Integrated energy system double-layer optimization operation method based on multivariable predictive control

Abstract

The present application relates to a kind of integrated energy system double-layer optimization operation method based on multivariable predictive control, comprising: the latest operation data of sampling system in dynamic scheduling layer, based on the dynamic characteristic model of equipment that can be used to describe the heating link of system equipment, according to the latest renewable energy and load demand forecast data within day, with the minimum economic cost and the minimum of dynamic supply-demand deviation index as optimization goal, with power balance constraint and equipment characteristic constraint as constraint, carry out within day rolling scheduling, provide the power scheduling instruction of each equipment in the next scheduling period;Real-time control layer takes the heat-electricity coupling link of gas internal combustion engine and its waste heat recovery equipment and heat pump as control object, based on multivariable predictive control method, the power scheduling instruction issued by dynamic scheduling layer is corrected in real time, using the power scheduling instruction after correction acts on system, can reduce the disturbance influence brought by prediction deviation, improves system energy supply accuracy and operating stability.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization operation technology, and in particular to a two-level optimization operation method for integrated energy systems based on multivariable predictive control. Background Technology

[0002] Renewable energy power generation, such as photovoltaic (PV) and wind power, is significantly affected by environmental factors, resulting in strong randomness and uncertainty in power output. This inevitably leads to a certain degree of error between predicted and actual power output data. Furthermore, as the penetration rate of PV and wind power continues to increase, the energy system operates in a highly disturbed environment for extended periods, posing greater challenges to the system's stable operation.

[0003] Currently, there is little in-depth research and discussion on feedback correction in scheduling layer strategies. The equipment models established by the scheduling layer are generally directly treated as actual equipment. However, because actual equipment often has complex dynamic characteristics and strong nonlinearity over a wide operating range, there is still a certain model deviation between the scheduling model and the actual equipment model. Therefore, the actual output of the system equipment may differ from that calculated by the scheduling layer.

[0004] The stable operation of integrated energy systems has always been a hot research topic. Integrated energy systems are typical multivariable, strongly coupled, and highly inertial systems, and are extremely susceptible to multiple / strong external disturbances. Conventional control algorithms may struggle to achieve satisfactory operational results. Therefore, fully exploring the thermoelectric coupling characteristics of the system and researching advanced control strategies and two-layer optimization operation methods including scheduling and control layers are crucial for reducing the impact of prediction deviations in renewable energy output and electrical load, as well as scheduling model deviations, on system operation, and ensuring stable and economical system operation. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a two-layer optimization operation method for integrated energy systems based on multivariate predictive control. The aim is to reduce the impact of prediction deviations in renewable energy generation and electrical load, as well as model deviations in the scheduling layer, on system operation, thereby ensuring reliable energy supply and stable operation of the system.

[0006] The technical solution adopted in this invention is as follows:

[0007] A two-layer optimization operation method for integrated energy systems based on multivariable predictive control, comprising operation optimization based on a dynamic scheduling layer and a real-time control layer, including:

[0008] The dynamic scheduling layer samples the latest operating data of the system, and based on the dynamic characteristic model of the equipment that can be used to describe the heating link of the system equipment, and according to the latest daily forecast data of renewable energy and load demand, with the optimization objectives of minimizing economic costs and minimizing dynamic supply and demand deviation indicators, and with power balance constraints and equipment characteristic constraints as constraints, performs intraday rolling scheduling and provides power scheduling instructions for each equipment in the next scheduling period.

[0009] The real-time control layer takes the gas internal combustion engine, other heat recovery equipment, and the thermoelectric coupling link of the heat pump as the control object. Based on the multivariate predictive control method, it corrects the power scheduling command issued by the dynamic scheduling layer in real time and uses the corrected power scheduling command to act on the system.

[0010] The further technical solution is as follows:

[0011] The dynamic characteristic model described above can be used to describe the heating process of a gas internal combustion engine or heat pump, and is represented in state-space form as follows:

[0012]

[0013] In the formula, x ep r is a state variable ep Q is the power scheduling command for the device. ep A represents the actual thermal output power of the equipment. ep B ep C ep Let be the coefficient matrix, and kk represent the kk-th sampling time.

[0014] The function expression of the optimization objective is: minJ=w1AIM1+w2AIM2, where J is the objective function value, AIM1 is the economic index, AIM2 is the dynamic supply and demand deviation index, including the dynamic deviation index of heat power and the dynamic deviation index of water supply temperature, and w1 and w2 are the corresponding weights.

