Coordinated Prediction Control Method for Natural Gas Cogeneration System with Non-constant Prediction Step Size

Abstract

The present invention discloses a coordinated prediction control method for a natural gas combined heat and power system with a non-constant prediction step length, which relates to the technical fields of natural gas combined heat and power systems and automatic control technologies. It solves the technical problem that the thermoelectric coordinated control performance of the system is not flexible and fast enough due to the significant time-scale differences between different energy conversion processes of the system. The key technical solution is to construct a quantitative relationship between the prediction step length and the number of prediction steps, so as to comprehensively characterize the dynamic characteristics of the system's heat and power processes with fewer prediction steps, and can effectively improve the flexibility and rapidity of the system's thermoelectric coordinated tracking control.

Description

Technical Field

[0001] This application relates to the technical fields of natural gas combined heat and power systems and automatic control technologies, specifically to the coordinated thermoelectric control technology of natural gas combined heat and power systems, and particularly to a coordinated predictive control method for natural gas combined heat and power systems with a non-constant prediction step size. Background Art

[0002] Facing the contradiction between the growing energy demand with the continuous development of social economy and the increasingly severe environmental pollution problems, the optimization of the energy system structure, energy conservation and emission reduction, flexible transformation, and improvement of energy supply security have become hot issues in the energy field of our country. Among them, promoting the further extensive implementation of the "three-in-one linkage" comprehensive transformation including energy-saving transformation, heating transformation, and flexibility transformation provides a basis for coal-fired power units to serve the construction of a low-carbon power system and achieve clean and efficient power supply of thermal power.

[0003] Through the reasonable matching and coupled operation with coal-fired power units, the natural gas combined heat and power system utilizes the energy cascade utilization of its energy-saving essence and the emission reduction essence of clean energy substitution, which is conducive to the improvement of the energy-saving and emission-reduction and flexible transformation capabilities of coal-fired power units. It is one of the important technical paths for coal-fired power units to carry out the "three-in-one linkage" comprehensive transformation. Therefore, it is urgent to study advanced operation and control strategies suitable for it to ensure the stable, flexible, and rapid energy supply of the system and demonstrate its system advantages.

[0004] At present, combined heat and power systems usually adopt the operation strategies of "electricity-determined heat" or "heat-determined electricity", lacking the analysis and utilization of the coupling characteristics of the system, resulting in the unexploited operation potential of the system and being difficult to meet the flexible operation requirements at the present stage. As a class of advanced model-based control algorithms, model predictive control has been developed and applied to the coordinated control of combined heat and power systems due to its good control performance for multi-variable systems with various complex constraints and uncertainties. Some scholars designed an operation strategy based on centralized model predictive control for a natural gas combined heat and power system with a micro gas turbine-heat pump as the core, realizing the balance between the heat and electricity supply and demand of the system. However, conventional model predictive control is difficult to overcome the adverse effects brought by the significant time-scale differences between the heat and electricity processes in combined heat and power systems, resulting in the system being unable to obtain an ideal thermoelectric coordinated control effect. Summary of the Invention

[0005] This application provides a coordinated predictive control method for natural gas combined heat and power systems with a non-constant prediction step size. Its technical purpose is to achieve a comprehensive characterization of the dynamic characteristics of the heat and electricity processes of the system by constructing a quantitative relationship between the prediction step size and the number of prediction steps, and to achieve the best balance between the prediction accuracy of the fast process and the prediction range of the slow process of the system, with a view to obtaining a good thermoelectric coordinated control effect of the system, providing an effective means and a new perspective for the flexible operation of natural gas combined heat and power systems.

