A decoupling control system and method for a supercritical carbon dioxide direct cycle nuclear power plant

Abstract

A decoupling control system and method for a supercritical carbon dioxide direct cycle nuclear power plant adopts a compressor inlet constant temperature and compressor inlet constant pressure control strategy, which is achieved by adjusting the cooler cold side valve opening through the compressor inlet temperature control system and the compressor inlet pressure control system to adjust the filling and discharge of working fluid in the capacity box; to ensure power supply quality, a drive shaft speed constant control strategy is adopted, which is achieved by adjusting the drive shaft auxiliary torque through the drive shaft speed control system; to ensure system efficiency, a core outlet constant temperature control strategy is adopted, which is achieved by adjusting the control valve opening through the core outlet temperature control system; to achieve the reactor following machine operation target, a core power matching demand power control strategy is adopted, which is achieved by adjusting the control rod speed through the power control system. Based on the neuron PID decoupling control technology, the control target that each output parameter in the system is affected only by a specific input parameter is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of supercritical carbon dioxide direct cycle nuclear power plants, and in particular to a decoupling control system and method for supercritical carbon dioxide direct cycle nuclear power plants. Background Art

[0002] A supercritical carbon dioxide direct cycle nuclear power system uses carbon dioxide as a working fluid, extracting nuclear heat directly from the nuclear reactor and converting this energy into electrical or mechanical energy through a supercritical carbon dioxide thermodynamic cycle. Because supercritical carbon dioxide exhibits a sudden change in physical properties at its critical point (7.38 MPa, 31.2°C), the compressor inlet operating conditions in a supercritical carbon dioxide direct cycle nuclear power system are set near this critical point. By leveraging the low compressibility of supercritical carbon dioxide at this operating condition and the high density difference between the compressor inlet and outlet caused by this sudden change in physical properties, the power consumption of the supercritical carbon dioxide compressor can be significantly reduced. Furthermore, as a primary coolant, carbon dioxide is safe, non-toxic, non-flammable, abundant, and naturally occurring. It possesses excellent physical and chemical stability and exhibits inert gas properties within the reactor's design temperature range. Compared to other conventional gaseous working fluids, supercritical carbon dioxide can more easily achieve a larger density difference, which facilitates the removal of residual heat from the core through natural circulation and significantly improves reactor reliability.

[0003] The supercritical carbon dioxide direct cycle nuclear power system is a highly promising new power system, boasting advantages such as simplified system design, compact structure, enhanced mobility, and high thermal efficiency. Its large-scale deployment will facilitate the clean and efficient use of my country's nuclear resources, significantly reduce greenhouse gas emissions, and be of vital importance in achieving the "dual carbon" goals. These advantages of the supercritical carbon dioxide direct cycle nuclear power system not only meet the development requirements of next-generation advanced energy conversion technologies but also align with future development trends in nuclear power plants, as described in the article "The Rise and Development of Supercritical Carbon Dioxide Nuclear Power Systems" by Huang Yanping et al., published in Atomic Energy Science and Technology, Vol. 57, No. 9, September 2023.

[0004] However, since the physical properties of supercritical carbon dioxide working fluid change dramatically near the critical point, and the supercritical carbon dioxide direct cycle nuclear power system has the characteristics of strong coupling of multiple devices, the control actions of each control system will have a strong mutual influence, making the regulation and control of its variable load, startup, shutdown and other processes more complicated. For example, the patent application with publication number CN112071457B discloses a load tracking method for a supercritical carbon dioxide direct cooling reactor system. When operating at full load, the CO2 working fluid flows into the turbine to expand and do work, releases part of the heat in the regenerator, and then enters the precooler for cooling. After entering the compressor for supercharging, it flows into the regenerator to absorb the energy of the turbine exhaust gas and returns to the reactor to complete the entire cycle process. When operating at partial load, the generator power control system is used to match the generator power with the grid demand. The compressor inlet temperature control system is used to control the compressor inlet temperature during load operation. The compressor blocking protection system is used to ensure that the compressor is away from the blocking operation area. The power of the reactor is automatically adjusted by the core power control system based on reactivity feedback, so that the reactor power is automatically adjusted following the change of the generator power, thereby realizing the load tracking capability of the supercritical carbon dioxide direct cooling reactor system within the full load range. However, this patent application has the defect that each control action affects each other. Therefore, it is necessary to select a suitable decoupling control method to reduce or eliminate the mutual influence of various control actions as much as possible, so that the fluctuation time and amplitude of the main parameters of the system can be controlled within a reasonable range under various loads and disturbances, avoiding overheating, overspeed, surge, etc., and maximizing the unit response speed, so that the system can quickly restore stability and ensure safe and stable operation of the system. It has important research significance and application value. Summary of the Invention

[0005] In order to overcome the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a decoupling control system and method for a supercritical carbon dioxide direct cycle nuclear power plant. In view of the multi-variable strong coupling characteristics of the supercritical carbon dioxide direct cycle nuclear power system, the mutual influence between different control channels is eliminated under its operating conditions such as variable load, startup, and shutdown, so that each output variable in the controlled system is only affected by a specific input variable, thereby achieving the decoupling control goal of the multi-variable strong coupling system and ensuring the safe and stable operation of the system.

