Multivariable fault-tolerant control method for maintaining thrust of liquid rocket engine

Abstract

The invention discloses a multivariable fault-tolerant control method for thrust maintenance of a liquid-propellant rocket engine, and the method comprises the steps: analyzing the fault mechanism and influence characteristics of the liquid-propellant rocket engine, and building a typical fault model; and by taking the mixing ratio as a constraint quantity, constructing a multivariable predictive optimization controller oriented to thrust maintenance, and designing a fault-tolerant control architecture to verify the performance of control logic under typical fault injection. According to the method, combustion chamber pressure reflecting thrust is taken as a controlled quantity, a mixing ratio is taken as a constraint quantity, a flow regulator valve core and a fuel main valve opening degree are taken as control quantities, offline training is performed based on a neural network prediction model, online rolling optimization is performed according to errors containing penalty functions and quadratic indexes of the control quantities, the control quantity which keeps better for the thrust is calculated, and the control quantity is optimized. And the maximum variable quantity of the mixing ratio is small, and the specific impulse is high. According to the method, transient loss and steady loss of thrust are reduced after the engine breaks down, the requirements of thrust keeping and mixing ratio constraint are met under the main-stage working condition, and the method has high fault-tolerant capacity for faults.

Description

Technical Field

[0001] The invention belongs to the field of liquid rocket engine control and simulation, and in particular relates to a multivariable fault-tolerant control method for maintaining thrust of a liquid rocket engine. Background Art

[0002] Liquid rocket engine (LRE) is the heart of space launch vehicle. It can convert the chemical energy of the liquid propellant into thermal energy in the thermal component, and convert the thermal energy into the kinetic energy of the ejected gas in the nozzle, thereby generating thrust. Due to the harsh working environment of the engine, various faults are easily induced to affect the rocket flight mission. Under different fault modes, the internal parameters of the engine will represent different change rules. Therefore, the research on the fault characteristics of liquid rocket engines is a hot topic at home and abroad. In order to analyze the fault characteristics of liquid rocket engines, in the case of the lack of engine test failure sample data due to the high cost of liquid rocket engine testing, conducting research on rocket engine fault modeling technology is the only way to improve the reliability and safety of the power system.

[0003] Thrust is the core parameter characterizing the state of a rocket engine. Its output is directly related to whether the rocket can fly safely and reliably, so it is one of the most important performance indicators of the engine. When an engine fails, the thrust will decline. At this time, the engine thrust maintenance fault-tolerant control technology will minimize the loss of thrust. The mixture ratio is defined as the ratio of the mass flow rate of the oxidizer and fuel consumed by the engine. During the thrust control process, the mixture ratio should not change too much, so as to control the overall propellant filling amount of the two tanks of the rocket and ensure the stability of the engine's own performance. Therefore, it is necessary to take the constraints of the mixture ratio into consideration to prevent the deviation of the mixture ratio from the design value.

[0004] Since the system of liquid rocket engines needs to maintain high working efficiency during the thrust control process, the mixture ratio remains stable during the entire working process, and the thrust and mixture ratio deviate from the design value after the engine fails, thrust maintenance and mixture ratio limitation need to be achieved, so high requirements are placed on the design of the engine control system. Conventional loop switching control methods use decentralized loops to form a multivariable control system, and do not consider the influence of coupling, which affects the thrust and mixture ratio maintenance effect to a certain extent. Therefore, in the context of frequent liquid rocket engine failures, it is necessary to study a reliable and safe engine thrust maintenance multivariable fault-tolerant control method based on typical fault modeling, eliminate the influence of coupling, take into account the control requirements of thrust and mixture ratio, and minimize the reliability and safety impact of engine failures. Summary of the invention

[0005] In view of the above technical problems, the present invention proposes a multivariable fault-tolerant control method for thrust maintenance of liquid rocket engines. Based on the research on the influence and characteristic analysis of typical engine faults, a typical fault model is established. Through offline training of the neural network prediction model, with the goal of minimizing the error and quadratic performance index of the control quantity containing the penalty function, a neural network optimization solution equation is established, and a thrust maintenance multivariable prediction optimization controller is designed. Finally, combined with the typical fault model, a thrust maintenance fault-tolerant control architecture based on the multivariable prediction optimization controller is constructed, and it is verified that the control method has better thrust maintenance and mixture ratio limitation effects under faults, and has better steady-state performance and dynamic performance.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance, comprising the following steps:

[0008] Step A), according to the typical failure modes of liquid rocket engines, determine the failure mechanisms of oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, combustion chamber gas leakage and characteristic parameters of failure influence, introduce failure factors to simulate failures for different typical failures, and establish a typical failure model of liquid rocket engines;

[0009] Step B), training a neural network prediction model with a data set consisting of an output state vector and an input control vector obtained according to a typical fault model of a liquid rocket engine, wherein the output state vector includes a mixture ratio, introducing a penalty function term for limiting the mixture ratio on the basis of an error and a quadratic index of a control quantity, performing a neural network optimization solution for thrust maintenance, and designing a thrust maintenance multivariable prediction optimization controller with high safety;

[0010] Step C), design a thrust maintenance fault-tolerant control architecture based on a multivariable predictive optimization controller, simulate the occurrence of typical faults such as oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, and combustion chamber gas leakage in a liquid rocket engine under the rated state of the main stage operating condition, and study and verify the thrust maintenance fault-tolerant control method and control performance.

[0011] As a further optimization scheme of the multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine of the present invention, the specific steps of step A) are as follows:

[0012] Step A1), by summarizing typical faults from the liquid rocket engine fault mode library, the fault influencing parameter characteristics of oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, and combustion chamber gas leakage faults are analyzed from the fault mechanism level. The analysis results are as follows: the influencing parameter characteristics of oxygen pre-compression turbine blade ablation fault are a decrease in turbine efficiency and speed, the influencing parameter characteristics of oxygen main pump cavitation fault are a decrease in pump head and pressure after the pump, and the influencing parameter characteristics of combustion chamber gas leakage fault are an increase in component outflow flow and a decrease in component pressure;

[0013] Step A2), establishing a typical engine fault model by introducing a fault factor parameter F to modify a mathematical model parameter D of an engine component, wherein F represents the severity of a fault of an engine component;

[0014] For oxygen pre-compression turbine blade ablation failure, due to the turbine efficiency η t A drop occurs, and the fault model is mathematically expressed as:

[0015]

[0016] Among them, η t (t) is the normal value of turbine efficiency, η′ t (t) is the actual value of turbine efficiency, t is the operating time, t f is the fault occurrence time, F1 is the turbine fault factor, when 0≤F1<1, it indicates that the turbine has a blade ablation fault; when F1=1, the turbine works normally;

[0017] For the oxygen main pump cavitation fault, due to the decrease in pump head ΔP, the fault model is mathematically expressed as:

[0018]

[0019] Among them, ΔP(t) is the normal value of the pump head, ΔP′(t) is the actual value of the pump head, and F2 is the pump failure factor. When 0≤F2<1, it indicates that the pump has a cavitation failure; when F2=1, the pump works normally;

[0020] For the combustion chamber gas leakage fault, since the outflow flow of the component in the mass conservation equation is equal to the sum of the component outlet flow and the leakage flow, that is, the component loses an additional part of the leakage flow, the fault model is mathematically expressed as:

[0021]

[0022] Among them, q ig (t), q lo (t), q lf (t) are the normal values ​​of the gas mass flow rate, liquid oxidant mass flow rate and liquid fuel mass flow rate flowing into the thermal component, respectively, and qeg (t) is the normal value of the gas mass flow rate out of the thermal component, is the actual value of the mass change rate of the working fluid of the thermal component, F3 is the failure factor of the thermal component, when F3>0, it indicates that a gas leakage failure has occurred in the thermal component; when F3=0, the thermal component works normally.

[0023] As a further optimization scheme of the multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine of the present invention, the specific steps of step B) are as follows:

[0024] Step B1), select the control variables as the flow regulator valve core opening α, the fuel main valve opening τ f ; The controlled variable is selected as the combustion chamber pressure P reflecting the thrust c ; Select output constraint variables including mixing ratio K c , then the engine multi-input and multi-output discretization system is expressed as:

[0025] x(t+1)=f(x(t),u(t))

[0026] Where x(t) = [P c K c ] T As the output state vector, u(t) = [α τ f ] T is the input control vector, where P c Denoted as x1(t), K c Denoted as x2(t);

[0027] The neural network prediction model training selects the current state vector x(t) and control vector u(t), and the output is the state vector x(t+1) at the next moment. The number of neurons in the input layer, hidden layer, and output layer are 4, 20, and 2, respectively. The training uses a typical fault model of a liquid rocket engine to generate 25,000 sets of training samples of the flow regulator valve core opening α and the fuel main valve opening τ near the rated state of the main stage operating condition. f And the combustion chamber pressure P c , mixing ratio K c Normalized dataset.