[0006] The above technical object of the present application is achieved by the following technical solutions:

[0007] A coordinated prediction control method for a natural gas combined heat and power supply system with a non-constant prediction step length, comprising:

[0008] Step 1: Conduct an open-loop step experiment on the mechanism model of the natural gas combined heat and power supply system to obtain a dynamic characteristic experiment curve; wherein, the natural gas combined heat and power supply system includes a micro gas turbine and a heat pump; the control quantities include the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor, and the controlled quantities include the net power generation and the hot water supply temperature;

[0009] Step 2: According to the data corresponding to the dynamic characteristic experiment curve, adopt the subspace identification method to identify and obtain a linear time-invariant state space model characterizing the dynamic characteristics of the natural gas combined heat and power supply system;

[0010] Step 3: Construct a quantitative relationship between the prediction step length and the number of prediction steps, and discretize the linear time-invariant state space model multiple times according to the quantitative relationship to obtain discrete state space models under different sampling periods;

[0011] Step 4: Successively amplify the discrete state space model into a discrete state space increment model, and then derive the prediction model from the discrete state space increment model;

[0012] Step 5: Define the control objective function of the finite-time domain optimization problem, and design the corresponding heat-electricity coordinated controller for the natural gas combined heat and power supply system in combination with the prediction model, and realize the heat-electricity coordinated tracking control of the system through the heat-electricity coordinated controller.

[0013] The beneficial effects of the present application are as follows:

[0014] (1) The heat-electricity coordinated prediction control method with a non-constant prediction step length described in the present application can effectively improve the flexibility and rapidity of the system's heat-electricity coordinated control by constructing a quantitative relationship between the prediction step length and the number of prediction steps.

[0015] (2) The prediction model construction method described in the present application comprehensively depicts the dynamic characteristics of the system's heat and electricity processes with fewer prediction steps, achieving the best balance between the prediction accuracy of the fast process and the prediction range of the slow process of the system. Description of the Drawings

[0016] Figure 1 It is a schematic structural diagram of the natural gas combined heat and power supply system in the specific embodiment of the present application;

[0017] Figure 2 It is a schematic structural diagram of the control structure in the specific embodiment of the present application;

[0018] Figure 3 This is a schematic diagram for comparing control algorithms in the specific implementation manner of this application;

[0019] Figure 4 This is a schematic diagram of the system open-loop step response curve in the specific implementation manner of this application;

[0020] Figure 5 This is a schematic diagram of the controlled variable result curve in the specific implementation manner of this application;

[0021] Figure 6 This is a schematic diagram of the control variable result curve in the specific implementation manner of this application. Specific implementation manner

[0022] Next, the technical solution of this application will be described in detail with reference to the accompanying drawings.

[0023] In the specific embodiment of this application, the natural gas combined heat and power system is as Figure 1 shown. In the following description, the abbreviation "system" always refers to the "natural gas combined heat and power system". As Figure 1 shown, the main equipment of this system includes a micro gas turbine (hereinafter referred to as "micro turbine") and an air source heat pump (hereinafter referred to as "heat pump"); other auxiliary equipment includes heat exchangers, busbars, pipelines, valves, circulating water pumps, controllers, and related electrical equipment. Among them, the micro turbine is a micro gas turbine operating in the combined heat and power mode, and the exhaust waste heat is used by the heat exchanger to produce hot water and send it to the water supply header; the hot water produced by the heat pump is mixed with the hot water produced by the waste heat of the micro turbine in the same water supply header, and then uniformly supplies hot water externally; all the electrical equipment of the system is connected to the same busbar and supplies power externally, without an external power grid.

[0024] Therefore, in order to achieve the control goal of quickly and smoothly tracking the thermal and electrical demands of users, this application proposes a coordinated predictive control method for a natural gas combined heat and power system with a non-constant prediction step size. More specifically, the control structure of the natural gas combined heat and power system described in this embodiment includes a thermoelectric coordinated controller, and the structure of this control structure is as Figure 2 shown. The physical object controlled by the thermoelectric coordinated controller is the natural gas combined heat and power system, including a micro gas turbine and a heat pump. Among them, the thermoelectric coordinated controller is designed by a model predictive control algorithm with a non-constant prediction step size, and by adjusting the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor, the net power generation and the hot water supply temperature of the natural gas combined heat and power system are changed.