[0006] To achieve the above objectives, the present invention adopts the following technical solutions:

[0007] A decoupling control system for a supercritical carbon dioxide direct cycle nuclear power plant includes a reactor 6. The outlet of a carbon dioxide pipeline of the reactor 6 is connected to the inlet of a turbine 10. The outlet of the carbon dioxide pipeline of the turbine 10 passes through the cold side of a regenerator 7, the secondary side of a cooler 8, a compressor 9, and the hot side of the regenerator 7 in sequence and returns to the inlet of the carbon dioxide of the reactor 6. The turbine shaft of the turbine 10 is connected to the rotating shaft of a generator 11.

[0008] A first temperature sensor and a compressor inlet temperature control system 1 are provided on the pipeline between the secondary side carbon dioxide outlet end of the cooler 8 and the carbon dioxide inlet end of the compressor 9. The signal output end of the first temperature sensor is connected to the signal input end of the compressor inlet temperature control system 1; the compressor inlet temperature control system 1 adjusts the opening of the first control valve 13 arranged at the cold water inlet end of the primary side of the cooler.

[0009] A first pressure sensor and a compressor inlet pressure control system 2 are provided on the pipeline between the secondary side carbon dioxide outlet of the cooler 8 and the carbon dioxide inlet of the compressor 9. The signal output end of the first pressure sensor is connected to the signal input end of the compressor inlet pressure control system 2, and the signal output end of the compressor inlet pressure control system 2 is connected to the signal end of the second control valve 14 at the outlet end of the capacity tank 12;

[0010] A second temperature sensor is provided on the pipe between the outlet end of the carbon dioxide pipe of the reactor 6 and the inlet end of the carbon dioxide pipe of the turbine 10. The signal end of the second temperature sensor is connected to the signal end of the core outlet temperature control system 3. The core outlet temperature control system 3 adjusts the opening of the third control valve 15 arranged between the cold side inlet end and the hot side outlet end of the regenerator 7.

[0011] The signal output of the first power sensor on the generator 11 passes through the signal terminal of the power control system 4, the signal terminal of the control drum 16 and the signal terminal of the reactor 6 in sequence, and the power control system 4 drives the control drum 16 to adjust the reactivity;

[0012] A first speed sensor is provided on the transmission shaft connecting the compressor 9 and the turbine 10. The signal terminal of the first speed sensor is connected to the signal terminal of the transmission shaft speed control system 5. The transmission shaft speed control system 5 adjusts the auxiliary torque of the transmission shaft.

[0013] The compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4 and the transmission shaft speed control system 5 are all equipped with a built-in PID neural network decoupling controller, which is constructed into a 10×15×5 forward network structure to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

[0014] The decoupling control method based on the above-mentioned supercritical carbon dioxide direct cycle nuclear power plant decoupling control system includes the following steps:

[0015] Step 1: Build a control system for a supercritical carbon dioxide direct cycle nuclear power plant: The compressor inlet temperature control system 1 maintains a constant compressor 9 inlet temperature by adjusting the cold side flow of the cooler 8; the compressor inlet pressure control system 2 maintains a constant compressor 9 inlet pressure by controlling the circulating working fluid in the capacity tank 12; the core outlet temperature control system 3 maintains a constant reactor 6 outlet temperature by adjusting the opening of the control valve 15; the power control system 4 achieves real-time matching of reactor power and load power by driving the reactivity introduced by the control drum 16; the compressor 9, turbine 10, and generator 11 rotate coaxially, and the drive shaft speed control system 5 maintains a constant drive shaft speed by adjusting the drive shaft auxiliary torque;

[0016] Step 2: The compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4, and the transmission shaft speed control system 5 in step 1 are constructed into a PID neural network decoupling control method with a 10×15×5 forward network structure;

[0017] Step 3: Based on the PID neural network decoupling control method constructed in step 2, the beetle swarm algorithm is used to globally search for the optimal initial weights of the five control systems: compressor inlet temperature control system 1, compressor inlet pressure control system 2, core outlet temperature control system 3, power control system 4, and transmission shaft speed control system 5;

[0018] Step 4: When the operating conditions of the supercritical carbon dioxide direct cycle nuclear power system change, in order to ensure the critical state of the working fluid at the compressor inlet, the PID neural network will take the difference between the output set value and the actual output value of the controlled system as input, and update the connection weights of the neurons in real time through the supervised Hebb learning rule, and adjust the strength of the proportional, integral, and differential effects in the system control process in real time, and finally obtain the input of the controlled system, complete the decoupling control of the supercritical carbon dioxide direct cycle nuclear power device, and convert the multivariable strong coupling system into multiple single-input and single-output systems, so as to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

[0019] The step 2 is constructed into a 10×15×5 forward network structure, that is, the compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4 and the transmission shaft speed control system 5 are five parallel sub-network structures, T c_ref 、P c_ref 、T core_ref 、N r_ref 、vref are the input set values ​​of the PID neural network decoupling controller; KV1, KV2, KV3, v r , T are the input values ​​of the controlled system respectively; T c 、P c 、T core 、N r and v are the actual output values ​​of the controlled system respectively; w si is the connection weight from the hidden layer to the output layer; s is the serial number of the parallel sub-network, s = 1, 2, 3, 4, 5; i is the serial number of the hidden layer neuron, i = 1, 2, 3.