[0028] Step B2), construct the neural network optimization solver equation as

[0029] u(t)=g(x(t),r(t),W)

[0030] Among them, r(t) is the system instruction input value at time t, W is the solver weight matrix; the solver adopts a three-layer network structure, with the input layer being x(t) and r(t), and the output layer being u(t), which is used to recursively calculate the predicted control quantity u' and the predicted output quantity x' in the prediction time domain;

[0031] For error vector and control vector quadratic performance indicators:

[0032]

[0033] Where, e(t) = x1(t) - r(t), P is the controller prediction time domain, e(t1+P) T Ze(t1+P) is the terminal error vector performance index, Q, R, and Z are the error vector, control vector, and terminal value error vector weighting coefficient matrices in the quadratic performance index functional, respectively, and are taken as diagonal matrices;

[0034] Let e(t) T Qe(t)+u(t) T Ru(t)=L(t), considering that the initial value x(t1) of the rolling optimization state is equal to the final value x'(t1'+P) of the rolling optimization state in the previous stage, the performance index augmentation function is obtained by introducing λ and q according to the Lagrange multiplier method:

[0035]

[0036] Among them, λ and q are Lagrange multipliers;

[0037] In order to find the control sequence that makes J as small as possible, it is necessary to recursively solve λ and q:

[0038] For t = t1 + P:

[0039]

[0040] For t = t1+P-1,…,t1+2,t1+1:

[0041]

[0042]

[0043] Continuously update the weights of the neural network optimization solver until ΔW = 0:

[0044]

[0045] W'=W+ΔW

[0046] Among them, α is the learning rate, and W' is the modified weight matrix. After the weight update is completed, the controller uses the neural network optimization solver equation calculation to only apply the actual control amount at the current moment to the liquid rocket engine, forming a rolling optimization.

[0047] Step B3), in the controller solution, the corresponding maximum and minimum constraint penalty function terms are introduced into the error vector and the control vector quadratic performance index, and the performance index becomes:

[0048]

[0049] At this time, L(t) becomes:

[0050] L(t)=e(t) T Qe(t)+u(t) T Ru(t)+σ1[max(x2(t)-x max ,0)] T [max(x2(t)-x max ,0)]

[0051] +σ2[-min(x2(t)-x min ,0)] T [-min(x2(t)-x min ,0)]

[0052] Among them, σ1 and σ2 represent the penalty factors exceeding the upper threshold and the lower threshold respectively; x max 、x min Respectively represent the mixing ratio K c The upper and lower limit thresholds are x max =1.08,x min =0.92.

[0053] As a further optimization scheme of the multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine of the present invention, the specific steps of step C) are as follows:

[0054] Step C1), based on the proposed liquid rocket engine thrust maintenance multivariable predictive optimization controller, combined with the established typical fault model, a thrust maintenance multivariable predictive optimization fault-tolerant control structure for typical faults is designed;

[0055] The control structure specifically includes three parts: the establishment of a typical engine fault model, a thrust-maintaining multivariable predictive optimization controller, and a liquid rocket engine object. The thrust-maintaining multivariable predictive optimization controller mainly includes a thrust-maintaining optimization solver, a prediction model, and a feedback correction module.

[0056] By giving the thrust command of the liquid rocket engine, the thrust maintenance optimization solver solves the corresponding combustion chamber pressure command. According to the predicted output x'(t+1), x'(t+2), ..., x'(t+P) within P steps given by the prediction model, the error of the penalty function and the quadratic index J of the control vector are added to optimize the solution. The optimal control quantity sequence u'(t), u'(t+1), ..., u'(t+P-1) within P steps is calculated and input into the prediction model. Only the first control quantity u(t) in the control quantity sequence, i.e. the current moment, is used. The liquid rocket engine actuator controls the engine propellant flow rate according to the flow regulator valve core opening and fuel main valve opening signals given by the controller. Under the access of the typical fault model, the engine output state quantity x is fed back to the thrust maintenance optimization solver and the feedback correction module in real time. The feedback correction module corrects the prediction model by calculating the difference between the actual output x(t+1) and the predicted output x'(t+1), and then performs a new optimization solution, so that the optimization result is not only based on the model, but also utilizes the feedback information of the engine.

[0057] Step C2), under the rated state of the main stage of the liquid rocket engine, based on the established fault model, the injection of typical faults such as oxygen pre-compression turbine blade erosion, oxygen main pump cavitation, and combustion chamber gas leakage are simulated respectively, and the loop switching PI thrust control method is used as a reference to study and verify the thrust maintenance fault-tolerant control architecture and control performance.