[0025] The above-mentioned coordinated predictive control method for a natural gas combined heat and power system with a non-constant prediction step size specifically includes the following steps:

[0026] Step 1: Conduct an open-loop step experiment on the mechanism model of the natural gas combined heat and power system to obtain the dynamic characteristic experiment curve; wherein, the natural gas combined heat and power system includes a micro gas turbine and a heat pump; the control variables include the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor, and the controlled variables include the net power generation and the hot water supply temperature.

[0027] Specifically, the calculation formula for the net power generation is expressed as:

[0028] N e = N MGT - N ASHP ;

[0029] Wherein, N e represents the net power generation; N MGT represents the power generation of the micro gas turbine; N ASHP represents the power consumption of the heat pump.

[0030] Step 2: According to the data corresponding to the dynamic characteristic experiment curve, adopt the subspace identification method to identify and obtain a linear time-invariant state space model characterizing the dynamic characteristics of the natural gas combined heat and power system.

[0031] Specifically, the linear time-invariant state space model is expressed as:

[0032]

[0033] Wherein, x c represents the state quantity of the system, u = [m f , r c T represents the control quantity of the system, y = [N e , T feed T represents the controlled quantity of the system; m f represents the fuel mass flow rate of the micro gas turbine, r c represents the rotational speed of the heat pump compressor, T feed represents the hot water supply temperature; A c , B c , C c respectively represent the corresponding coefficient matrices.

[0034] Step 3: Construct a quantitative relationship between the prediction step length and the number of prediction steps, and discretize the linear time-invariant state space model multiple times according to the quantitative relationship to obtain discrete state space models under different sampling periods.

[0035] Specifically, construct a quantitative relationship between the prediction step length and the number of prediction steps, that is, construct a quantitative relationship between the model discrete sampling period and the prediction step length of the prediction model recurrence process. The model discrete sampling period is the number of prediction steps, then this quantitative relationship is expressed as:​​

[0036] Δt i = 1·2 i-1 。

[0037] The discrete state - space model is expressed as:

[0038]

[0039] where, x d (k) represents the state quantity of the system at time k, u(k) represents the control quantity of the system at time k, and y(k) represents the controlled quantity of the system at time k; Δt i represents the discrete sampling period of the model, and i represents the prediction step length of the prediction model recursion process.

[0040] Step 4: Amplify the discrete state - space model into a discrete state - space increment model in sequence, and then derive the prediction model from the discrete state - space increment model to comprehensively characterize the dynamic characteristics of the system's thermal and electrical processes.

[0041] Specifically, the discrete state - space increment model is expressed as:

[0042]

[0043] where, Δu(k)=u(k) - u(k - 1); The value of x(k) is estimated by the Kalman filter in each control period; Δx d (k)=x d (k)-x d (k - 1); I represents the corresponding identity matrix, and O represents the corresponding zero matrix.

[0044] Define the prediction step number P and the control step number M, and M≤P; Assume that the control increment remains unchanged within the prediction step number P and outside the control step number M. Then, by recursively calculating P steps of the discrete state - space increment model under different sampling periods, the prediction model of the system within the future prediction time domain N P is obtained. Then the prediction model is expressed as:

[0045] Y(k)=S x ·x(k)+S u ·ΔU(k);

[0046]

[0047]

[0048] where, N P =∑Δt i 。

[0049] Step 5: Define the control objective function for the finite-horizon optimization problem, and design the thermoelectric coordination controller corresponding to the natural gas combined heat and power system in combination with the prediction model, and realize the thermoelectric coordination tracking control of the system through the thermoelectric coordination controller.

[0050] Specifically, the control objective function is expressed as:

[0051]

[0052] s.t.u min ≤u(k + j)≤u max ,Δu min ≤Δu(k + j)≤Δu max ;

[0053] where y set (i) represents the set value of the controlled variable; Q represents the output weight matrix; R represents the control weight matrix; u min represents the lower limit of the control quantity, u max represents the upper limit of the control quantity; Δu min represents the lower limit of the control increment; Δu max represents the upper limit of the control increment.