[0020] The specific algorithm process of the beetle swarm algorithm in step 3 is as follows:

[0021] Initialize the parameters of the longhorn beetle group, set the number of longhorn beetles in the group, randomly assign the speed, position, and flight direction parameters of each longhorn beetle, and define the following fitness function:

[0022]

[0023] Where T is the maximum number of iterations of the beetle swarm algorithm, length is the number of PIDNN training times, n is the number of beetles in the beetle swarm, r is the target signal value, and y is the output of the system.

[0024] The fitness value of each beetle's position in the beetle population is calculated based on the fitness function, and the optimal position of each beetle and the global optimal position are updated based on the fitness value of each beetle's position.

[0025] Update the inertia weight:

[0026]

[0027] Where T is the maximum number of iterations, θ t is the inertia weight, θ max and θ min are the maximum and minimum values ​​of θ respectively.

[0028] Update the speed of each longhorn beetle:

[0029]

[0030] Where, represents the position of the i-th longicorn in the t-th iteration of the q-th dimension, represents the speed of the i-th longhorn beetle in the t-th iteration in the q-th dimension, is the individual extreme position, is the extreme position of the population, b1 and b2 are two learning factors, z3 and z4 are random numbers in [0,1];

[0031] Update the increment function:

[0032]

[0033] Where, represents the increment function, δ t It represents the step size in the t-th iteration of the longhorn beetle, which decreases as the number of iterations increases and is updated according to the following formula:

[0034] δ t =z1δ t-1 +δ 0

[0035] In the formula, z1 is a positive constant, δ 0 is the initial value of the step length, and the positions of the left and right tentacles of each longhorn beetle are updated as follows:

[0036]

[0037]

[0038] Where, d t represents the distance between the two whiskers in the tth iteration;

[0039] Update the position of each longhorn beetle:

[0040]

[0041] Where β is a positive constant. After multiple iterations, the beetle swarm algorithm completes the global search and obtains the global optimal solution.

[0042] The PID neural network in step 4 takes the difference between the output set value and the actual output value of the controlled system as input. The PID neural network input variables x1, x2, and x3 will be expressed in incremental form:

[0043]

[0044] According to the input variables x1, x2, and x3 of the PID neural network, the input quantity of the controlled system will be calculated as follows:

[0045]

[0046] Where K is the neuron proportional coefficient, K>0; w1, w2, w3 are the neuron connection weights, which are approximately equal to the proportional, integral, and differential coefficients of the PID controller.

[0047] The supervised Hebb learning rule in step 4 is an algorithm rule that selects the supervised Hebb learning rule as the connection weight, specifically:

[0048]

[0049] Where η p is the proportional learning rate; η i is the integrated learning rate; η d is the differential learning rate;

[0050] The connection weights of the neural network are adjusted in real time through the supervised Hebb learning rule until the input error signal is close to zero. At this time, the input of the controlled system is the correct input. That is, by adjusting the strength of the proportional, integral and differential effects of the PID neural network, the mutual influence between different control channels is eliminated.

[0051] In step 4, the difference between the output set value and the actual output value of the controlled system is used as input. In the five control systems, namely, the compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4, and the transmission shaft speed control system 5, the difference input and the PID neural network decoupling control method are specifically as follows:

[0052] The difference between the compressor inlet temperature set value of the compressor inlet temperature control system 1 and the measured value of the first temperature sensor is calculated, and the difference signal is sent to the PID neural network decoupling controller in the compressor inlet temperature control system 1; the decoupling controller sends its output signal to the first control valve 13;

[0053] The difference between the compressor inlet pressure set value of the compressor inlet pressure control system 2 and the measured value of the first pressure sensor is made, and the difference signal is sent to the PID neural network decoupling controller in the compressor inlet pressure control system 2; the decoupling controller sends its output signal to the second control valve 14 of the capacity tank;

[0054] The core outlet temperature set value of the core outlet temperature control system 3 is subtracted from the measured value of the second temperature sensor, and the difference signal is sent to the PID neuron network decoupling controller in the core outlet temperature control system 3; the decoupling controller sends its output signal to the third control valve 15;

[0055] The power setting value of the power control system 4 is subtracted from the measured value of the first power sensor, and the difference signal is sent to the PID neuron network decoupling controller in the power control system 4; the decoupling controller sends its output signal to the control drum 16;

[0056] The difference between the transmission shaft speed set value of the transmission shaft speed control system 5 and the measured value of the first speed sensor is sent to the PID neuron network decoupling controller in the transmission shaft speed control system 5; the decoupling controller sends its output signal to the actuator to adjust the transmission shaft auxiliary torque.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1. To ensure the critical state of the working fluid at the compressor inlet, a constant compressor inlet temperature and constant compressor inlet pressure control strategy is adopted, and the compressor inlet temperature control system is used to adjust the cooler cold side valve opening and the compressor inlet pressure control system is used to adjust the filling and discharge of the working fluid in the capacity box; to ensure the power supply quality, a constant drive shaft speed control strategy is adopted, and the drive shaft auxiliary torque is adjusted through the drive shaft speed control system; to ensure system efficiency, a constant core outlet temperature control strategy is adopted, and the core outlet temperature control system is used to adjust the control valve opening; to achieve the reactor follow-up operation target, a core power matching demand power control strategy is adopted, and the power control system is used to adjust the control rod speed.