[0015] The economic indicators AIM1 specifically include:

[0016]

[0017] In the formula, Δt is the scheduling step size, and N k K represents the number of time-domain steps during the rolling process. fuel For the biogas fuel cost of a gas-fired internal combustion engine, K ice For the operation and maintenance costs of gas internal combustion engines, K hp For the operation and maintenance costs of heat pumps, K bat For the operation and maintenance costs of the battery, K hs For the operation and maintenance costs of the thermal storage tank, D ice,f P is the mass flow rate of the fuel consumed. iceP is the output electrical power of a gas internal combustion engine. hp P is the electrical power consumed by the heat pump. bat Q represents the power output value of the battery. hs Let i represent the power output value of the thermal storage tank, and i represent the i-th scheduling time period.

[0018] The dynamic supply-demand deviation index AIM2 specifically includes:

[0019]

[0020] In the formula, Δt kk For the sampling period, N kk The number of samples in each scheduling time period, Q hp (i,j),T hp (i,j), Q hx (i,j) and T hx (i,j) represent the heat pump output power, heat pump outlet hot water temperature, waste heat recovery power of the gas internal combustion engine, and waste heat recovery hot water temperature at the j-th sampling time within the i-th scheduling time period, respectively. T supply G represents the temperature of the hot water after mixing with the heat pump water and the waste heat recovery water. hp G represents the hot water flow rate of the heat pump. hx Q represents the hot water flow rate in the waste heat recovery process. load To meet the user's heat load requirements.

[0021] The power balance constraints include system electrical power and thermal power balance constraints; the equipment characteristic constraints include operating characteristic constraints of the gas internal combustion engine and other heat recovery links, operating characteristic constraints of heat pump equipment, characteristic constraints of battery equipment, and characteristic constraints of thermal storage tank equipment.

[0022] The multivariate predictive control method includes:

[0023] S1: Select input and output variables to conduct an open-loop step response experiment and identify the transfer function model of the integrated energy system. The output variables are the remaining electric power and the total hot water supply temperature, and the input variables are the internal combustion engine electric power command adjustment amount and the heat pump heat power command adjustment amount.

[0024] S2: Define the augmented state-space model of the transfer function model:

[0025]

[0026]

[0027] In the formula, y(kk) is the system output, and Δx d(kk) is the change in state variable at time kk, x(kk) is the amplified state variable, Δu(kk) is the change in control variable at time kk, and A d B d C d These are the coefficient matrices corresponding to the state-space model of the controlled object, where A, B, and C are the coefficient matrices after state augmentation, and I... p×p It is a p×p dimensional identity matrix, O d It is a zero matrix;

[0028] S3: Set controller parameters, including prediction time domain P, control time domain M, error weight matrix Q, and control weight matrix R;

[0029] S4: Estimate the current state of the system based on the current system output y(kk);

[0030] S5: Establish the prediction equation Y(kk)=Fx(kk)+ΩΔU(kk) to predict the output Y(kk) of the system at P future times, where ΔU(kk) represents the future sequence of control variable changes, and F and Ω are the coefficient matrices corresponding to the prediction equation;

[0031] in,

[0032]

[0033] S6: Build Performance Metrics Solve for ΔU(kk), and select the optimal control change Δu(kk) at time kk as the output to ensure that the controlled variable can track the setpoint, while the fluctuation of the control variable is not too drastic. Where Y... r (kk)=[y r (kk+1),y r (kk+2),…,y r (kk+P)] T y represents a sequence of setpoints at future times. r The set value for a single moment. These are the quadratic forms when the weight matrix takes Q and R, respectively;

[0034] S7: Calculate the control correction amount u(kk) = u(kk-1) + Δu(kk), u(kk) = [u1(kk), u2(kk)] T u1(kk) and u2(kk) are the control correction values ​​for the power scheduling commands of the gas internal combustion engine and the heat pump, respectively.

[0035] S8: Modify the power scheduling command issued by the scheduling layer by the control correction amount u(kk). The modified power command is applied to the system. The system output information is sampled and updated at the next moment. Repeat steps S4 to S8.

