Fire storage collaborative peak regulation process coordination control strategy based on iterative learning and active disturbance rejection control

Abstract

The invention relates to an iterative learning and active-disturbance-rejection-control-based coordinated control strategy for a fire storage coordinated peak regulation process. The invention discloses an iterative learning and active-disturbance-rejection-control-based coordinated control strategy for the fire storage coordinated peak regulation process. The method comprises the following steps: firstly, according to molten salt heat storage-heat release and steam-water dynamic characteristics, simplifying a coupled molten salt heat storage and supercritical unit collaborative peak regulation process into a five-input five-output coupled system; secondly, integrating iterative learning and active disturbance rejection control to construct a coordination control technology to promote multivariable stationary response in a coordination peak regulation process; and finally, verifying the effectiveness of the fused salt heat storage and supercritical unit collaborative peak regulation process coupling control technology based on a simulation platform. According to the coordinated control technology for the coordinated peak regulation process, safe and stable operation of the fused salt heat storage device and the supercritical unit can be achieved at the same time, the deep peak regulation capacity and economic benefits are improved, and construction of a novel power system in China can be accelerated and promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of coordinated control of the coordinated peak-shaving process of integrated molten salt heat storage and thermal power generating units, and more specifically, to a coordinated control strategy for the coordinated peak-shaving process of coupled molten salt heat storage and supercritical units. Background Art

[0002] With the launch and continuous deployment of the "Shagohuang" new energy base plan, the proportion of new energy power in my country's power system is increasing. However, the intermittent nature of new energy power has an increasingly greater impact on the stable and safe operation of the power grid. To this end, it is urgent to build a bridge that can balance the vigorous development of new energy power and the safe operation of the power grid. The integrated system of large-scale energy storage devices and thermal power generating units has become an important means to promote deeper and faster deep peak regulation of thermal power generating units due to its fast response speed, low carbon and high efficiency. In particular, the high reliability, large capacity and high scalability of molten salt heat storage technology have been successfully applied on a large scale in the field of assisting large coal-fired units in deep peak regulation.

[0003] The supercritical unit coupled with molten salt heat storage cleverly utilizes the advantages of the fast heat storage and release speed of the molten salt device, flexibly changes the electric power provided by the supercritical unit to the power grid, promotes the consumption of new energy power, and thus improves the safe operation capability of the new power system. In order to improve the energy ladder utilization efficiency of the supercritical unit, the double-tank molten salt heat storage system is configured in the heat recycling system of the unit. After receiving the peak-shaving command from the power grid, the unit quickly reduces its output power, and the remaining power or extraction steam is used to heat the molten salt in the heat storage system. The heat energy carried by the molten salt is then used to heat the deoxygenated water and become a supplementary industrial steam supply. It can be seen that the coordinated peak-shaving process of the supercritical unit coupled with molten salt heat storage involves the phase change of the working fluid in the two major systems of the molten salt heat storage device and the supercritical unit, and the variable value change range and rate of different equipment measurement points are large, and multiple variables are strongly coupled. The pressure and temperature control effect of the working fluid inside the molten salt heat storage device and the supercritical unit has a direct impact on safe temperature and economic operation. Therefore, taking the supercritical unit coupled with molten salt heat storage as the research object, it is very necessary to design a coordinated control strategy with strong robustness, good tracking and anti-disturbance performance based on iterative learning-auto-disturbance rejection control to improve the safe operation level of the integrated system.

[0004] Active disturbance rejection control is an advanced control paradigm that treats the unmeasurable uncertainties inside and outside the system as total disturbances for estimation and compensation. The tracking differentiator, extended state observer, and nonlinear state error feedback control law constitute the trilogy of active disturbance rejection control. The traditional proportional-integral-differential controller is difficult to alleviate the contradiction between response speed and overshoot. Active disturbance rejection control balances response speed and overshoot by adding a tracking differentiator to arrange the transition process and giving a reasonable control signal. As the core component of active disturbance rejection control, the extended state observer estimates and compensates the total disturbance in real time, realizes the feedback linearization of the dynamic system, and changes the control object into an integral series type. The nonlinear state error feedback control law calculates the control signal based on the observation results and transmits it to the actuator to ensure the set value tracking and disturbance suppression performance of the closed-loop control system. In addition, in order to accurately track the changing peak-shaving instructions, an iterative learning module is added before the extended state observer of the active disturbance rejection control to dynamically adjust the control law. As a type of learning control strategy, iterative learning can achieve complete tracking of highly uncertain and complex dynamic systems with a small amount of prior knowledge. Therefore, integrating active disturbance rejection control and iterative learning methods to construct a multivariable coordinated control strategy is conducive to improving the deep peak-shaving capacity of supercritical units coupled with molten salt heat storage and promoting the construction of a high-proportion new energy power system. Summary of the invention