[0058] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0059] (1) The present invention proposes a liquid rocket engine thrust maintenance multivariable controller, which can be combined with an established engine typical fault model to form a thrust maintenance fault-tolerant control architecture based on a multivariable predictive optimization controller. The control architecture has strong fault tolerance for engine faults and has good steady-state and dynamic performance.

[0060] (2) The present invention proposes a multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine, which takes the mixture ratio as an additional constraint and uses a neural network to construct a prediction model and an optimization solver. This method can eliminate the coupling effect, has a strong thrust maintenance and mixture ratio limitation effect after a fault occurs, and makes the engine specific impulse higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a structure diagram of a multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine;

[0062] Figure 2 It is a schematic diagram of the method of establishing the ablation failure model of the oxygen pre-compression turbine blade of a liquid rocket engine;

[0063] Figure 3 It is a schematic diagram of the method of establishing the cavitation fault model of the oxygen main pump of a liquid rocket engine;

[0064] Figure 4 It is a schematic diagram of the method of establishing a gas leakage fault model in the combustion chamber of a liquid rocket engine;

[0065] Figure 5 is the neural network prediction model training error graph;

[0066] Figure 6 This is a comparison diagram of the thrust maintenance control effect of a liquid rocket engine under oxygen pre-compression turbine blade ablation failure, including (a) flow regulator valve core opening, (b) fuel main valve opening, (c) combustion chamber pressure, (d) thrust, (e) mixture ratio, (f) specific impulse, (g) liquid oxygen main valve flow, fuel main valve flow, (h) main turbine speed, oxygen main pump after-pump pressure, (i) fuel first-stage pump, fuel second-stage pump after-pump pressure, (j) gas generator pressure;

[0067] Figure 7 This is a comparison diagram of the thrust maintenance control effect under the cavitation fault of the oxygen main pump of a liquid rocket engine, including (a) flow regulator valve core opening, (b) fuel main valve opening, (c) combustion chamber pressure, (d) thrust, (e) mixture ratio, (f) specific impulse, (g) liquid oxygen main valve flow, fuel main valve flow, (h) main turbine speed, oxygen main pump after-pump pressure, (i) fuel first-stage pump, fuel second-stage pump after-pump pressure, (j) gas generator pressure;

[0068] Figure 8 This is a comparison chart of the thrust maintenance control effect under the gas leakage fault of the liquid rocket engine combustion chamber, including (a) flow regulator valve core opening, (b) fuel main valve opening, (c) combustion chamber pressure, (d) thrust, (e) mixture ratio, (f) specific impulse, (g) liquid oxygen main valve flow, fuel main valve flow, (h) main turbine speed, oxygen main pump after-pump pressure, (i) fuel first-stage pump, fuel second-stage pump after-pump pressure, (j) gas generator pressure. DETAILED DESCRIPTION

[0069] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0070] The specific implementation mode of the present invention takes the design of a multivariable fault-tolerant control method for thrust maintenance of a certain type of liquid oxygen / kerosene staged combustion cycle liquid rocket engine of the Long March series of launch vehicles as an example. The control structure is as follows: Figure 1 As shown, the specific steps include:

[0071] Step A), according to the typical failure modes of liquid rocket engines, determine the failure mechanisms of oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, combustion chamber gas leakage and characteristic parameters of failure influence, introduce failure factors to simulate failures for different typical failures, and establish a typical failure model of liquid rocket engines;

[0072] Step B), training a neural network prediction model with a data set consisting of an output state vector and an input control vector obtained according to a typical fault model of a liquid rocket engine, wherein the output state vector includes a mixture ratio, introducing a penalty function term for limiting the mixture ratio on the basis of an error and a quadratic index of a control quantity, performing a neural network optimization solution for thrust maintenance, and designing a thrust maintenance multivariable prediction optimization controller with high safety;

[0073] Step C), design a thrust maintenance fault-tolerant control architecture based on a multivariable predictive optimization controller, simulate the occurrence of typical faults such as oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, and combustion chamber gas leakage in a liquid rocket engine under the rated state of the main stage operating condition, and study and verify the thrust maintenance fault-tolerant control method and control performance.

