Search

CN-122000945-A - Multi-energy collaborative prediction control method and system based on digital twin

CN122000945ACN 122000945 ACN122000945 ACN 122000945ACN-122000945-A

Abstract

The invention discloses a multi-energy collaborative prediction control method and system based on digital twin, and relates to the technical field of industrial control; the method comprises the steps of inputting historical loads of each energy source in a plurality of periods and weather forecast data of the next period into a trained LSTM forecast model to conduct multi-energy source load forecast, outputting to obtain forecast values of each energy source load of the next period, constructing an optimization model, solving the optimization model to obtain a preliminary scheduling plan, conducting simulation deduction on control instructions in the preliminary scheduling plan to obtain a forecast system state of the next period, correcting the preliminary scheduling plan based on the forecast system state of the next period if a result of the simulation deduction deviates from the forecast, and obtaining a final control instruction set.

Inventors

  • YANG YONGJIAN
  • WU SHANG

Assignees

  • 中宜港能(上海)能源发展有限公司

Dates

Publication Date
20260508
Application Date
20260130

Claims (10)

  1. 1. The digital twin-based multi-energy collaborative prediction control method is characterized by comprising the following steps of: inputting the historical load of each energy source in a plurality of historical periods and the weather forecast data of the next period into a trained LSTM forecast model to conduct multi-energy source load forecast, and outputting to obtain the forecast value of each energy source load of the next period; based on a pre-selected decision variable, each energy load predicted value of the next period and preset constraint parameters, an optimization model is built, and the optimization model is solved to obtain a preliminary scheduling plan; If the result of the simulation deduction is deviated from the expected value, the preliminary scheduling plan is corrected based on the state of the system of the next period prediction, and a final control instruction set is obtained; and executing the control instruction based on the final control instruction set, and performing feedback monitoring and model correction.
  2. 2. The digital twin-based multi-energy collaborative prediction control method according to claim 1, wherein an optimization objective function of an optimization model includes a stability-related term; constraint conditions of the optimization model are consistent with constraint parameters.
  3. 3. The digital twin based multi-energy collaborative prediction control method according to claim 2, wherein stability related terms include minimizing the magnitude of each device output change, minimizing the rate of change of stored energy state of charge, minimizing the extent to which each energy network operating parameter deviates from a preset safety margin, minimizing grid frequency and voltage fluctuation magnitude.
  4. 4. The digital twinning-based multi-energy collaborative prediction control method according to claim 2, wherein the constraint parameters include an upper generator output limit, a lower generator output limit, an upper energy storage charging power limit, an upper energy storage discharging power limit, an energy storage capacity, an upper and lower energy storage SOC limit, an upper chiller output limit, an upper boiler output limit, an upper load adjustment limit, a coupling constraint parameter, and a safety margin requirement.
  5. 5. The digital twin-based multi-energy collaborative prediction control method according to claim 1, wherein the method for obtaining the next period predicted system state comprises: pre-constructing a virtual system model, taking the predicted value of each energy load and the planned control action as inputs, and outputting to obtain the predicted system state of the next period; The virtual system model is a digital twin body of the whole system, wherein the planned control action is a control instruction corresponding to an initial scheduling scheme, the initial scheduling scheme is obtained by continuing the control instruction actually executed in the previous period and calculating the control instruction of the equipment by combining with each energy load predicted value.
  6. 6. The digital twin-based multi-energy collaborative prediction control method according to claim 5, wherein system state evolution calculation is performed in a virtual system model according to energy conservation law; The virtual system model respectively establishes a supply and demand balance equation according to electric power, cold energy, heat energy and fuel gas, wherein the total supply quantity in the electric power balance equation is equal to the sum of total electric load and electric power network loss, the total supply quantity is the sum of renewable energy output predicted value, generator output command and energy storage discharge power, the total electric load is the sum of electric load predicted value, the total supply quantity in the cold energy balance equation is equal to the cold load predicted value, the total supply quantity is the sum of refrigerator output command and cold energy storage discharge quantity, the total supply quantity in the heat energy balance equation is equal to the sum of heat load predicted value, the total supply quantity in the fuel gas balance equation is equal to the sum of fuel gas load predicted value and fuel gas equipment consumption, and the total supply quantity is the sum of fuel gas supply equipment output command and fuel gas energy storage discharge quantity.
  7. 7. The digital twin based multi-energy collaborative prediction control method according to claim 2, wherein determining whether the result of analog deduction is a deviation from expected includes: in the next period pre-estimated system state, if the simulation value of any key index exceeds the corresponding preset target value allowable range or the distance between the simulation value of any key index and the constraint boundary is smaller than a preset safety threshold, judging that the simulation value of any key index deviates from the expected value; the constraint boundary is the upper and lower limit values of constraint parameters; The key indexes comprise energy storage charge state, generator output, refrigerator output, boiler output, power grid voltage, power grid frequency, heat supply network water supply and return temperature, heat supply network water supply and return pressure, air network pressure and air network flow; The preset target value is a standard range of safe and stable operation of the system, and the preset safety threshold value is a buffer zone between the constraint boundary and the preset target value allowable range.
  8. 8. The digital twin based multi-energy collaborative prediction control method according to claim 2, wherein the method of deriving a final control instruction set includes: Constructing a secondary optimization model, wherein the secondary optimization model comprises minimizing the sum of the balance deviation of supply and demand of multiple energy sources, minimizing the weighted sum of the adjustment quantity of the control instructions of each device, minimizing the weighted sum of the change rate of the control instructions of each device and minimizing the sum of penalty items deviating from the safety margin; solving the secondary optimization model to obtain the adjustment quantity of the control instruction; And superposing the adjustment quantity of the control instruction on the control instruction of the preliminary scheduling plan to obtain a final control instruction set.
  9. 9. The digital twin based multi-energy collaborative prediction control method according to claim 1, wherein the decision variables used to construct the optimization model include generator output commands, stored energy charge and discharge commands, load adjustment commands, cold supply output commands, heat supply output commands, and gas supply equipment output commands.
  10. 10. A digital twin-based multi-energy collaborative prediction control system for implementing the digital twin-based multi-energy collaborative prediction control method according to any one of claims 1-9, comprising: The acquisition module is used for acquiring the energy loads in real time; The prediction module is used for inputting the historical load of each energy source in a plurality of periods and the weather prediction data of the next period into a trained LSTM prediction model to perform multi-energy load prediction, and outputting to obtain the predicted value of each energy source load of the next period; The solving module is used for constructing an optimization model based on a pre-selected decision variable, each energy load predicted value of the next period and a preset constraint parameter, and solving the optimization model to obtain a preliminary scheduling plan; The correction module is used for carrying out simulated deduction on the control instruction in the preliminary scheduling plan to obtain a next period estimated system state, and correcting the preliminary scheduling plan based on the next period estimated system state if the simulated deduction result is deviated from the expected result to obtain a final control instruction set; and the execution module executes the control instruction based on the final control instruction set and performs feedback monitoring and model correction.