[0005] The present invention aims to provide a coordinated control strategy for the peak-shaving process of supercritical units coupled with molten salt heat storage, so as to enhance the depth and speed of large-scale thermal power generating units participating in peak-shaving and provide a solid guarantee for the consumption of large-scale new energy electricity. This strategy reasonably takes into account the complex dynamic characteristics of molten salt heat storage devices and supercritical units, and designs a coordinated control method for the peak-shaving process by combining self-disturbance rejection control and iterative learning methods that can handle nonlinearity and uncertainty. Iterative learning-self-disturbance rejection control accurately estimates the uncertainty and disturbance inside and outside the unit and compensates for it quickly, comprehensively considers the coupling relationship between multiple variables in the system, and dynamically adjusts the control law to achieve complete tracking of dynamic peak-shaving instructions.

[0006] The coordinated control strategy of the fire-storage coordinated peak-shaving process based on iterative learning and anti-disturbance control proposed in this invention consists of the following three steps:

[0007] S1: Analyze the difficulties of multivariable control in the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage;

[0008] S2: Establish a coordinated control strategy for the peak-shaving process of supercritical units coupled with molten salt heat storage;

[0009] S3: Rely on the simulation platform to verify and analyze the effectiveness of the designed coordinated control strategy for the collaborative peak-shaving process.

[0010] S1: Supercritical units coupled with molten salt heat storage, on the premise of not affecting operational safety and economy, flexibly utilize the storage-release function of molten salt heat storage devices to further reduce the electric power provided by the units to the power grid, promote the units to participate in deep peak regulation to increase revenue, and provide support for the development of new energy power. In order to achieve complete tracking of dynamic peak regulation instructions and suppress system internal and external disturbances in the peak regulation process, the coordinated control strategy of the coordinated peak regulation process of supercritical units coupled with molten salt heat storage is designed by iterative learning-self-disturbance rejection control with outstanding target tracking performance and anti-disturbance performance, strong robust performance and dynamic learning performance. Since the dynamic characteristics of the working fluids in the two systems in the deep peak regulation process are difficult to quantitatively describe, the molten salt of the molten salt heat storage device and the steam-water flow, pressure and temperature of the supercritical units change rapidly and over a large range, and inappropriate control laws will cause the failure of the control logic of the coordinated control system of the unit, overheating of the boiler heating surface, and boiler fire extinguishing. In order to improve the control effect of molten salt and steam-water pressure and temperature to achieve safe and efficient operation of the coordinated peak regulation process of supercritical units coupled with molten salt heat storage, a multivariable coordinated control strategy must be designed. In order to facilitate the design of control strategies, key variables that can accurately characterize the operating status of molten salt heat storage devices and supercritical units are extracted, and the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage is simplified into a coupled system with five inputs and five outputs. The five input variables are coal feed rate, water feed rate, main steam valve opening, low-temperature molten salt flow rate, and deoxygenated water flow rate; the five output variables are output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam supply flow rate. Each input variable and output variable has a strong coupling characteristic and a large difference in numerical dimensions, which places high requirements on the comprehensive performance of the control strategy.

[0011] Based on the analysis of S1, it can be seen that during the deep peak-shaving process, the values ​​of various variables of the supercritical units coupled with molten salt heat storage change rapidly and dramatically, and there are many uncertain disturbances. Therefore, the iterative learning with good target tracking performance and disturbance suppression performance is combined with the self-disturbance rejection control and dynamic learning characteristics to accurately estimate and compensate for the total disturbance in real time, so as to achieve complete tracking of the dynamic peak-shaving instructions. An iterative learning-self-disturbance rejection controller with the same structure is designed for each output loop of the supercritical unit coupled with molten salt heat storage, and the flexible factor is used to characterize the coupling relationship of each loop. The iterative learning-self-disturbance rejection controllers corresponding to the output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature and supplementary industrial steam supply flow loops are respectively recorded as IL-ADRC 1 IL-ADRC 2 IL-ADRC 3 IL-ADRC 4 IL-ADRC 5 ,. IL-ADRC of the output power circuit 1 Take this as an example to explain the design principle. The design process of other controllers is similar. Then step S2 can be concretely described as follows:

[0012] S2.1: The peak load regulation process of the output power loop of a supercritical coal-fired unit coupled with molten salt heat storage can be described by a second-order process:

[0013]

[0014] N E is the controlled quantity of the system, i.e., the output power of the supercritical unit coupled with molten salt heat storage. The second-order derivative of the output power with respect to the time variable t is recorded as u B is the control quantity, i.e., the coal feeding quantity; g is a lumped function including uncertainties such as time variable t, nonlinearity, internal and external disturbances d; b≠0 is the uncertainty of the system input gain; the total disturbance f of the system including unknown nonlinearity, uncertainty, external disturbance and internal dynamic is defined as:

[0015] f=g+(bb 0 ) B (2)

[0016] Among them, b 0 is the estimated value of b; in order to accurately respond to the peak load regulation instruction of the power grid, an iterative learning-ADRC control is designed to estimate and compensate the total disturbance f of the system.

[0017] Nonlinear ADRC consists of a tracking differentiator, an extended state observer, and a nonlinear error feedback control law. In order to reasonably balance the contradiction between rapid response and overshoot, a tracking differentiator is arranged inside the ADRC to generate a continuous and noise-free setpoint tracking signal and its differential form. The expression of the tracking differentiator is:

[0018]

[0019] Among them, v(t) is the output of the tracking differentiator, that is, the set value tracking signal, is the derivative of the setpoint tracking signal v(t).

[0020] The extended state observer is the core component of the ADRC. It estimates the states of various variables in the coordinated peak-shaving process of the supercritical unit coupled with molten salt heat storage, and adds an extended state to estimate the total disturbance signal f. The three state quantities x of the extended state observer are 1 、x 2 and x 3 The general form is:

[0021]

[0022] Where β=[L 1 L 2 ] is the gain matrix of the extended state observer; is the state variable x1 Derivatives of estimates; is the state variable x 2 Derivatives of estimates; is the state variable x 2 Estimated value; is the state variable x 3 Derivatives of estimates; is the state variable x 3 Estimated value.

[0023] The nonlinear state error feedback control law is the last component of the ADRC, which is responsible for calculating the control quantity, i.e., the coal feed rate u, based on the information provided by the tracking differentiator and the extended state observer. B (t), the expression is:

[0024]

[0025] Wherein, e(t) is the difference between the peak load regulation instruction of the power grid and the output power of the supercritical unit coupled with molten salt heat storage, i.e., the load tracking error; is the derivative of the load tracking error; k 1 , k 2 and k 3 is the parameter of the control law; α is the nonlinear term index used to enhance the robustness of the active disturbance rejection control law.

[0026] Since the peak load instructions and disturbances of the power grid are dynamic, the control law of the supercritical unit coupled with molten salt heat storage must be adaptively adjusted to respond to load demand. In this context, an iterative learning module with excellent learning control performance for nonlinear systems is added before the extended state observer to estimate the uncertain total disturbance f of the system. The iterative learning process is as follows:

[0027]

[0028] in, Therefore, w r (t,k) can be considered as an estimate of the total disturbance f of the system. By adding the correction term in each iteration The estimation accuracy of f can be quickly improved; k is the kth sample, a and ε are constant factors.

[0029] Define the iterative estimation correction σ(t,k) as:

[0030]

[0031] Then formula (6) can be rewritten as:

[0032]

[0033] Taking the derivative of σ(t,k) with respect to t, we can obtain:

[0034]

[0035] Then the estimation error of the system can be expressed as:

[0036]

[0037] in, is the nth-order derivative of the grid peak load regulation command with respect to time t.

[0038] The control law can be designed as:

[0039]

[0040] Among them, β≥0 is the feedback gain, w x (t,k) is defined as follows:

[0041]

[0042] but

[0043]

[0044] Among them, f and are bounded, and is locally uniformly continuous.

[0045] So far, the design of the coordinated control strategy for the thermal storage collaborative peak-shaving process based on iterative learning and active disturbance rejection control has been completed.

[0046] Based on the coordinated control strategy of the fire-storage coordinated peak-shaving process based on iterative learning and anti-disturbance control designed in step S2.1, the effectiveness of the designed coordinated control strategy of the coordinated peak-shaving process is verified and analyzed on the simulation platform in step S3. The specific process is as follows:

[0047] S3.1: 12,000 sets of actual operating data of the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage are selected, and a peak-shaving process state space model with a five-input and five-output structure is obtained based on the subspace identification method as the controlled object model of coordinated control.