[0074] The detailed steps of step A) are as follows:

[0075] Step A1), from the liquid rocket engine multiple failure mode library, turbine blade ablation, pump cavitation and thermal component gas leakage failures occurring in the turbine pump system and thermal component system are selected as typical failures for study, and the failure mechanism and influencing parameter characteristics are analyzed. The results are shown in the following table:

[0076] Table 1 Characteristics of parameters affecting typical failure modes of liquid rocket engines

[0077]

[0078] It can be seen from the table that the influencing parameter characteristics of oxygen pre-compression turbine blade ablation fault are the decrease in turbine efficiency and speed, the influencing parameter characteristics of oxygen main pump cavitation fault are the decrease in pump head and pressure after the pump, and the influencing parameter characteristics of combustion chamber gas leakage fault are the increase in component outflow flow and the decrease in component pressure. Based on the influence of each typical fault mode on the change of engine parameters, the typical engine fault model is established;

[0079] Step A2), establishing a typical engine fault model by introducing a fault factor parameter F to modify a mathematical model parameter D of an engine component, wherein F represents the severity of a fault of an engine component;

[0080] For oxygen pre-compression turbine blade ablation failure, based on the fault influencing parameter characteristic analysis, due to the turbine efficiency η t When a drop occurs, the fault model is established as follows Figure 2As shown, the fault model is mathematically described as:

[0081]

[0082] Among them, η t (t) is the normal value of turbine efficiency, η′ t (t) is the actual value of turbine efficiency, t is the operating time, t f is the fault occurrence time, F1 is the turbine fault factor, when 0≤F1<1, it indicates that the turbine has a blade ablation fault; when F1=1, the turbine works normally;

[0083] For the oxygen main pump cavitation fault, based on the fault influencing parameter characteristic analysis, due to the decrease in pump head ΔP, the fault model is established as follows: Figure 3 As shown, the fault model is mathematically described as:

[0084]

[0085] Among them, ΔP(t) is the normal value of the pump head, ΔP′(t) is the actual value of the pump head, and F2 is the pump failure factor. When 0≤F2<1, it indicates that the pump has a cavitation failure; when F2=1, the pump works normally;

[0086] For the combustion chamber gas leakage fault, based on the fault influencing parameter characteristic analysis, since the component outflow flow in the mass conservation equation is equal to the sum of the component outlet flow and the leakage flow, that is, the component loses an additional part of the leakage flow, the fault model is established as follows Figure 4 As shown, the fault model is mathematically described as:

[0087]

[0088] Among them, q ig (t), q lo (t), q lf (t) are the normal values ​​of the gas mass flow rate, liquid oxidant mass flow rate and liquid fuel mass flow rate flowing into the thermal component, respectively, and q eg (t) is the normal value of the gas mass flow rate out of the thermal component, is the actual value of the mass change rate of the working fluid of the thermal component, F3 is the failure factor of the thermal component, when F3>0, it indicates that a gas leakage failure has occurred in the thermal component; when F3=0, the thermal component works normally.

[0089] The detailed steps of step B) are as follows:

[0090] Step B1), adjust the engine flow regulator valve core opening α and the fuel main valve opening τ f As the control variable; the combustion chamber pressure P reflecting the thrust c As the controlled variable; mixing ratio Kc As the output constraint variable, for the engine multi-input and multi-output discretization system:

[0091] x(t+1)=f(x(t),u(t))

[0092] Where x(t) = [P c K c ] T As the output state vector, u(t) = [α τ f ] T is the input control vector, where P c Denoted as x1(t), K c Denoted as x2(t);

[0093] The neural network prediction model training adopts a single hidden layer feedforward BP neural network structure. The network input is the current state vector x(t) and control vector u(t) in the above formula, and the output is the state vector x(t+1) at the next moment. The number of neurons in the input layer, hidden layer and output layer are 4, 20 and 2 respectively. The training first uses the typical fault model of liquid rocket engine to generate 25000 sets of training samples of flow regulator valve core opening α and fuel main valve opening τ near the rated state of the main stage working condition. f And the combustion chamber pressure P c , mixing ratio K c The normalized data set is ; the learning rate is set to 0.05 and the momentum coefficient is set to 0.1 for training. The neural network training error is as follows Figure 5 As shown in the figure, it can be seen that when the iteration reaches 378 times, the training error reaches 9.985×10 -9 , it is believed that the neural network training converges;

[0094] Step B2), establish the neural network optimization solver equation as

[0095] u(t)=g(x(t),r(t),W)

[0096] Among them, r(t) is the system command input value at time t, and W is the solver weight matrix; the solver adopts a single hidden layer feedforward BP neural network structure, with the input layer being x(t) and r(t), and the output layer being u(t), which is used to recursively calculate the predicted control quantity u' and the predicted output quantity x' in the prediction time domain;