[0036] The estimation of the system's current state using the Kalman filter algorithm includes:

[0037] Perform state prediction in one step: It is the change in control quantity at time kk-1. It is the state estimate at time kk-1. It is the state estimate at time kk from time kk-1;

[0038] Calculate the covariance matrix of the one-step state prediction: P kk / kk-1 =AP kk-1 A T +Q kk-1 P kk-1 and P kk / kk-1 These are the state covariance matrices at time kk-1 and at time kk, calculated recursively, respectively, Q. kk-1 It is the process noise covariance matrix at time kk-1;

[0039] Calculate the filter gain matrix: K kk =P kk / kk-1 C T (CP kk / kk-1 C T +R kk ) -1 K kk It is the filter gain matrix, R kk It is the covariance matrix for measuring noise;

[0040] The state is corrected based on the current measurement value:

[0041] Calculate the covariance matrix of the state estimate: P kk =(IK kk C)P kk / kk-1 I is the identity matrix, P kk It is the state covariance matrix at time kk.

[0042] The selection of the prediction time domain P should cover the main dynamic characteristics of the system, and the control time domain M should not be greater than the prediction time domain P; the error weight matrix Q and the control weight matrix R represent the weight coefficients of the deviation term and the control increment term, respectively, and are both diagonal matrices.

[0043] The device power scheduling instructions issued by the dynamic scheduling layer include the battery discharge power P. bat,dis and charging power P bat,ch The heat dissipation power Q of the heat storage tank hs,dis and thermal storage power Q hs,ch The power P of a gas internal combustion engine ice0 and the power Q of the heat pumphp0 .

[0044] The control correction variable u(kk) is used to correct the device power scheduling command issued by the scheduling layer, including:

[0045] P ice0,ac (kk)=P ice0 (k)+u1(kk), Q hp0,ac (kk)=Q hp0 (k)+u2(kk),

[0046] In the formula, P ice0,ac Q represents the corrected power output of a gas-fired internal combustion engine. hp0,ac This is the corrected power value for the heat pump.

[0047] The beneficial effects of this invention are as follows:

[0048] Under the framework of a two-layer operation strategy, the real-time control layer based on the multivariate predictive control method can correct the equipment power commands issued by the dynamic scheduling layer in real time. This can reduce the disturbance caused by the prediction deviation of renewable energy power generation and electrical load, and help ensure the accuracy of system power supply and the stability of operation.

[0049] This invention achieves effective control of the thermoelectric coupling link of an integrated energy system through a multivariable predictive control method. It takes into account the characteristics of multiple variables and strong thermoelectric coupling within the system, making the control system more rational. Furthermore, it ensures the unbiased performance of the controller through an augmented state-space model.

[0050] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. Attached Figure Description

[0051] Figure 1 This is a schematic diagram illustrating the principle of the optimized operation method according to an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of the integrated energy system structure according to an embodiment of the present invention.

[0053] Figure 3 This is a schematic diagram of the heating component of the main equipment in the integrated energy system according to an embodiment of the present invention.

[0054] Figure 4 The results represent the operating results on the power side using a traditional single-layer strategy.

[0055] Figure 5 The results are shown for the operation of the power side using the optimized operation method of the embodiments of the present invention.

[0056] Figure 6The results are for the hot water temperature side using a traditional single-layer strategy.

[0057] Figure 7 The results are shown for the operation of the hot water temperature side using the optimized operation method of the present invention. Detailed Implementation

[0058] The specific embodiments of the present invention are described below with reference to the accompanying drawings.

[0059] See Figure 1 This application discloses a two-layer optimization operation method for an integrated energy system based on multivariable predictive control, which optimizes operation based on a dynamic scheduling layer and a real-time control layer, including:

[0060] The dynamic scheduling layer samples the latest operating data of the system, and based on the dynamic characteristic model of the equipment that can be used to describe the heating link of the system equipment, and according to the latest daily forecast data of renewable energy and load demand, with the optimization objectives of minimizing economic costs and minimizing dynamic supply and demand deviation indicators, and with power balance constraints and equipment characteristic constraints as constraints, performs intraday rolling scheduling and provides power scheduling instructions for each equipment in the next scheduling period.

[0061] The real-time control layer takes the gas internal combustion engine, other heat recovery equipment, and the thermoelectric coupling link of the heat pump as the control object. Based on the multivariate predictive control method, it corrects the power dispatching command issued by the dynamic dispatching layer in real time. The corrected power dispatching command is then applied to the system to reduce the impact of prediction deviations of new energy sources and electric loads and model deviations of the dispatching layer on the system operation.