[0048] S3.2: When the grid peak load regulation command is issued, the supercritical unit coupled with molten salt heat storage quickly reduces its output power, and the remaining electric energy or turbine exhaust steam is extracted to heat the molten salt. At the same time, the deoxygenated water is heated and vaporized by the heat carried by the molten salt to become supplementary industrial steam supply. During the peak load regulation process, the output power, separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam flow should change smoothly, and reasonably coordinate and respond to load commands. The flexible and multi-level utilization of heat effectively improves the deep peak load regulation capability and economic benefits of the supercritical unit coupled with molten salt heat storage.

[0049] S3.3: Observe the response effects of the five controlled quantities of the supercritical unit coupled with molten salt heat storage, namely, output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature and supplementary industrial steam flow rate.

[0050] S3.4: Quantitatively evaluate the control performance of the coordinated peak-shaving process of the units based on the designed coordinated control strategy, including the tracking error, adjustment time, overshoot and smoothness of the control quantity of the five output variables, and comprehensively evaluate the coordinated peak-shaving performance of the units. Beneficial effects of the present invention:

[0051] From the perspective of process control method design, the present invention designs a coordinated control strategy for the collaborative peak-shaving process of supercritical units coupled with molten salt heat storage, which is used to accurately and smoothly control the molten salt and steam-water pressure, temperature, and flow rate, and improve the ability of molten salt heat storage devices and supercritical units to coordinate and cooperate in deep peak-shaving, which meets the urgent needs of the construction of a high-proportion new energy power system.

[0052] The present invention comprehensively considers the dynamic characteristics of the molten salt heat storage device and the unit in the design of the coordinated control strategy for the collaborative peak-shaving process of the supercritical unit coupled with molten salt heat storage, constructs a simplified five-input and five-output model that accurately characterizes the operating state, and integrates iterative learning and self-disturbance rejection control to improve the real-time and complete tracking performance of the unit for dynamic peak-shaving instructions, providing solid support for the large-scale consumption of new energy electricity. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a simplified structure of the coordinated control strategy of the fire-storage coordinated peak-shaving process based on iterative learning and self-disturbance rejection control of the present invention;

[0054] To facilitate the design of control strategies, the supercritical unit coupled with molten salt heat storage is simplified into a five-input and five-output coupled system, with strong coupling between the variables. To achieve safe and efficient deep peak regulation, an iterative learning-auto-disturbance rejection controller is designed for each of the five output loops of the unit: output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam flow, which are denoted as IL-ADRC. 1 IL-ADRC 2 IL-ADRC 3 IL-ADRC 4 IL-ADRC 5 In addition, combined with the flexibility factor k 1 , k 2 , k 3 , k 4 and k 5Characterize the coupling relationship between the output variables. When the power grid peak load regulation command is issued, the supercritical unit coupled with molten salt heat storage quickly reduces its output power to the set value under the coordinated control strategy, and at the same time, the four output variables of separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam flow rate change smoothly to a stable state. DETAILED DESCRIPTION

[0055] The specific implementation modes of the present invention will be further described below with reference to the accompanying drawings.

[0056] Please refer to the attached figure in the instruction manual. Figure 1 , Figure 1 The invention simplifies the structure of the coordinated control strategy of the fire-storage coordinated peak-shaving process based on iterative learning and active disturbance rejection control, and adopts iterative learning-active disturbance rejection controller IL-ADRC 1 IL-ADRC 2 IL-ADRC 3 IL-ADRC 4 IL-ADRC 5 For output power N E (MW), separator outlet temperature T I (℃), main steam pressure P st (MPa), high temperature tank inlet molten salt temperature T HM (℃) and supplementary industrial steam flow D SW (m 3 / h) five output variables are controlled. In addition, combined with the flexibility factor k 1 , k 2 , k 3 , k 4 , k 5 The coupling between the output variables is characterized and fed back to the control quantity. The stable coordinated control among the five output variables effectively improves the coordinated peak-shaving capability of the supercritical units coupled with molten salt heat storage.