[0097] For error vector and control vector quadratic performance indicators:

[0098]

[0099] Where, e(t) = x1(t) - r(t), P is the controller prediction time domain, e(t1+P) TZe(t1+P) is the terminal error vector performance index, Q, R, and Z are the error vector, control vector, and terminal value error vector weighting coefficient matrices in the quadratic performance index functional, respectively, and are taken as diagonal matrices;

[0100] Let e(t) T Qe(t)+u(t) T Ru(t)=L(t), considering that the initial value x(t1) of the rolling optimization state is equal to the final value x'(t1'+P) of the rolling optimization state in the previous stage, the performance index augmentation function is obtained by introducing λ and q according to the Lagrange multiplier method:

[0101]

[0102] Among them, λ and q are Lagrange multipliers;

[0103] In order to find the control sequence that makes J as small as possible, it is necessary to recursively solve λ and q:

[0104] For t = t1 + P:

[0105]

[0106] For t = t1+P-1,…,t1+2,t1+1:

[0107]

[0108]

[0109] Continuously update the weights of the neural network optimization solver until ΔW = 0:

[0110]

[0111] W'=W+ΔW

[0112] Among them, α is the learning rate, and W' is the modified weight matrix. After the weight update is completed, the controller uses the neural network optimization solver equation calculation to only apply the actual control amount at the current moment to the liquid rocket engine, forming a rolling optimization.

[0113] Step B3), in order to deal with the output constraint, the mixing ratio K c The predicted output is limited by the limit threshold in the prediction time domain. In the controller solution, the corresponding maximum and minimum constraint penalty function terms are introduced into the error vector and control vector quadratic performance index, and the performance index becomes:

[0114]

[0115] At this time, L(t) becomes:

[0116] L(t)=e(t) T Qe(t)+u(t) T Ru(t)+σ1[max(x2(t)-x max ,0)] T [max(x2(t)-x max ,0)]+σ2[-min(x2(t)-x min ,0)] T [-min(x2(t)-x min ,0)]

[0117] Among them, σ1 and σ2 represent the penalty factors for exceeding the upper threshold and the lower threshold, respectively, which are large positive numbers and usually have values ​​more than two orders of magnitude larger than other weighting coefficients; x max 、x min Respectively represent the mixing ratio K c The upper and lower limit thresholds are set according to the adjustable range of the mixture ratio of the liquid rocket engine. max =1.08,x min =0.92. After introducing the penalty function term to the performance index, at each moment in the entire prediction time domain, if the predicted output is within the limit threshold, the penalty function term is 0, which has no effect on the optimization of the performance index J; if the predicted output is outside the limit threshold, the penalty function term increases significantly, causing the performance index J to increase significantly, and the controller aims to solve the control amount that makes the performance index J extremely small, so that the output is limited within its threshold. Therefore, the controller can monitor the mixing ratio K in real time. c Perform limit control, that is, when receiving a trend signal that the mixture ratio exceeds the limit threshold, the controller will solve the control quantity to avoid the mixture ratio exceeding the threshold, so that the deviation between the mixture ratio and the design value is small, ensuring the performance stability and working safety of the engine.

[0118] The detailed steps of step C) are as follows:

[0119] Step C1), based on the proposed liquid rocket engine thrust maintenance multivariable predictive optimization controller, combined with the established typical fault model, a thrust maintenance multivariable predictive optimization fault-tolerant control structure for typical faults is designed, such as Figure 1 As shown;

[0120] As can be seen from the figure, the thrust maintenance fault-tolerant control architecture based on the multivariable predictive optimization controller consists of three parts, namely the establishment of the typical engine fault model, the thrust maintenance multivariable predictive optimization controller and the liquid rocket engine object. The thrust maintenance multivariable predictive optimization controller mainly includes the thrust maintenance optimization solver, the prediction model and the feedback correction module.

[0121] By giving the thrust command of the liquid rocket engine, the thrust maintenance optimization solver solves the corresponding combustion chamber pressure command. According to the predicted output x'(t+1), x'(t+2), ..., x'(t+P) within P steps given by the prediction model, the error of the penalty function and the quadratic index J of the control vector are added to optimize the solution. The optimal control quantity sequence u'(t), u'(t+1), ..., u'(t+P-1) within P steps is calculated and input into the prediction model. Only the first control quantity u(t) in the control quantity sequence, i.e. the current moment, is used. The liquid rocket engine actuator controls the engine propellant flow rate according to the flow regulator valve core opening and fuel main valve opening signals given by the controller. Under the access of the typical fault model, the engine output state quantity x is fed back to the thrust maintenance optimization solver and the feedback correction module in real time. The feedback correction module corrects the prediction model by calculating the difference between the actual output x(t+1) and the predicted output x'(t+1), and then performs a new optimization solution, so that the optimization result is not only based on the model, but also utilizes the feedback information of the engine.