[0062] This application reduces the impact of prediction deviations of new energy sources and electrical loads and model deviations of the scheduling layer on system operation by using a mechanism of real-time correction of power commands issued by the scheduling layer in real time through the real-time control layer, thereby improving the stability and reliability of the integrated energy system's power supply.

[0063] The technical solution of this application is further illustrated below with specific embodiments.

[0064] This embodiment presents a two-level optimization operation method for integrated energy systems based on multivariable predictive control, targeting integrated energy systems with the following structure: Figure 2 As shown, the integrated energy system includes a gas internal combustion engine, a waste heat exchanger, a heat pump, a battery, a thermal storage tank, photovoltaic panels, and a fan. Among these, the photovoltaic panels, fan, gas internal combustion engine, and battery meet the user's electricity demand, while the heat pump and thermal storage tank meet the heat load demand. The structure of the heating components of the gas internal combustion engine and heat pump is shown below. Figure 3 As shown, the gas internal combustion engine adopts a PID closed-loop control system to adjust the fuel quantity to control the output electric power, which in turn affects the output heat power of the downstream waste heat exchanger. The heat pump adopts a PID closed-loop control system to adjust the speed of the heat pump compressor to control the output heat power.

[0065] The operation method of this embodiment includes:

[0066] Step 1: Based on the latest operational data of the dynamic scheduling layer sampling system, and the dynamic characteristic models of the equipment used to describe the heating process of the system, and according to the latest intraday renewable energy and load demand forecast data, with the optimization objectives of minimizing economic costs and dynamic supply-demand deviation indicators, and constrained by power balance constraints and equipment characteristic constraints, intraday rolling scheduling is performed to provide power scheduling instructions for each device in the next scheduling period. Specifically, this includes:

[0067] A rolling time-domain optimization method is adopted, with a scheduling step size of 5 minutes and a rolling time domain size of 4 hours, to construct a dynamic characteristic model that can be used to describe the heating process of a gas internal combustion engine or heat pump, and it is represented in the form of a state-space model:

[0068]

[0069] In the formula, x ep r is a state variable ep Q is the power scheduling command for the device. ep This refers to the actual thermal power output of the equipment; when the equipment is a gas internal combustion engine, r ep For power command P ice0 Q ep The heat power Q of waste heat recovery and utilization from gas internal combustion engines hx When the equipment is a heat pump, r ep For heat pump thermal power command Q hp0 Q ep The output heat power Q of the heat pump hp A ep B ep C ep Let be the coefficient matrix of the state-space model, where kk represents the kk-th sampling time.

[0070] The objective function for designing the dynamic scheduling layer is: min J = w1AIM1 + w2AIM2, where J is the objective function value, AIM1 is the economic indicator, AIM2 is the dynamic supply and demand deviation indicator, including the dynamic deviation indicators of heat power and water supply temperature, and w1 and w2 are the corresponding weights.

[0071] The economic indicators AIM1 specifically include:

[0072]

[0073] In the formula, Δt is the scheduling step size, and N k K represents the number of time-domain steps during the rolling process. fuel The cost of biogas fuel for a gas-fired internal combustion engine, in yuan / kg, K.ice The operating and maintenance cost of a gas-fired internal combustion engine, in yuan / kWh, K. hp The operating and maintenance cost of a heat pump is expressed in yuan / kWh, K. bat The operating and maintenance cost of the battery is expressed in yuan / kWh, K. hs The operating and maintenance cost of the thermal storage tank is RMB / kWh, D ice,f P is the mass flow rate of the fuel consumed. ice P is the output electrical power of a gas internal combustion engine. hp P is the electrical power consumed by the heat pump. bat Q represents the power output value of the battery. hs Let i represent the power output value of the thermal storage tank, and i represent the i-th scheduling time period.

[0074] The dynamic supply-demand deviation index AIM2 specifically includes:

[0075]

[0076] In the formula, Δt kk For the sampling period, N kk The number of samples in each scheduling time period, Q hp (i,j),T hp (i,j), Q hx (i,j) and T hx (i,j) represent the heat pump output power, heat pump outlet hot water temperature, waste heat recovery power of the gas internal combustion engine, and waste heat recovery hot water temperature at the j-th sampling time within the i-th scheduling time period, respectively. T supply G represents the temperature of the hot water after mixing with the heat pump water and the waste heat recovery water. hp G represents the hot water flow rate of the heat pump. hx Q represents the hot water flow rate in the waste heat recovery process. load To meet the user's heat load requirements.