[0057] S1: Analyze the difficulties of multivariable control in the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage;

[0058] S2: Establish a coordinated control strategy for the peak-shaving process of supercritical units coupled with molten salt heat storage;

[0059] S3: Rely on the simulation platform to verify and analyze the effectiveness of the designed coordinated control strategy for the collaborative peak-shaving process.

[0060] S1: Supercritical units coupled with molten salt heat storage, on the premise of not affecting operational safety and economy, flexibly utilize the storage-release function of molten salt heat storage devices to further reduce the electric power provided by the units to the power grid, promote the units to participate in deep peak regulation to increase revenue, and provide support for the development of new energy power. In order to achieve complete tracking of dynamic peak regulation instructions and suppress system internal and external disturbances in the peak regulation process, the coordinated control strategy of the coordinated peak regulation process of supercritical units coupled with molten salt heat storage is designed by iterative learning-self-disturbance rejection control with outstanding target tracking performance and anti-disturbance performance, strong robust performance and dynamic learning performance. Since the dynamic characteristics of the working fluids in the two systems in the deep peak regulation process are difficult to quantitatively describe, the molten salt of the molten salt heat storage device and the steam-water flow, pressure and temperature of the supercritical units change rapidly and over a large range, and inappropriate control laws will cause the failure of the control logic of the coordinated control system of the unit, overheating of the boiler heating surface, and boiler fire extinguishing. In order to improve the control effect of molten salt and steam-water pressure and temperature to achieve safe and efficient operation of the coordinated peak regulation process of supercritical units coupled with molten salt heat storage, a multivariable coordinated control strategy must be designed. In order to facilitate the design of control strategies, key variables that can accurately characterize the operating status of molten salt heat storage devices and supercritical units are extracted, and the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage is simplified into a coupled system with five inputs and five outputs. The five input variables are coal feed rate, water feed rate, main steam valve opening, low-temperature molten salt flow rate, and deoxygenated water flow rate; the five output variables are output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam supply flow rate. Each input variable and output variable has a strong coupling characteristic and a large difference in numerical dimensions, which places high requirements on the comprehensive performance of the control strategy.

[0061] Based on the analysis of S1, it can be seen that during the deep peak-shaving process, the values ​​of various variables of the supercritical units coupled with molten salt heat storage change rapidly and dramatically, and there are many uncertain disturbances. Therefore, the iterative learning with good target tracking performance and disturbance suppression performance is combined with the self-disturbance rejection control and dynamic learning characteristics to accurately estimate and compensate for the total disturbance in real time, so as to achieve complete tracking of the dynamic peak-shaving instructions. An iterative learning-self-disturbance rejection controller with the same structure is designed for each output loop of the supercritical unit coupled with molten salt heat storage, and the flexible factor is used to characterize the coupling relationship of each loop. The iterative learning-self-disturbance rejection controllers corresponding to the output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature and supplementary industrial steam supply flow loops are respectively recorded as IL-ADRC 1 IL-ADRC 2 IL-ADRC 3 IL-ADRC 4 IL-ADRC 5 ,. IL-ADRC of the output power circuit 1 Take this as an example to explain the design principle. The design process of other controllers is similar. Then step S2 can be concretely described as follows:

[0062] S2.1: The peak load regulation process of the output power loop of a supercritical coal-fired unit coupled with molten salt heat storage can be described by a second-order process:

[0063]

[0064] N E is the controlled quantity of the system, i.e., the output power of the supercritical unit coupled with molten salt heat storage. The second-order derivative of the output power with respect to the time variable t is recorded as u B is the control quantity, i.e., the coal feeding quantity; g is a lumped function including uncertainties such as time variable t, nonlinearity, internal and external disturbances d; b≠0 is the uncertainty of the system input gain; the total disturbance f of the system including unknown nonlinearity, uncertainty, external disturbance and internal dynamic is defined as:

[0065] f=g+(bb 0 ) B (16)

[0066] Among them, b 0 is the estimated value of b; in order to accurately respond to the peak load regulation instruction of the power grid, an iterative learning-ADRC control is designed to estimate and compensate the total disturbance f of the system.

[0067] Nonlinear ADRC consists of a tracking differentiator, an extended state observer, and a nonlinear error feedback control law. In order to reasonably balance the contradiction between rapid response and overshoot, a tracking differentiator is arranged inside the ADRC to generate a continuous and noise-free setpoint tracking signal and its differential form. The expression of the tracking differentiator is:

[0068]

[0069] Among them, v(t) is the output of the tracking differentiator, that is, the set value tracking signal, is the derivative of the setpoint tracking signal v(t).