[0054] Design the thermoelectric coordination controller through the above model predictive control algorithm with non-constant prediction step length. Then the physical object controlled by the thermoelectric coordination controller is the natural gas combined heat and power system; this thermoelectric coordination controller adjusts the net power generation and the hot water supply temperature of the natural gas combined heat and power system by regulating the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor.

[0055] Specifically, the basic implementation process of realizing the thermoelectric coordination tracking control of the system through the thermoelectric coordination controller is as follows: at each sampling time k, the thermoelectric coordination controller estimates the system state x(k) at the current time through the Kalman filter; then obtains the future P-step output prediction sequence according to the prediction model; then solves the given finite-horizon optimization problem to determine the optimal control sequence ΔU(k); finally, implements the first group of elements of this optimal control sequence on the natural gas combined heat and power system to adjust the net power generation and the hot water supply temperature of the natural gas combined heat and power system.

[0056] It should be noted that when Δt i = const, the model predictive control with non-constant prediction step length degenerates into the conventional model predictive control with constant prediction step length.

[0057] The following takes an actual system as an example for calculation, and the specific process is as follows:

[0058] First, based on the established mechanism model of the natural gas combined heat and power system, an open-loop step experiment is conducted on it to obtain the corresponding dynamic characteristic experiment curves, such as Figure 4 as shown. Subsequently, according to the corresponding dynamic characteristic experiment data, a linear time-invariant state space model can be obtained by using the subspace identification method. Then, a quantitative relationship between the prediction step length and the number of prediction steps is constructed. Based on the constructed quantitative relationship between the prediction step length and the number of prediction steps, the linear time-invariant state space model shown is discretized multiple times to obtain discrete state space models under different sampling periods. After appropriate transformation, the discrete state space model is augmented to obtain a discrete state space increment model.

[0059] Define the number of prediction steps P and the number of control steps M, and M ≤ P. Assume that the control increment remains unchanged within the number of prediction steps P and outside the number of control steps M. Then, by recursively calculating P steps in combination with the discrete state space increment models under different sampling periods, a prediction model of the system within the future prediction time domain N P is obtained.

[0060] Finally, define the control objective function of the finite time domain optimization problem, construct the finite time domain optimization problem of the thermoelectric coordinated control of the natural gas combined heat and power system based on the prediction model and the quantitative relationship, optimize and solve the finite time domain optimization problem, and control the natural gas combined heat and power system.

[0061] To verify the superiority of the model predictive control strategy with non-constant prediction step length (Non-constant Prediction-step MPC, NMPC) proposed in this embodiment, it is compared with two conventional model predictive control strategies (MPC1, MPC2) and the conventional PI control strategy (PID). The comparison schematic diagram is as Figure 3 shown. In the conventional PI control strategy, the parameters of each PI controller are shown in Table 1. In the three model predictive control strategies, the parameters of the MPC controller are shown in Table 2.

[0062] The simulation results are respectively as Figures 5 to 6 shown, where Figure 5 is the schematic diagram of the result curve of the controlled quantity in the simulation experiment, Figure 6 is the schematic diagram of the result curve of the control quantity in the simulation experiment. Further, the integrated absolute error (IAE) index of the controlled quantity under different control strategies and the simulation calculation time under different MPC control strategies are calculated respectively. The calculation results are shown in Table 3.

[0063] As Figure 5 shown, the results show that under the NMPC control strategy proposed in this paper, the net power generation of the system (N e ) and the hot water supply temperature (T feed) It can track its corresponding set value quickly and stably, and has a small overshoot during the adjustment process. At the same time, in the dynamic adjustment process of the NMPC control strategy proposed in this paper, by formulating a better combination of control quantities (such as Figure 6 shown), it can effectively reduce the adverse effects brought by the thermoelectric coupling characteristics of the system to the adjustment process. When the set value of the hot water supply temperature changes, the fluctuation of its net power generation (N e ) is significantly smaller than that of the conventional PI control strategy.

