offline training environment based on the digital twin model, converting a multi-machine regulation task of the electrically driven compressed air station into a plurality of optimization targets, applying a deep reinforcement learning algorithm based on each optimization target, performing offline training on an upper energy management decision maker in the offline training environment, and performing offline training on a middle strategy generator based on a supervised learning algorithm and a simulated learning algorithm; Based on the trained upper-layer energy management decision maker, applying the current state information, the gas consumption prediction information and the electricity price prediction information of the electric drive compressed air station to decide an air compressor unit operation strategy and target interval information of the electric drive compressed air station; And based on the trained middle-layer strategy generator, applying the gas consumption prediction information and the target interval information to perform constraint optimization on the operation strategy of the air compressor unit, generating a control strategy, and based on a bottom-layer execution controller, applying the control strategy to control the electric-drive compressed air station. According to the multi-machine regulation and control method of the electric drive compressed air station provided by the invention, based on each optimization target, a deep reinforcement learning algorithm is applied, and under the offline training environment, an upper energy management decision maker is subjected to offline training, and the method comprises the following steps: Under the offline training environment, constructing a reward function based on each optimization target, and constructing a state space based on the current state information, the gas consumption prediction information and the electricity price prediction information and an action space based on the air compressor unit operation strategy, wherein the reward function comprises comprehensive energy consumption, operation cost, constraint violation degree and start-stop regulation information of the electric drive compressed air station; And based on the reward function, the state space and the action space, applying a deep reinforcement learning algorithm to perform offline training on the upper-layer energy management decision-maker. According to the multi-machine regulation method of the electric drive compressed air station provided by the invention, the construction steps of the data drive model of each air compressor unit comprise the following steps: for any air compressor unit, acquiring the running state quantity of the air compressor in the any air compressor unit, the inlet and outlet pressure, the inlet and outlet tempera