[0070] The extended state observer is the core component of the ADRC. It estimates the states of various variables in the coordinated peak-shaving process of the supercritical unit coupled with molten salt heat storage, and adds an extended state to estimate the total disturbance signal f. The three state quantities x of the extended state observer are 1 、x 2 and x 3 The general form is:

[0071]

[0072] Where β=[L 1 L 2 ] is the gain matrix of the extended state observer; is the state variable x1 Derivatives of estimates; is the state variable x 2 Derivatives of estimates; is the state variable x 2 Estimated value; is the state variable x 3 Derivatives of estimates; is the state variable x 3 Estimated value.

[0073] The nonlinear state error feedback control law is the last component of the ADRC, which is responsible for calculating the control quantity, i.e., the coal feed rate u, based on the information provided by the tracking differentiator and the extended state observer. B (t), the expression is:

[0074]

[0075] Wherein, e(t) is the difference between the peak load regulation instruction of the power grid and the output power of the supercritical unit coupled with molten salt heat storage, i.e., the load tracking error; is the derivative of the load tracking error; k 1 , k 2 and k 3 is the parameter of the control law; α is the nonlinear term index used to enhance the robustness of the active disturbance rejection control law.

[0076] Since the peak load instructions and disturbances of the power grid are dynamic, the control law of the supercritical unit coupled with molten salt heat storage must be adaptively adjusted to respond to load demand. In this context, an iterative learning module with excellent learning control performance for nonlinear systems is added before the extended state observer to estimate the uncertain total disturbance f of the system. The iterative learning process is as follows:

[0077]

[0078] in, Therefore, w r (t,k) can be considered as an estimate of the total disturbance f of the system. By adding the correction term in each iteration The estimation accuracy of f can be quickly improved; k is the kth sample, a and ε are constant factors.

[0079] Define the iterative estimation correction σ(t,k) as:

[0080]

[0081] Then formula (20) can be rewritten as:

[0082]

[0083] Taking the derivative of σ(t,k) with respect to t, we can obtain:

[0084]

[0085] Then the estimation error of the system can be expressed as:

[0086]

[0087] in, is the nth-order derivative of the grid peak load regulation command with respect to time t.

[0088] The control law can be designed as:

[0089]

[0090] Among them, β≥0 is the feedback gain, w x (t,k) is defined as follows:

[0091]

[0092] but

[0093]

[0094] f and are bounded, and is locally uniformly continuous.

[0095] So far, the design of the coordinated control strategy for the thermal storage collaborative peak-shaving process based on iterative learning and active disturbance rejection control has been completed.

[0096] Based on the coordinated control strategy of the fire-storage coordinated peak-shaving process based on iterative learning and anti-disturbance control designed in step S2.1, the effectiveness of the designed coordinated control strategy is verified and analyzed on the simulation platform in step S3. The specific process is as follows:

[0097] This example is based on the actual peak load data of the 660MW ultra-supercritical unit + 40MW / 80MWh molten salt thermal storage system of Jiangsu Jingjiang Power Plant. The method steps include:

[0098] In this example, the operating conditions of the coordinated peak-shaving process of the supercritical coal-fired unit coupled with molten salt heat storage are as follows:

[0099] Table 1 Operating conditions of 40MW / 80MWh+660MW supercritical coal-fired unit coupled with molten salt thermal storage

[0100] S3.1: 10,000 sets of actual historical operation data of the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage are selected, and the state space model of the unit peak-shaving process is obtained by combining the subspace identification method.

[0101] S3.2: Based on the state space model of the supercritical unit coordinated peak-shaving process coupled with molten salt heat storage obtained by identification, a coordinated control strategy is designed. The five output variables, output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature and supplementary industrial steam flow, are respectively determined by IL-ADRC 1 IL-ADRC 2 IL-ADRC 3 IL-ADRC 4 IL-ADRC 5 Take control.

[0102] S3.3: When the grid peak load regulation command is issued, the supercritical unit coupled with molten salt heat storage adjusts the five output variables of output power, separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam supply flow based on the designed coordinated peak load regulation process coordinated control strategy to quickly respond to load demand while maintaining the safe and stable operation of the molten salt heat storage system and the unit.

[0103] S3.4: Observe the response curves of output power, separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam supply flow during the coordinated peak regulation process.