[0064] Table 1 - Controller parameters of the conventional PI control strategy

[0065] Controller type <![CDATA[m f -N e > <![CDATA[r c -T feed > Proportional gain (P) 6.672e-06 1.772 Integral gain (I) 1.239e-06 0.1752 Sampling period (s) 1 1

[0066] Table 2 - Controller parameters in three MPC control strategies

[0067]

[0068] Table 3 - Calculation results of IAE index and simulation calculation time

[0069] IAE <![CDATA[N e > <![CDATA[T feed > Simulation calculation time (s) PID 1145.58 59.89 / MPC1 135.09 352.74 3.0914 MPC2 195.38 37.08 4.6051 NMPC 94.94 52.35 3.1540

[0070] The calculation results in Table 3 also clearly and quantitatively verify the characteristics and advantages of the NMPC control strategy proposed in this paper:

[0071] 1) Since the conventional PI control strategy cannot fully consider the interaction between systems, the control of the electrical side index is affected by the slow process of the system (thermal side characteristics), and its control effect is significantly inferior to that of the MPC control strategy (the IAE value of its net power generation (N e ) is much larger than the IAE value of the net power generation (N e ) under the MPC control strategy);

[0072] 2) Compared with the other two conventional MPC control strategies, the NMPC control strategy proposed in this paper effectively realizes the best trade-off between the prediction accuracy of the fast process and the prediction range of the slow process of the system by constructing a quantitative relationship between the prediction step length and the number of prediction steps, and obtains the optimal thermoelectric coordination comprehensive control performance: on the one hand, the NMPC control strategy proposed in this paper comprehensively depicts the dynamic characteristics of the fast process (electrical side) and slow process (thermal side) of the system through a long prediction time domain, effectively reducing the adverse effects of the slow process of the system on the control of the fast process, and obtaining the optimal control effect on the electrical side and an almost optimal control effect on the thermal side; on the other hand, the NMPC control strategy proposed in this paper has a small number of prediction steps, which is beneficial to improving the calculation efficiency, so its calculation time is close to the optimal.

[0073] In summary, the coordinated prediction control method for natural gas combined heat and power systems with non-constant prediction step lengths proposed in this application has more excellent thermoelectric coordination comprehensive control performance.

[0074] The above are exemplary embodiments of the present application, and the protection scope of the present application is defined by the claims and their equivalents.

Claims

1. A coordinated prediction control method for a natural gas combined heat and power system with a non-constant prediction step size, characterized in that, Including: Step 1: Conduct an open-loop step experiment on the mechanism model of the natural gas combined heat and power system to obtain the dynamic characteristic experiment curve. Among them, the natural gas combined heat and power system includes a micro gas turbine and a heat pump. The control variables include the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor, and the controlled variables include the net power generation and the hot water supply temperature. Step 2: According to the data corresponding to the dynamic characteristic experiment curve, use the subspace identification method to identify and obtain a linear time-invariant state space model representing the dynamic characteristics of the natural gas combined heat and power system. Step 3: Construct a quantitative relationship between the prediction step length and the number of prediction steps, and discretize the linear time-invariant state space model multiple times according to the quantitative relationship to obtain discrete state space models under different sampling periods. Step 4: Successively expand the discrete state space model into a discrete state space increment model, and then derive the prediction model from the discrete state space increment model. Step 5: Define the control objective function of the finite-time domain optimization problem, and design the corresponding heat-electricity coordination controller for the natural gas combined heat and power system in combination with the prediction model, and realize the heat-electricity coordination tracking control of the system through the heat-electricity coordination controller.

2. The method according to claim 1, characterized in that, In the said Step 1, the calculation formula of the net power generation is expressed as: N e = N MGT -N ASHP ; Among them, N e represents the net power generation; N MGT represents the power generation of the micro gas turbine; N ASHP represents the power consumption of the heat pump.