[0059] 2. The decoupling system of the supercritical carbon dioxide direct cycle nuclear power plant aims to solve the multi-parameter and strong coupling problems of the supercritical carbon dioxide direct cycle system. Based on the neuron PID decoupling control technology, the system is converted into multiple single-input and single-output controls, eliminating the mutual influence between different control channels and achieving the control goal that each output parameter in the system is only affected by a specific input parameter.

[0060] In summary, the decoupling system and method for a supercritical carbon dioxide direct cycle nuclear power plant proposed in this invention are easy to implement in engineering. The five control systems transform the supercritical carbon dioxide direct cycle nuclear power plant from a multivariable, tightly coupled system into multiple single-input, single-output control systems. This achieves the control objective of ensuring that each output variable in the multivariable coupled system is affected only by a specific input variable. Furthermore, it is expected to offer good economic efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the connection of the entire system of the present invention.

[0062] Figure 2 This is the PID single neuron network control structure diagram.

[0063] Figure 3 This is the PID neural network decoupling control structure diagram. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0065] Reference Figure 1A decoupling control system for a supercritical carbon dioxide direct cycle nuclear power plant includes a reactor 6, wherein the outlet of a carbon dioxide pipeline of the reactor 6 is connected to the inlet of a carbon dioxide pipeline of a turbine 10. The outlet of the carbon dioxide pipeline of the turbine 10 passes through the cold side of a regenerator 7, the secondary side of a cooler 8, a compressor 9, and the hot side of the regenerator 7 in sequence and returns to the carbon dioxide inlet of the reactor 6; the turbine shaft of the turbine 10 is connected to the rotating shaft of a generator 11;

[0066] A first temperature sensor and a compressor inlet temperature control system 1 are provided on the pipeline between the secondary side carbon dioxide outlet end of the cooler 8 and the carbon dioxide inlet end of the compressor 9. The signal output end of the first temperature sensor is connected to the signal input end of the compressor inlet temperature control system 1; the compressor inlet temperature control system 1 adjusts the opening of the first control valve 13 arranged at the cold side inlet end of the cooler 8.

[0067] A first pressure sensor and a compressor inlet pressure control system 2 are provided on the pipeline between the secondary side carbon dioxide outlet of the cooler 8 and the carbon dioxide inlet of the compressor 9. The signal output end of the first pressure sensor is connected to the signal input end of the compressor inlet pressure control system 2, and the signal output end of the compressor inlet pressure control system 2 is connected to the signal end of the second control valve 14 at the outlet end of the capacity tank 12; the opening of the second control valve 14 is controlled;

[0068] A second temperature sensor is provided on the pipe between the outlet end of the carbon dioxide pipe of the reactor 6 and the inlet end of the carbon dioxide pipe of the turbine 10. The signal end of the second temperature sensor is connected to the signal end of the core outlet temperature control system 3. The core outlet temperature control system 3 adjusts the opening of the third control valve 15 arranged between the cold side inlet end and the hot side outlet end of the regenerator 7.

[0069] The signal output of the first power sensor on the generator 11 passes through the signal terminal of the power control system 4, the signal terminal of the control drum 16 and the signal terminal of the reactor 6 in sequence, and the power control system 4 drives the control drum 16 to adjust the reactivity;

[0070] A first speed sensor is provided on the transmission shaft connecting the compressor 9 and the turbine 10. The signal terminal of the first speed sensor is connected to the signal terminal of the transmission shaft speed control system 5. The transmission shaft speed control system 5 adjusts the auxiliary torque of the transmission shaft.

[0071] The compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4 and the transmission shaft speed control system 5 are all equipped with a built-in PID neural network decoupling controller, which is constructed into a 10×15×5 forward network structure to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

[0072] The operating principle of the above system is as follows: when operating at full load, the circulating fluid is heated by the reactor 6 and then flows into the turbine 10, where it expands and produces work, driving the generator 11 to rotate. Subsequently, the circulating fluid passes through the regenerator 7, releases some heat, enters the cooler 8, is cooled to the target temperature, enters the compressor 9 for pressurization, and then passes through the regenerator 7 to absorb the heat of the hot-side circulating fluid before returning to the reactor 6, completing the entire cycle.

[0073] When operating at partial load, the compressor inlet temperature control system 1 is used to control the cold side feed water flow of the cooler 8 to maintain the constant inlet temperature of the compressor 9; the compressor inlet pressure control system 2 is used to control the filling and discharge capacity of the capacity tank 12 to maintain the constant compressor inlet pressure; the core outlet temperature control system 3 is used to adjust the opening of the third control valve 15 to maintain the constant core outlet temperature; the power control system 4 is used to drive the control drum 16 to ensure that the core output power matches the required power; the drive shaft speed control system 5 is used to adjust the drive shaft auxiliary torque to maintain the drive shaft speed constant.