[0104] S3.5: Calculate the regulation error, regulation time, and response smoothness of the output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam flow rate during the coordinated peak regulation of supercritical units coupled with molten salt heat storage.

[0105] The statistical results show that based on the coordinated control strategy of the thermal storage coordinated peak-shaving process based on iterative learning and self-disturbance rejection control designed by the present invention, the unit can accurately meet the peak-shaving instructions of the power grid. The load tracking error is only about 1.5% of the load demand instruction, which effectively improves the deep peak-shaving capability of the supercritical unit and helps promote the construction of a high-proportion new energy power system.

Claims

1. A coordinated control strategy for the peak load regulation process of thermal energy storage based on iterative learning and anti-disturbance control is characterized by: The following steps are involved: S1: Analyze the difficulties of multivariable control in the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage; S2: Establish a coordinated control strategy for the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage; the details are as follows: S2.1: The peak-shaving process of the output power loop of a supercritical coal-fired unit coupled with molten salt heat storage can be described by a second-order process: N E is the controlled quantity of the system, i.e., the output power of the supercritical unit coupled with molten salt heat storage. The second-order derivative of the output power with respect to the time variable t is recorded as u B is the control quantity, i.e., the coal feeding quantity; g is a lumped function including uncertainties such as time variable t, nonlinearity, internal and external disturbances d; b≠0 is the uncertainty of the system input gain; the total disturbance f of the system including unknown nonlinearity, uncertainty, external disturbance and internal dynamic is defined as: f=g+(b-b0)u B (2) Where b0 is the estimated value of b. In order to accurately respond to the peak-shaving command of the power grid, an iterative learning-ADRC is designed to estimate and compensate for the total disturbance f of the system. Nonlinear ADRC consists of a tracking differentiator, an extended state observer, and a nonlinear error feedback control law. In order to reasonably balance the contradiction between rapid response and overshoot, a tracking differentiator is arranged inside the ADRC to generate a continuous and noise-free setpoint tracking signal and its differential form. The expression of the tracking differentiator is: Among them, v(t) is the output of the tracking differentiator, that is, the set value tracking signal, is the derivative of the setpoint tracking signal v(t). The extended state observer is the core component of the ADRC. It estimates the states of various variables in the peak-shaving process of the supercritical unit coupled with molten salt heat storage, and adds an extended state to estimate the total disturbance signal f. The general form of the three state quantities x1, x2 and x3 of the extended state observer is: Where β = [L1 L2] is the gain matrix of the extended state observer; is the derivative of the estimated value of the state variable x1; is the derivative of the estimated value of the state variable x2; is the estimated value of the state variable x2; is the derivative of the estimated value of the state variable x3; Estimate the value of the state variable x3. The nonlinear state error feedback control law is the last component of the ADRC, which is responsible for calculating the control quantity, i.e., the coal feed rate u, based on the information provided by the tracking differentiator and the extended state observer. B (t), the expression is: Wherein, e(t) is the difference between the peak load regulation instruction of the power grid and the output power of the supercritical unit coupled with molten salt heat storage, i.e., the load tracking error; is the derivative of the load tracking error; k1, k2 and k3 are the parameters of the control law; α is the nonlinear term index used to enhance the robustness of the active disturbance rejection control law. Since the peak load instructions and disturbances of the power grid are dynamic, the control law of the supercritical unit coupled with molten salt heat storage must be adaptively adjusted to respond to load demand. In this context, an iterative learning module with excellent learning control performance for nonlinear systems is added before the extended state observer to estimate the uncertain total disturbance f of the system. The iterative learning process is as follows: in, Therefore, w r (t,k) can be considered as an estimate of the total disturbance f of the system. By adding the correction term in each iteration The estimation accuracy of f can be quickly improved; k is the kth sample, a and ε are constant factors. Define the iterative estimation correction σ(t,k) as: Then formula (20) can be rewritten as: Taking the derivative of σ(t,k) with respect to t, we can obtain: Then the estimation error of the system can be expressed as: in, is the nth-order derivative of the grid peak load regulation command with respect to time t. The control law can be designed as: Among them, β≥0 is the feedback gain, w x (t,k) is defined as follows: but f and are bounded, and is locally uniformly continuous. So far, the design of the coordinated control strategy for the collaborative peak-shaving process of supercritical units coupled with molten salt heat storage has been completed. Based on the Mamba-2 intelligent modeling scheme for the deep peak-shaving process of supercritical coal-fired units coupled with molten salt heat storage designed in step S2.1, the effectiveness of the designed modeling scheme is verified and analyzed in step S3 based on the simulation platform: S3: Rely on the simulation platform to verify and analyze the effectiveness of the designed coordinated control strategy for the collaborative peak-shaving process.