[0077] The constraints for constructing the dynamic scheduling layer include power balance constraints and equipment characteristic constraints:

[0078] The electrical and thermal power balance constraints are:

[0079]

[0080] In the formula, P wind P represents the predicted power of the wind turbine. pv P represents the predicted power of photovoltaic power. load and Q load These are the predicted power values ​​for electrical load and thermal load, respectively.

[0081] Operating characteristic constraints of gas internal combustion engines and other heat recovery processes

[0082]

[0083] In the formula, P ice,min and P ice,max T represents the lower and upper limits of the power output of a gas-fired internal combustion engine, respectively; hx T represents the temperature of the hot water after waste heat recovery. hx,min and T hx,max ΔT represents the lower and upper limits of the hot water temperature for waste heat recovery, respectively; hx,min and ΔT hx,max Indicates the lower and upper limits of the rate of change of hot water temperature; T hx,w1 Indicates the hot water inlet temperature; c pw This is the specific heat capacity of hot water;

[0084] Constraints on the operating characteristics of heat pump equipment:

[0085]

[0086] In the formula, T hp Q represents the outlet hot water temperature of the heat pump. hp,min and Q hp,max Indicates the lower and upper limits of the heat pump's thermal power; T hp,min and T hp,max Indicates the lower and upper limits of the hot water temperature of the heat pump; ΔT hp,min and ΔT hp,max Indicates the lower and upper limits of the rate of change of hot water temperature; T hp,cw1 Indicates the hot water inlet temperature; cop hp This indicates the energy efficiency ratio of the heat pump;

[0087] Battery equipment characteristic constraints:

[0088]

[0089] In the formula, x bat The state of charge of the battery; C bat The capacity of the battery is expressed in kWh; P bat,ch and P bat,dis These are the charging power command value and discharging power command value of the battery, respectively, in kW; η bat,ch and η bat,dis These represent charging efficiency and discharging efficiency, respectively; x bat,min and x bat,max These represent the upper and lower limits of the state of charge; the 0-1 logic variable δ. bat,ch and δ bat,dis These represent the charging and discharging indicators, respectively; that is, when the battery is charging, δ bat,ch =1, when the battery is discharging, δ bat,dis =1; P bat,min and P bat,maxThese are the upper and lower limits of the charging and discharging power, in kW;

[0090] The characteristics of the thermal storage tank equipment are constrained as follows:

[0091]

[0092] In the formula, x hs C is the state quantity of the thermal storage tank. hs Q represents the capacity of the thermal storage tank, in kWh; hs,ch and Q hs,dis These represent the heat storage power and heat release power of the thermal storage tank, respectively, in kW; η hs,ch and η hs,dis These are the heat charging efficiency and heat dissipation efficiency, respectively. hs,min and x hs,max These represent the upper and lower limits of the thermal storage tank's state parameters; the 0-1 logic variable δ hs,ch and δ hs,dis These represent the thermal storage operation state and the thermal release operation state, respectively; Q hs,min and Q hs,max These represent the upper and lower limits of the heat dissipation and charging power, in kW.

[0093] Step Two: The real-time control layer takes the gas-fired internal combustion engine, other heat recovery equipment, and the thermoelectric coupling link of the heat pump as the control objects. Based on the multivariate predictive control method, it corrects the power commands issued by the dynamic scheduling layer in real time to reduce the impact of prediction deviations of new energy sources and electrical loads and model deviations of the scheduling layer on system operation. Specifically, this includes:

[0094] The specific implementation process of the multivariate predictive control method is as follows:

[0095] S1: Obtain the transfer function model of the integrated energy system. Under steady-state conditions, using the internal combustion engine electric power command and heat pump thermal power command as input variables, and the remaining electric power and total hot water supply temperature as output variables, a step response experiment is conducted, with a sampling period Δt set. kk The data is preprocessed to identify the transfer function model, which is then converted into a discrete state-space model.

[0096]

[0097] In the formula, y(kk) is the system output, and x d (kk) is the state variable at time kk, u(kk) is the control variable at time kk, and A d B d C d These are the coefficient matrices corresponding to the state space model of the controlled object;

[0098] S2: Obtain the augmented state-space model of the discrete state-space model:

[0099]

[0100]

[0101] In the formula, Δx d (kk) is the state change at time kk, x(kk) is the amplified state variable, Δu(kk) is the control change at time kk, and A, B, and C are the coefficient matrices after state amplification. p×p It is a p×p dimensional identity matrix, O d It is a zero matrix.