[0074] The decoupling control method based on the above-mentioned supercritical carbon dioxide direct cycle nuclear power plant decoupling control system includes the following steps:

[0075] Step 1: Build a control system for a supercritical carbon dioxide direct cycle nuclear power plant: The compressor inlet temperature control system 1 maintains a constant compressor 9 inlet temperature by adjusting the cold side flow of the cooler 8; the compressor inlet pressure control system 2 maintains a constant compressor 9 inlet pressure by controlling the circulating working fluid in the capacity tank 12; the core outlet temperature control system 3 maintains a constant reactor 6 outlet temperature by adjusting the opening of the control valve 15; the power control system 4 achieves real-time matching of reactor power and load power by driving the reactivity introduced by the control drum 16; the compressor 9, turbine 10, and generator 11 rotate coaxially, and the drive shaft speed control system 5 maintains a constant drive shaft speed by adjusting the drive shaft auxiliary torque;

[0076] Step 2: The compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4, and the transmission shaft speed control system 5 in step 1 are constructed into a PID neural network decoupling control method with a 10×15×5 forward network structure;

[0077] Step 3: Based on the PID neural network decoupling control method constructed in step 2, the beetle swarm algorithm is used to globally search for the optimal initial weights of the five control systems: compressor inlet temperature control system 1, compressor inlet pressure control system 2, core outlet temperature control system 3, power control system 4, and transmission shaft speed control system 5;

[0078] Step 4: When the operating conditions of the supercritical carbon dioxide direct cycle nuclear power system change, in order to ensure the critical state of the working fluid at the compressor inlet, the PID neural network will take the difference between the output set value and the actual output value of the controlled system as input, and update the connection weights of the neurons in real time through the supervised Hebb learning rule, and adjust the strength of the proportional, integral, and differential effects in the system control process in real time, and finally obtain the input of the controlled system, complete the decoupling control of the supercritical carbon dioxide direct cycle nuclear power device, and convert the multivariable strong coupling system into multiple single-input and single-output systems, so as to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

[0079] See also Figure 3 , the step 2 is constructed into a 10×15×5 forward network structure, that is, the compressor inlet temperature control system 1, the compressor inlet pressure control system 2, the core outlet temperature control system 3, the power control system 4 and the transmission shaft speed control system 5 are five parallel sub-network structures, which is a five-input, five-output coupled system. Therefore, the PID neural network decoupling controller is designed as a 10×15×5 three-layer forward network structure, T c_ref 、P c_ref 、T core_ref 、N r_ref 、v ref are the input set values ​​of the PID neural network decoupling controller; KV1, KV2, KV3, v r , T are the input values ​​of the controlled system respectively; T c 、P c 、T core 、N r and v are the actual output values ​​of the controlled system respectively; w si is the connection weight from the hidden layer to the output layer; s is the serial number of the parallel sub-network, s = 1, 2, 3, 4, 5; i is the serial number of the hidden layer neuron, i = 1, 2, 3.

[0080] The beetle swarm algorithm in step 3 combines the characteristics of the particle swarm algorithm and the beetle whisker algorithm, and searches for the optimal solution globally by simulating the foraging mechanism of the beetle swarm. The specific algorithm process is as follows:

[0081] Initialize the parameters of the longhorn beetle group, set the number of longhorn beetles in the group, randomly assign the speed, position, and flight direction parameters of each longhorn beetle, and define the following fitness function:

[0082]

[0083] Where T is the maximum number of iterations of the beetle swarm algorithm, length is the number of PIDNN training times, n is the number of beetles in the beetle swarm, r is the target signal value, and y is the output of the system.

[0084] The fitness value of each beetle's position in the beetle population is calculated based on the fitness function, and the optimal position of each beetle and the global optimal position are updated based on the fitness value of each beetle's position.

[0085] Update the inertia weight:

[0086]

[0087] Where T is the maximum number of iterations, θ t is the inertia weight, θ max and θ min are the maximum and minimum values ​​of θ respectively.

[0088] Update the speed of each longhorn beetle:

[0089]

[0090] Where, represents the position of the i-th longicorn in the t-th iteration of the q-th dimension, represents the speed of the i-th longhorn beetle in the t-th iteration in the q-th dimension, is the individual extreme position, is the extreme position of the population, b1 and b2 are two learning factors, z3 and z4 are random numbers in [0,1];

[0091] Update the increment function:

[0092]

[0093] Where, represents the increment function, δ t It represents the step size in the t-th iteration of the longhorn beetle, which decreases as the number of iterations increases and is updated according to the following formula:

[0094] δ t =z1δ t-1 +δ 0

[0095] In the formula, z1 is a positive constant, δ 0 is the initial value of the step length, and the positions of the left and right tentacles of each longhorn beetle are updated as follows:

[0096]

[0097]

[0098] Where, d t represents the distance between the two whiskers in the tth iteration;

[0099] Update the position of each longhorn beetle:

[0100]

[0101] Where β is a positive constant. After multiple iterations, the beetle swarm algorithm completes the global search and obtains the global optimal solution.