2. According to claim 1, a coordinated control strategy for the fire-storage coordinated peak-shaving process based on iterative learning and active disturbance rejection control is characterized by: The difficulties of multivariable control in the coordinated peak-shaving process of supercritical units coupled with molten salt heat storage described in step S1 are as follows: S1: Supercritical units coupled with molten salt heat storage, on the premise of not affecting operational safety and economy, flexibly utilize the storage-release function of molten salt heat storage devices to further reduce the electric power provided by the units to the power grid, promote the units to participate in deep peak regulation to increase revenue, and provide support for the development of new energy power. In order to achieve complete tracking of dynamic peak regulation instructions and suppress system internal and external disturbances in the peak regulation process, the coordinated control strategy of the coordinated peak regulation process of supercritical units coupled with molten salt heat storage is designed by iterative learning-self-disturbance rejection control with outstanding target tracking performance and anti-disturbance performance, strong robust performance and dynamic learning performance. Since the dynamic characteristics of the working fluids in the two systems in the deep peak regulation process are difficult to quantitatively describe, the molten salt of the molten salt heat storage device and the steam-water flow, pressure and temperature of the supercritical units change rapidly and over a large range, and inappropriate control laws will cause the failure of the control logic of the coordinated control system of the unit, overheating of the boiler heating surface, boiler fire extinguishing and other accidents. In order to improve the control effect of molten salt and steam-water pressure and temperature to achieve safe and efficient operation of the peak regulation process of supercritical units coupled with molten salt heat storage, a multivariable coordinated control strategy must be designed. In order to facilitate the design of control strategies, key variables that can accurately characterize the operating status of molten salt heat storage devices and supercritical units are extracted, and the peak-shaving process of supercritical units coupled with molten salt heat storage is simplified into a coupled system with five inputs and five outputs. The five input variables are coal feed rate, water feed rate, main steam valve opening, low-temperature molten salt flow rate, and deoxygenated water flow rate; the five output variables are output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam flow rate. Each input variable has a strong coupling characteristic with the output variable and the numerical dimension gap is large, which places high demands on the comprehensive performance of the control strategy.

3. According to claim 1, a coordinated control strategy for the fire-storage coordinated peak-shaving process based on iterative learning and active disturbance rejection control is characterized by: Based on the coordinated control strategy of the collaborative peak load regulation process designed in step S2, the effectiveness of the proposed coordinated control strategy is verified and analyzed in step S3 by relying on the simulation platform, including: S3.1: 10,000 sets of actual historical operation data of the deep peak-shaving process of supercritical units coupled with molten salt heat storage are selected, and the state space model of the unit peak-shaving process is obtained by combining the subspace identification method. S3.2: Based on the identified state space model of the supercritical unit coordinated peak-shaving process coupled with molten salt heat storage, a coordinated control strategy is designed. The five output variables of output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature and supplementary industrial steam flow are controlled by IL-ADRC1, IL-ADRC2, IL-ADRC3, IL-ADRC4 and IL-ADRC5 respectively. S3.3: When the grid peak load regulation command is issued, the supercritical unit coupled with molten salt heat storage adjusts the five output variables of output power, separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam supply flow based on the designed coordinated peak load regulation process coordinated control strategy to quickly respond to load demand while maintaining the safe and stable operation of the molten salt heat storage system and the unit. S3.4: Observe the response curves of output power, separator outlet temperature, main steam pressure, high temperature tank inlet molten salt temperature and supplementary industrial steam supply flow during the peak regulation process. S3.5: Calculate the regulation error, regulation time, and response smoothness of the output power, separator outlet temperature, main steam pressure, high-temperature tank inlet molten salt temperature, and supplementary industrial steam flow rate during the peak regulation process of the supercritical unit coupled with molten salt heat storage. The statistical results show that based on the coordinated control strategy of the thermal storage coordinated peak-shaving process based on iterative learning and self-disturbance rejection control designed by the present invention, the unit can accurately meet the peak-shaving instructions of the power grid. The load tracking error is only about 1.5% of the load demand instruction, which effectively improves the deep peak-shaving capability of the supercritical unit and helps promote the construction of a high-proportion new energy power system.

Images