[0102] S3: Set the controller parameters, including the prediction time domain P, the control time domain M, the error weight matrix Q, and the control weight matrix R. The prediction time domain P should cover the main dynamic characteristics of the system, and the control time domain M should not be greater than the prediction time domain P. The error weight matrix Q and the control weight matrix R represent the weight coefficients of the deviation term and the control increment term, respectively, and are both diagonal matrices, expressed as: Q = diag(q1,…,q P R = diag(r1,…,r) M );

[0103] S4: Based on the current system output y(kk), the Kalman filter algorithm is used to estimate the current state of the system:

[0104] Perform state prediction in one step: It is the change in control quantity at time kk-1. It is the state estimate at time kk-1. It is the state estimate at time kk from time kk-1;

[0105] Calculate the covariance matrix of the one-step state prediction: P kk / kk-1 =AP kk-1 A T +Q kk-1 P kk-1 and P kk / kk-1 These are the state covariance matrices at time kk-1 and at time kk, calculated recursively, respectively, Q. kk-1 It is the process noise covariance matrix at time kk-1;

[0106] Calculate the filter gain matrix: K kk =P kk / kk-1 C T (CP kk / kk-1 C T +R kk ) -1 K kk It is the filter gain matrix, R kk It is the covariance matrix for measuring noise;

[0107] The state is corrected based on the current measurement value:

[0108] Calculate the covariance matrix of the state estimate: P kk =(IK kk C)P kk / kk-1 In the formula, I is the identity matrix, and P kk It is the state covariance matrix at time kk.

[0109] S5: Establish prediction equation (10) to predict the output Y(kk) of the system at P future times:

[0110] Y(kk)=Fx(kk)+ΩΔU(kk) (10)

[0111] In the formula, ΔU(kk) represents the future sequence of changes in control variables, and F and Ω are the coefficient matrices corresponding to the prediction equation.

[0112] in,

[0113]

[0114] S6: Build Performance Metrics Solve for ΔU(kk), and select the optimal control change Δu(kk) at time kk as the output to ensure that the controlled variable can track the setpoint, while the fluctuation of the control variable is not too drastic. Where Y... r (kk)=[y r (kk+1),y r (kk+2),…,y r (kk+P)] T y represents a sequence of setpoints at future times. r The set value for a single moment. These are the quadratic forms when the weight matrix takes Q and R, respectively;

[0115] S7: Calculate the control correction amount u(kk) = u(kk-1) + Δu(kk), u(kk) = [u1(kk), u2(kk)] T u1(kk) and u2(kk) are the control correction values ​​for the power scheduling commands of the gas internal combustion engine and the heat pump, respectively.

[0116] S8: Modify the power scheduling command issued by the scheduling layer by the control correction amount u(kk). The modified power command is applied to the system. At the next moment, sample and update the system's output variables (remaining power and total hot water temperature). Repeat steps S4 to S8.

[0117] Specifically, the device power scheduling instructions issued by the dynamic scheduling layer include the battery discharge power P.bat,dis and charging power P bat,ch The heat dissipation power Q of the heat storage tank hs,dis and thermal storage power Q hs,ch The power P of a gas internal combustion engine ice0 and the power Q of the heat pump hp0 .

[0118] Among them, the control correction amount u(kk) will affect the power P of the gas internal combustion engine issued by the dispatching layer. ice0 and the power Q of the heat pump hp0 The corrected calculation formula is as follows:

[0119] P ice0,ac (kk)=P ice0 (k)+u1(kk), Q hp0,ac (kk)=Q hp0 (k)+u2(kk)

[0120] In the formula, P ice0,ac Q represents the corrected power output of a gas-fired internal combustion engine. hp0,ac This is the corrected power value for the heat pump.

[0121] The effectiveness of the optimized operation method of this embodiment will be further illustrated by specific calculation examples below.