[0102] In step 4, the PID neural network takes the difference between the output set value and the actual output value of the controlled system as input. The PID neural network input variables x1, x2, and x3 are expressed in incremental form:

[0103]

[0104] According to the input variables x1, x2, and x3 of the PID neural network, the input quantity of the controlled system will be calculated as follows:

[0105]

[0106] Where K is the neuron proportional coefficient, K>0; w1, w2, w3 are the neuron connection weights, which are approximately equal to the proportional, integral, and differential coefficients of the PID controller.

[0107] See also Figure 2 In the present invention, the actual output value y of the controlled system is fed back to the input of the control system in real time. By comparing it with the input setpoint r, an error value e is obtained. This error value e is converted to x1, x2, and x3 through state transitions, and then weighted by w1, w2, and w3 to obtain the output value u. The PID neural network structure is continuously trained based on the error value e, adjusting the network connection weights to continuously reduce the error value e until it approaches zero. At this point, the output value u of the PID neural network structure is the correct input that can achieve the ideal setpoint for the controlled system. This output value u of the PID neural network structure is then input into the controller's actuator. After a series of execution actions, all output values ​​y of the controlled system will reach the target setpoint r, achieving decoupling control for the controlled system.

[0108] The supervised Hebb learning rule is an algorithm rule that selects the supervised Hebb learning rule as the connection weight, specifically:

[0109]

[0110] Where η p is the proportional learning rate; η i is the integrated learning rate; η d is the differential learning rate;

[0111] The connection weights of the neural network are adjusted in real time through the supervised Hebb learning rule until the input error signal is close to zero. At this time, the input of the controlled system is the correct input. That is, by adjusting the strength of the proportional, integral and differential effects of the PID neural network, the mutual influence between different control channels is eliminated. The control goal of the supercritical carbon dioxide direct cycle nuclear system is achieved that each output parameter is only affected by a specific input parameter.

[0112] When operating at full load, the circulating fluid is heated by the reactor 6 and then flows into the turbine 10 to expand and perform work, driving the generator 11 to rotate. Subsequently, the circulating fluid passes through the regenerator 7 to release some of its heat, enters the cooler 8 to be cooled to the target temperature, enters the compressor 9 for pressurization, and then passes through the regenerator 7 to absorb the heat of the hot-side circulating fluid before returning to the reactor 6, completing the entire cycle.

[0113] When operating at partial load, the compressor inlet temperature control system 1 is used to control the cold side feed water flow of the cooler 8 to maintain the constant inlet temperature of the compressor 9; the compressor inlet pressure control system 2 is used to control the filling and discharge capacity of the capacity tank 12 to maintain the constant compressor inlet pressure; the core outlet temperature control system 3 is used to adjust the opening of the third control valve 15 to maintain the constant core outlet temperature; the power control system 4 is used to drive the control drum 16 to ensure that the core output power matches the required power; the drive shaft speed control system 5 is used to adjust the drive shaft auxiliary torque to maintain the drive shaft speed constant.

[0114] See also Figure 3 In order to ensure the critical state of the working fluid at the compressor inlet, it is necessary to ensure that the working fluid temperature at the compressor inlet is maintained at the critical temperature. A compressor inlet temperature control system 1 is designed. The compressor inlet temperature set value is subtracted from the measured value of the first temperature sensor, and the difference signal is sent to the PID neural network decoupling controller in the compressor inlet temperature control system 1; the decoupling controller sends its output signal to the first control valve 13.

[0115] In order to ensure the critical state of the working fluid at the compressor inlet, it is necessary to ensure that the working fluid pressure at the compressor inlet is maintained at the critical pressure. The compressor inlet pressure control system 2 is designed. The compressor inlet pressure set value is subtracted from the measured value of the first pressure sensor, and the difference signal is sent to the PID neuron network decoupling controller in the compressor inlet pressure control system 2; the decoupling controller sends its output signal to the second control valve 14 of the capacity box.

[0116] To ensure the highest possible loop efficiency, the core outlet working fluid temperature must be maintained constant. Therefore, a core outlet temperature control system 3 is designed. The core outlet temperature setpoint is subtracted from the value measured by the second temperature sensor. This difference signal is sent to a PID neural network decoupling controller within core outlet temperature control system 3. The decoupling controller then sends its output signal to third control valve 15.

[0117] In order to ensure the stack-following operation strategy, it is necessary to ensure that the reactor power matches the required power in real time. A power control system 4 is designed, and the power set value is subtracted from the measured value of the first power sensor. The difference signal is sent to the PID neural network decoupling controller in the power control system 4; the decoupling controller sends its output signal to the control drum 16.

[0118] In order to ensure the power supply quality, it is necessary to ensure that the transmission shaft speed remains constant. A transmission shaft speed control system 5 is designed. The transmission shaft speed set value is subtracted from the measured value of the first speed sensor, and the difference signal is sent to the PID neural network decoupling controller in the transmission shaft speed control system 5; the decoupling controller sends its output signal to the actuator to adjust the transmission shaft auxiliary torque.