[0122] Step C2), under the rated working condition of the main stage of the liquid rocket engine, based on the established fault model, such as Figure 2 , Figure 3 , Figure 4 The following figure shows the simulation of the injection of typical faults of oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, and combustion chamber gas leakage, and the reference of the loop switching PI thrust control method is used to study and verify the thrust maintenance fault-tolerant control architecture and control performance;

[0123] The loop switching PI control method involved in the comparison adopts a decentralized loop to form a multivariable control system, which is composed of a thrust PI single variable control loop and a mixture ratio PI single variable control loop. The loop switching logic is that when the mixture ratio exceeds the limit, the thrust single variable control loop is switched to the thrust and mixture ratio decentralized control loop;

[0124] like Figure 2 As shown in the figure, oxygen is injected into the engine at 60 seconds to pre-compress the turbine blade ablation fault, and the fault factor F1 is set to 0.9. The simulation results are as follows Figure 6 The performance index results are shown in Table 2. It can be seen that after the oxygen pre-compression turbine blade ablation failure occurs in the engine, the present invention and the loop switching control can both reduce the steady-state loss of thrust under the fault condition. Since the mixture ratio is not exceeded, the loop switching control only controls the thrust single variable with the flow regulator, while the present invention optimizes based on the thrust maintenance target, eliminates the coupling effect, and solves two control quantities as the engine input, so the thrust transient loss is smaller, the time required to recover to the rated value is shorter, and the absolute value of the maximum change of the mixture ratio is smaller, and the specific impulse is higher.

[0125] Table 2 Comparison of thrust retention effect under typical faults of a certain type of liquid rocket engine main stage rated operating conditions

[0126]

[0127] like Figure 3 As shown in the figure, at 60s, the engine oxygen injection main pump cavitation fault is detected, and the fault factor F2 is set to 0.95. The simulation results are as follows Figure 7 The performance index results are shown in Table 2. It can be seen that after the engine has an oxygen main pump cavitation fault, the present invention and the loop switching control can both reduce the steady-state loss of thrust in the event of a fault, wherein the loop switching control switches to the limiting loop to control the mixture ratio after the mixture ratio exceeds the threshold, and the present invention solves the online rolling optimization solution with a mixture ratio constraint, and solves the control quantity output that is better for thrust maintenance and mixture ratio limitation to act on the engine, so the thrust transient loss is smaller, the thrust recovery is faster, the mixture ratio always keeps a distance from the boundary, the absolute value of the maximum change of the mixture ratio is smaller, and the specific impulse is also higher.

[0128] like Figure 4 As shown in the figure, at 60s, the engine is injected with combustion chamber gas leakage fault, and the fault factor F3 is set to 0.3. The simulation results are as follows Figure 8 As shown, the performance index results are shown in Table 2. It can be seen that after the combustion chamber gas leakage fault occurs in the engine, since the present invention uses a neural network to construct a prediction model and an optimization solver, and the neural network has the characteristics of strong robustness, the prediction model feedback correction based on the neural network and the rolling implementation of the optimization solver can make up for the uncertainty caused by model mismatch and external disturbance in time. Therefore, when the engine has a more serious thrust decline fault, the control method proposed by the present invention shows a strong fault tolerance. In the control process, the specific performance of the present invention is that compared with the loop switching control, the transient loss and steady-state loss of thrust are smaller, the maximum deviation of the mixture ratio from the design value is also smaller, and the specific impulse is also higher.

[0129] It can be seen that, thanks to the establishment of the neural network prediction model and the optimization solver and the constrained online rolling optimization strategy, the multivariable fault-tolerant control method for thrust maintenance of a liquid rocket engine proposed in the present invention is more effective for thrust maintenance under typical engine failures than conventional methods, and the transient thrust loss is reduced by 10% to 40%. The deviation of the engine's mixture ratio from the design value is smaller, and the specific impulse is higher, which can achieve performance maintenance and working safety under typical engine failures.