Description

Multi-energy collaborative prediction control method and system based on digital twin Technical Field The invention relates to the technical field of industrial control, in particular to a digital twin-based multi-energy collaborative prediction control method and system. Background In a multi-energy collaborative operation control scene of high-proportion renewable energy grid connection, widely participated energy storage equipment and flexible load, a closed-loop control system covering prediction, optimization and execution is required to be constructed for realizing dynamic balance and efficient utilization of energy supply and demand so as to cope with random fluctuation of a source load side. The technology has the core principle that a prediction module establishes a prediction model of load and renewable energy output by collecting historical operation data and real-time state information to provide data support for a subsequent optimization decision, an optimization module generates an optimal strategy of power output adjustment, energy storage charge-discharge control and load adjustment by combining system operation constraint and economic environment-friendly targets based on a prediction result, and an execution module issues an optimization instruction to each energy device and feeds back actual operation data of the device to the prediction module to realize dynamic updating and closed loop iteration of the model. However, because the construction of the prediction model directly depends on the execution result of the previous round of control instruction, the control action generated by the optimization module can change the actual running state of the system, so that the data fed back to the prediction module deviates from the natural running trend, and the next round of prediction result generates deviation, and the optimization module further amplifies the deviation between the running state of the system and the natural trend by taking the deviation prediction value as the basis to formulate a control strategy, so as to form an endogenous feedback loop for mutually strengthening the prediction deviation and the control action. The problem can cause zigzag deviation of the predictive value and the actual value, the power of the energy equipment fluctuates back and forth on the operation constraint boundary, the expected energy-saving and efficiency-improving target cannot be achieved, the number of charge and discharge cycles of the energy storage equipment can be increased abnormally, the running instability of the system is aggravated, and the risks of equipment faults and energy supply interruption are increased. In view of the above, the present invention provides a digital twin-based multi-energy collaborative prediction control method and system to solve the above problems. Disclosure of Invention In order to overcome the defects in the prior art and achieve the purposes, the invention provides a multi-energy collaborative prediction control method based on digital twinning, which comprises the following steps: inputting the historical load of each energy source in a plurality of historical periods and the weather forecast data of the next period into a trained LSTM forecast model to conduct multi-energy source load forecast, and outputting to obtain the forecast value of each energy source load of the next period; based on a pre-selected decision variable, each energy load predicted value of the next period and preset constraint parameters, an optimization model is built, and the optimization model is solved to obtain a preliminary scheduling plan; If the result of the simulation deduction is deviated from the expected value, the preliminary scheduling plan is corrected based on the state of the system of the next period prediction, and a final control instruction set is obtained; and executing the control instruction based on the final control instruction set, and performing feedback monitoring and model correction. Further, the decision variables used for constructing the optimization model include a generator output command, an energy storage charge-discharge command, a load adjustment command, a cooling output command, a heating output command and a gas supply equipment output command. Further, the optimization objective function of the optimization model contains stability-related terms; constraint conditions of the optimization model are consistent with constraint parameters. Further, the stability related items include minimizing the output variation amplitude of each device, minimizing the energy storage state of charge variation rate, minimizing the degree of deviation of each energy network operation parameter from a preset safety margin, and minimizing the power grid frequency and voltage fluctuation amplitude. Further, the constraint parameters comprise an upper generator output limit, a lower generator output limit, an upper energy storage charging power limit, an upp