[0119] The above content is only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A decoupling control system for a supercritical carbon dioxide direct cycle nuclear power plant, comprising a reactor (6), wherein the outlet end of a carbon dioxide pipeline of the reactor (6) is connected to the inlet end of a carbon dioxide pipeline of a turbine (10), and the outlet end of the carbon dioxide pipeline of the turbine (10) is returned to the carbon dioxide inlet end of the reactor (6) through the cold side of a regenerator (7), the secondary side of a cooler (8), a compressor (9), and the hot side of the regenerator (7); the turbine shaft of the turbine (10) is connected to the rotating shaft of a generator (11); and characterized in that: A first temperature sensor and a compressor inlet temperature control system (1) are provided on a pipeline between the secondary side carbon dioxide outlet of the cooler (8) and the carbon dioxide inlet of the compressor (9), and a signal output end of the first temperature sensor is connected to a signal input end of the compressor inlet temperature control system (1); The compressor inlet temperature control system (1) adjusts the opening of a first control valve (13) arranged at the cold water inlet end of the primary side of the cooler; A first pressure sensor and a compressor inlet pressure control system (2) are provided on a pipeline between the secondary side carbon dioxide outlet of the cooler (8) and the carbon dioxide inlet of the compressor (9); a signal output end of the first pressure sensor is connected to a signal input end of the compressor inlet pressure control system (2); and a signal output end of the compressor inlet pressure control system (2) is connected to a signal end of a second control valve (14) at an outlet end of a capacity tank (12); A second temperature sensor is provided on the pipeline between the outlet end of the carbon dioxide pipeline of the reactor (6) and the carbon dioxide inlet end of the turbine (10), and a signal end of the second temperature sensor is connected to a signal end of a core outlet temperature control system (3). The core outlet temperature control system (3) adjusts the opening of a third control valve (15) arranged between the cold side inlet end and the hot side outlet end of the regenerator (7); The signal output of the first power sensor on the generator (11) passes through the signal terminal of the power control system (4), the signal terminal of the control drum (16) and the signal terminal of the reactor (6) in sequence, and the power control system (4) drives the control drum (16) to adjust the reactivity; A first speed sensor is provided on the transmission shaft connecting the compressor (9) and the turbine (10), and a signal end of the first speed sensor is connected to a signal end of a transmission shaft speed control system (5), and the transmission shaft speed control system (5) adjusts the auxiliary torque of the transmission shaft; The compressor inlet temperature control system (1), the compressor inlet pressure control system (2), the core outlet temperature control system (3), the power control system (4) and the transmission shaft speed control system (5) are all equipped with a built-in PID neuron network decoupling controller, which is constructed into a 10×15×5 forward network structure to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

2. The decoupling control method of a decoupling control system of a supercritical carbon dioxide direct cycle nuclear power plant according to claim 1, characterized in that: The following steps are involved: Step 1: Constructing a control system for a supercritical carbon dioxide direct cycle nuclear power plant: a compressor inlet temperature control system (1) maintains a constant compressor (9) inlet temperature by adjusting the flow rate of the cooler (8) on the cold side; a compressor inlet pressure control system (2) maintains a constant compressor (9) inlet pressure by controlling the circulating working medium of the capacity box (12); a core outlet temperature control system (3) maintains a constant reactor (6) outlet temperature by adjusting the opening of a control valve (15); a power control system (4) achieves real-time matching of reactor power and load power by introducing reactivity through a drive control drum (16); the compressor (9), turbine (10) and generator (11) rotate coaxially, and a transmission shaft speed control system (5) maintains a constant transmission shaft speed by adjusting the transmission shaft auxiliary torque; Step 2: The compressor inlet temperature control system (1), the compressor inlet pressure control system (2), the core outlet temperature control system (3), the power control system (4) and the transmission shaft speed control system (5) in step 1 are constructed into a PID neural network decoupling control method with a 10×15×5 forward network structure; Step 3: Based on the PID neural network decoupling control method constructed in step 2, the beetle swarm algorithm is used to globally search for the optimal initial weights of the five control systems: compressor inlet temperature control system (1), compressor inlet pressure control system (2), core outlet temperature control system (3), power control system (4), and transmission shaft speed control system (5); Step 4: When the operating conditions of the supercritical carbon dioxide direct cycle nuclear power system change, in order to ensure the critical state of the working fluid at the compressor inlet, the PID neural network will take the difference between the output set value and the actual output value of the controlled system as input, and update the connection weights of the neurons in real time through the supervised Hebb learning rule, and adjust the strength of the proportional, integral, and differential effects in the system control process in real time, and finally obtain the input of the controlled system, complete the decoupling control of the supercritical carbon dioxide direct cycle nuclear power device, and convert the multivariable strong coupling system into multiple single-input and single-output systems, so as to achieve the control goal that each output parameter of the controlled system is only affected by a specific input parameter.