[0122] right Figure 2 The integrated energy system shown was subjected to optimized operation simulation, with simulation parameters set as shown in Table 1:

[0123] Table 1 Simulation Parameters

[0124] parameter numerical values unit Sampling period 5 Second Error weight matrix <![CDATA[I2]]> - Control Matrix <![CDATA[5I2]]> - Prediction Time Domain 200 - Control Time Domain 1 - <![CDATA[K fuel ]]> 2.1 Yuan / kg <![CDATA[K hp ]]> 0.08 Yuan / kWh <![CDATA[K bat ]]> 0.05 Yuan / kWh <![CDATA[K hs ]]> 0.01 Yuan / kWh <![CDATA[K ice ]]> 0.11 Yuan / kWh <![CDATA[w1]]> 1 - <![CDATA[w2]]> 3.6 -

[0125] The two-layer optimization operation method implemented in this paper is compared with the conventional single-layer operation strategy scheme that only has a dynamic scheduling layer. In this method, the dynamic scheduling layer is solved using the Gurobi solver, and the real-time control layer is solved using a quadratic programming method. The calculation results on the power side are as follows: Figure 4 , Figure 5 As shown, by comparison Figure 4 and Figure 5 As can be seen, the dual-layer optimized operation method proposed in this embodiment, through the real-time correction method of the control layer, can promptly correct equipment commands, thus better meeting the actual electrical load demand and exhibiting a smaller deviation between power supply and demand compared to the single-layer operation strategy. The calculation structures for the hot water temperature side are as follows: Figure 6 and Figure 7 As shown, by comparison Figure 6 and Figure 7 It can be seen that the two-layer optimization operation method can ensure that the hot water supply temperature better tracks the changes in the set value through real-time correction of the control layer, and reduce the impact of scheduling model deviation on the system heating.

[0126] It will be understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A two-level optimization operation method for an integrated energy system based on multivariable predictive control, characterized in that, Operational optimization is based on a dynamic scheduling layer and a real-time control layer, including: The dynamic scheduling layer samples the latest operating data of the system, and based on the dynamic characteristic model of the equipment that can be used to describe the heating link of the system equipment, and according to the latest daily forecast data of renewable energy and load demand, with the optimization objectives of minimizing economic costs and minimizing dynamic supply and demand deviation indicators, and with power balance constraints and equipment characteristic constraints as constraints, performs intraday rolling scheduling and provides power scheduling instructions for each equipment in the next scheduling period. The real-time control layer takes the gas internal combustion engine, other heat recovery equipment, and the thermoelectric coupling link of the heat pump as the control object. Based on the multivariate predictive control method, it corrects the power scheduling command issued by the dynamic scheduling layer in real time and uses the corrected power scheduling command to act on the system.

2. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 1, characterized in that, The dynamic characteristic model described above can be used to describe the heating process of a gas internal combustion engine or heat pump, and is represented in state-space form as follows: In the formula, x ep r is a state variable ep Q is the power scheduling command for the device. ep A represents the actual thermal output power of the equipment. ep B ep C ep Let be the coefficient matrix, and kk represent the kk-th sampling time.

3. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 1, characterized in that, The function expression of the optimization objective is: minJ=w1AIM1+w2AIM2, where J is the objective function value, AIM1 is the economic index, AIM2 is the dynamic supply and demand deviation index, including the dynamic deviation index of heat power and the dynamic deviation index of water supply temperature, and w1 and w2 are the corresponding weights. The economic indicators AIM1 specifically include: In the formula, Δt k For the scheduling step size, N k K represents the number of time-domain steps during the rolling process. fuel For the biogas fuel cost of a gas-fired internal combustion engine, K ice For the operation and maintenance costs of gas internal combustion engines, K hp For the operation and maintenance costs of heat pumps, K bat For the operation and maintenance costs of the battery, K hs For the operation and maintenance costs of the thermal storage tank, D ice,f P is the mass flow rate of the fuel consumed. ice P is the output electrical power of a gas internal combustion engine. hp P is the electrical power consumed by the heat pump. bat Q represents the power output value of the battery. hs represents the power output value of the thermal storage tank; i represents the i-th scheduling time period; The dynamic supply-demand deviation index AIM2 specifically includes: In the formula, Δt kk For the sampling period, N kk The number of samples in each scheduling time period, Q hp (i,j),T hp (i,j), Q hx (i,j) and T hx (i,j) represent the heat pump output power, heat pump outlet hot water temperature, waste heat recovery power of the gas internal combustion engine, and waste heat recovery hot water temperature at the j-th sampling time within the i-th scheduling time period, respectively. T supply G represents the temperature of the hot water after mixing with the heat pump water and the waste heat recovery water. hp G represents the hot water flow rate of the heat pump. hx Q represents the hot water flow rate in the waste heat recovery process. load To meet the user's heat load requirements.

4. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 1, characterized in that, The power balance constraints include system electrical power and thermal power balance constraints; the equipment characteristic constraints include operating characteristic constraints of the gas internal combustion engine and other heat recovery links, operating characteristic constraints of heat pump equipment, characteristic constraints of battery equipment, and characteristic constraints of thermal storage tank equipment.

5. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 1, characterized in that, The multivariate predictive control method includes: S1: Select input and output variables to conduct an open-loop step response experiment and identify the transfer function model of the integrated energy system. The output variables are the remaining electric power and the total hot water supply temperature, and the input variables are the internal combustion engine electric power command adjustment amount and the heat pump heat power command adjustment amount. S2: Define the augmented state-space model of the transfer function model: In the formula, y(kk) is the system output, and Δx d (kk) is the change in state variable at time kk, x(kk) is the amplified state variable, Δu(kk) is the change in control variable at time kk, and A d B d C d These are the coefficient matrices corresponding to the state-space model of the controlled object, where A, B, and C are the coefficient matrices after state augmentation, and I... p×p It is a p×p dimensional identity matrix, O d It is a zero matrix; S3: Set controller parameters, including prediction time domain P, control time domain M, error weight matrix Q, and control weight matrix R; S4: Estimate the current state of the system based on the current system output y(kk); S5: Establish the prediction equation Y(kk)=Fx(kk)+ΩΔU(kk) to predict the output Y(kk) of the system at P future times, where ΔU(kk) represents the future sequence of control variable changes, and F and Ω are the coefficient matrices corresponding to the prediction equation; in, S6: Build Performance Metrics Solve for ΔU(kk), and select the optimal control change Δu(kk) at time kk as the output to ensure that the controlled variable can track the setpoint, while the fluctuation of the control variable is not too drastic. Where Y... r (kk)=[y r (kk+1),y r (kk+2),…,y r (kk+P)] T y represents a sequence of setpoints at future times. r The set value for a single moment. These are the quadratic forms when the weight matrix takes Q and R, respectively; S7: Calculate the control correction amount u(kk) = u(kk-1) + Δu(kk), u(kk) = [u1(kk), u2(kk)] T u1(kk) and u2(kk) are the control correction values ​​for the power scheduling commands of the gas internal combustion engine and the heat pump, respectively. S8: Modify the power scheduling command issued by the scheduling layer by the control correction amount u(kk). The modified power command is applied to the system. The system output information is sampled and updated at the next moment. Repeat steps S4 to S8.

6. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 5, characterized in that, The estimation of the system's current state using the Kalman filter algorithm includes: Perform state prediction in one step: Δu(kk-1) is the change in the control quantity at time kk-1. It is the state estimate at time kk-1. It is the state estimate at time kk from time kk-1; Calculate the covariance matrix of the one-step state prediction: P kk / kk-1 =AP kk-1 A T +Q kk-1 P kk-1 and P kk / kk-1 These are the state covariance matrices at time kk-1 and at time kk, calculated recursively, respectively, Q. kk-1 It is the process noise covariance matrix at time kk-1; Calculate the filter gain matrix: K kk =P kk / kk-1 C T (CP kk / kk-1 C T +R kk ) -1 K kk It is the filter gain matrix, R kk It is the covariance matrix for measuring noise; The state is corrected based on the current measurement value: Calculate the covariance matrix of the state estimate: P kk =(IK kk C)P kk / kk-1 I is the identity matrix, P kk It is the state covariance matrix at time kk.

7. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 5, characterized in that, The selection of the prediction time domain P should cover the main dynamic characteristics of the system, and the control time domain M should not be greater than the prediction time domain P; the error weight matrix Q and the control weight matrix R represent the weight coefficients of the deviation term and the control increment term, respectively, and are both diagonal matrices.

8. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 5, characterized in that, The device power scheduling instructions issued by the dynamic scheduling layer include the battery discharge power P. bat,dis and charging power P bat,ch The heat dissipation power Q of the heat storage tank hs,dis and thermal storage power Q hs,ch The power P of a gas internal combustion engine ice0 and the power Q of the heat pump hp0 .

9. The two-level optimization operation method for integrated energy systems based on multivariable predictive control according to claim 8, characterized in that, The control correction variable u(kk) is used to correct the device power scheduling command issued by the scheduling layer, including: P ice0,ac (kk)=P ice0 (k)+u1(kk),Q hp0,ac (kk)=Q hp0 (k)+u2(kk), In the formula, P ice0,ac Q represents the corrected power output of a gas-fired internal combustion engine. hp0,ac This is the corrected power value for the heat pump.

Images