3. The method according to claim 1, wherein In the said Step 2, the linear time-invariant state space model is expressed as: where, x c represents the state quantity of the system, u = [m f , r c T represents the control quantity of the system, y = [N e , T feed T represents the controlled quantity of the system; m f represents the mass flow rate of the fuel quantity of the micro gas turbine, r c represents the rotational speed of the heat pump compressor, T feed represents the hot water supply temperature; A c , B c , C c represent the corresponding coefficient matrices respectively.​​ 4. The method according to claim 3, wherein In the said Step 3, the discrete state space model is expressed as: where x d (k) represents the state quantity of the system at time k, u(k) represents the control quantity of the system at time k, and y(k) represents the controlled quantity of the system at time k; Δt i represents the discrete sampling period of the model, and i represents the prediction step length of the prediction model recursion process; The discrete state space increment model is expressed as: where, Δu(k) = u(k) - u(k - 1); The value of x(k) is estimated by a Kalman filter in each control period; Δx d (k) = x d (k) - x d (k - 1); I represents the corresponding identity matrix, and O represents the corresponding zero matrix; Define the prediction step P and the control step M, where M ≤ P; assume that the control increment remains unchanged within the prediction step P and outside the control step M, then recursively calculate P steps by combining the discrete state space increment models under different sampling periods to obtain the prediction model of the system within the future prediction time domain N P The prediction model is expressed as: Y(k) = S x ·x(k) + S u ·ΔU(k); Among them, N P = ∑Δt i .

5. The method according to claim 4, wherein In the said Step 5, the control objective function is expressed as: s.t.u min ≤u(k + j)≤u max ,Δu min ≤Δu(k + j)≤Δu max ; where y set (i) represents the setpoint of the controlled variable; Q represents the output weight matrix; R represents the control weight matrix; u min represents the lower limit of the control variable, u max represents the upper limit of the control variable; Δu min represents the lower limit of the control increment; Δu max represents the upper limit of the control increment.

6. The method according to claim 5, wherein In the said Step 3, construct a quantitative relationship between the prediction step length and the number of prediction steps, that is, construct a quantitative relationship between the model discrete sampling period and the prediction step length in the prediction model recurrence process, and this quantitative relationship is expressed as: Δt i = 1·2 i-1 .

7. The method according to any one of claims 1-6, characterized in that The natural gas combined heat and power system further includes a heat exchanger, a bus, pipelines, valves, a circulating water pump, a controller, and related electrical equipment; the control structure of the natural gas combined heat and power system includes a heat-electricity coordination controller. The micro gas turbine is a micro gas turbine operating in the combined heat and power mode, and its exhaust waste heat is used by the heat exchanger to produce hot water and send it to the water supply header; the heat pump is an air source heat pump, and the hot water produced by it is mixed with the hot water produced by the waste heat of the micro gas turbine in the same water supply header and then uniformly supplies hot water to the outside. The electrical equipment is all connected to the same bus and supplies power to the outside uniformly without an external power grid.

8. The method according to any one of claims 1 to 6, characterized in that The physical object controlled by the heat-electricity coordination controller is the natural gas combined heat and power system, and the heat-electricity coordination controller adjusts the fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor to adjust the net power generation and the hot water supply temperature of the natural gas combined heat and power system.

9. The method according to claim 8, characterized in that, In the said Step 5, realizing the heat-electricity coordination tracking control of the system through the heat-electricity coordination controller includes: In each control period, use the Kalman filter to estimate the state quantity of the natural gas combined heat and power system, and then combine the prediction model to obtain the change trajectory of the controlled quantity in the future prediction time domain. Send the change trajectory and reference trajectory of the controlled quantity within the future prediction time domain to the thermoelectric coordination controller, solve the given finite-time domain optimization problem, and obtain the optimal control sequence for the current control period, that is, the optimal fuel mass flow rate of the micro gas turbine and the rotational speed of the heat pump compressor. Implement the first set of elements of this optimal control sequence on the natural gas combined heat and power system to adjust the net power generation and hot water supply temperature of the natural gas combined heat and power system.

Images