3. The decoupling control method according to claim 2, characterized in that: The step 2 is constructed into a 10×15×5 forward network structure, that is, the compressor inlet temperature control system (1), the compressor inlet pressure control system (2), the core outlet temperature control system (3), the power control system (4) and the transmission shaft speed control system (5) are five parallel sub-network structures, T c_ref 、P c_ref 、T core_ref 、N r_ref 、v ref are the input set values ​​of the PID neural network decoupling controller; KV1, KV2, KV3, v r , T are the input values ​​of the controlled system respectively; T c 、P c 、T core 、N r and v are the actual output values ​​of the controlled system respectively; w si is the connection weight from the hidden layer to the output layer; s is the serial number of the parallel sub-network, s = 1, 2, 3, 4, 5; i is the serial number of the hidden layer neuron, i = 1, 2, 3.

4. The decoupling control method according to claim 2, characterized in that: The specific algorithm process of the beetle swarm algorithm in step 3 is as follows: Initialize the parameters of the longhorn beetle group, set the number of longhorn beetles in the group, randomly assign the speed, position, and flight direction parameters of each longhorn beetle, and define the following fitness function: Where T is the maximum number of iterations of the beetle swarm algorithm, length is the number of PIDNN training times, n is the number of beetles in the beetle swarm, r is the target signal value, and y is the output of the system; The fitness value of each beetle's position in the beetle population is calculated based on the fitness function, and the optimal position of each beetle and the global optimal position are updated based on the fitness value of each beetle's position. Update the inertia weight: Where T is the maximum number of iterations, θ t is the inertia weight, θ max and θ min are the maximum and minimum values ​​of θ respectively; Update the speed of each longhorn beetle: Where, represents the position of the i-th longicorn in the t-th iteration of the q-th dimension, represents the speed of the i-th longhorn beetle in the t-th iteration in the q-th dimension, is the individual extreme position, is the extreme position of the population, b1 and b2 are two learning factors, z3 and z4 are random numbers in [0,1]; Update the increment function: Where, represents the increment function, δ t It represents the step size in the t-th iteration of the longhorn beetle, which decreases as the number of iterations increases and is updated according to the following formula: d t =z1δ t-1 +d 0 In the formula, z1 is a positive constant, δ 0 is the initial value of the step length, and the positions of the left and right tentacles of each longhorn beetle are updated as follows: Where, d t represents the distance between the two whiskers in the tth iteration; Update the position of each longhorn beetle: Where β is a positive constant. After multiple iterations, the beetle swarm algorithm completes the global search and obtains the global optimal solution.

5. The decoupling control method according to claim 2, characterized in that: The PID neural network in step 4 takes the difference between the output set value and the actual output value of the controlled system as input. The PID neural network input variables x1, x2, and x3 will be expressed in incremental form: According to the input variables x1, x2, and x3 of the PID neural network, the input quantity of the controlled system will be calculated as follows: Where K is the neuron proportional coefficient, K>0; w1, w2, w3 are the neuron connection weights, which are approximately equal to the proportional, integral, and differential coefficients of the PID controller.

6. The decoupling control method according to claim 2, characterized in that: The supervised Hebb learning rule in step 4 is an algorithm rule that selects the supervised Hebb learning rule as the connection weight, specifically: Where η p is the proportional learning rate; η i is the integrated learning rate; η d is the differential learning rate; The connection weights of the neural network are adjusted in real time through the supervised Hebb learning rule until the input error signal is close to zero. At this time, the input of the controlled system is the correct input. That is, by adjusting the strength of the proportional, integral and differential effects of the PID neural network, the mutual influence between different control channels is eliminated.

7. The decoupling control method according to claim 4, characterized in that: In step 4, the difference between the output set value and the actual output value of the controlled system is used as input. In the five control systems, namely, the compressor inlet temperature control system (1), the compressor inlet pressure control system (2), the core outlet temperature control system (3), the power control system (4), and the transmission shaft speed control system (5), the difference input and the PID neural network decoupling control method are specifically as follows: The compressor inlet temperature setting value of the compressor inlet temperature control system (1) is subtracted from the measurement value of the first temperature sensor, and the difference signal is sent to the PID neural network decoupling controller in the compressor inlet temperature control system (1); the decoupling controller sends its output signal to the first control valve (13); the compressor inlet pressure setting value of the compressor inlet pressure control system (2) is subtracted from the measurement value of the first pressure sensor, and the difference signal is sent to the PID neural network decoupling controller in the compressor inlet pressure control system (2); the decoupling controller sends its output signal to the second control valve (14) of the capacity box; the core outlet temperature setting value of the core outlet temperature control system (3) is subtracted from the measurement value of the second temperature sensor, The difference signal is sent to the PID neural network decoupling controller in the core outlet temperature control system (3); the decoupling controller sends its output signal to the third control valve (15); the power setting value of the power control system (4) is subtracted from the measured value of the first power sensor, and the difference signal is sent to the PID neural network decoupling controller in the power control system (4); the decoupling controller sends its output signal to the control drum (16); the transmission shaft speed setting value of the transmission shaft speed control system (5) is subtracted from the measured value of the first speed sensor, and the difference signal is sent to the PID neural network decoupling controller in the transmission shaft speed control system (5); the decoupling controller sends its output signal to the actuator to adjust the transmission shaft auxiliary torque.

Images