[0130] Based on the characteristic analysis of the influencing parameters of the typical failure mode of liquid rocket engines, the present invention establishes a typical failure model of the engine; the combustion chamber pressure P c is the controlled quantity, the mixing ratio K cis the constraint amount, the flow regulator valve core and fuel main valve opening α, τ f For the control quantity, a multivariable predictive optimization controller for thrust maintenance is established through offline training of a neural network prediction model, and optimization of a neural network based on errors and quadratic indicators of control quantity with penalty functions; a multivariable fault-tolerant control architecture for thrust maintenance is proposed by combining a typical fault model with the controller. Simulations show that the present invention eliminates the influence of coupling, has better thrust maintenance and mixture ratio limiting effects under faults, has smaller transient and steady-state thrust losses, has stronger fault tolerance, and makes the engine specific impulse relatively high.

[0131] It should be pointed out that the above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes and substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance, characterized in that: The following steps are involved: Step A), according to the typical failure modes of liquid rocket engines, determine the failure mechanisms of oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, combustion chamber gas leakage and characteristic parameters of failure influence, introduce failure factors to simulate failures for different typical failures, and establish a typical failure model of liquid rocket engines; Step B), training a neural network prediction model with a data set consisting of an output state vector and an input control vector obtained according to a typical fault model of a liquid rocket engine, wherein the output state vector includes a mixture ratio, introducing a penalty function term for limiting the mixture ratio on the basis of an error and a quadratic index of a control quantity, performing a neural network optimization solution for thrust maintenance, and designing a thrust maintenance multivariable prediction optimization controller with high safety; Step C), design a thrust maintenance fault-tolerant control architecture based on a multivariable predictive optimization controller, simulate the occurrence of typical faults such as oxygen pre-compression turbine blade ablation, oxygen main pump cavitation, and combustion chamber gas leakage in a liquid rocket engine under the rated state of the main stage operating condition, and study and verify the thrust maintenance fault-tolerant control method and control performance.

2. A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance as claimed in claim 1, characterized in that: The influencing parameter characteristics of oxygen pre-compression turbine blade ablation failure are the decrease in turbine efficiency and speed, the influencing parameter characteristics of oxygen main pump cavitation failure are the decrease in pump head and pressure after the pump, and the influencing parameter characteristics of combustion chamber gas leakage failure are the increase in component outflow flow and the decrease in component pressure.

3. A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance as claimed in claim 2, characterized in that: The typical engine fault model is established by introducing the fault factor parameter F to modify the mathematical model parameter D of the engine component, where F represents the severity of the fault of the engine component. For oxygen pre-compression turbine blade ablation failure, due to the turbine efficiency η t A drop occurs, and the fault model is mathematically described as: Among them, η t (t) is the normal value of turbine efficiency, η′ t (t) is the actual value of turbine efficiency, t is the operating time, t f is the fault occurrence time, F1 is the turbine fault factor, when 0≤F1<1, it indicates that the turbine has a blade ablation fault; when F1=1, the turbine works normally; For the oxygen main pump cavitation fault, due to the decrease in pump head ΔP, the fault model is mathematically described as: Among them, ΔP(t) is the normal value of the pump head, ΔP′(t) is the actual value of the pump head, and F2 is the pump failure factor. When 0≤F2<1, it indicates that the pump has a cavitation failure; when F2=1, the pump works normally; For combustion chamber gas leakage failure, since the outflow rate of the component in the mass conservation equation is equal to the sum of the component outlet flow and the leakage flow, that is, the component loses an additional part of the leakage flow, the fault model is mathematically described as: Among them, q ig (t), q lo (t), q lf (t) are the normal values ​​of the gas mass flow rate, liquid oxidant mass flow rate and liquid fuel mass flow rate flowing into the thermal component, respectively, and q eg (t) is the normal value of the gas mass flow rate out of the thermal component, is the actual value of the mass change rate of the working fluid of the thermal component, F3 is the failure factor of the thermal component, when F3>0, it indicates that a gas leakage failure has occurred in the thermal component; when F3=0, the thermal component works normally.

4. A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance as claimed in claim 1, characterized in that: The specific steps of step B) are as follows: Step B1), select the control variables as the flow regulator valve core opening α, the fuel main valve opening τ f ; The controlled variable is selected as the combustion chamber pressure P reflecting the thrust c ; Select output constraint variables including mixing ratio K c , then the engine multi-input and multi-output discretization system is expressed as: x(t+1)=f(x(t),u(t)) Where x(t) = [P c K c ] T As the output state vector, u(t) = [α τ f ] T is the input control vector, where P c Denoted as x1(t), K c Denoted as x2(t); The neural network prediction model training selects the current state vector x(t) and control vector u(t), and outputs the state vector x(t+1) at the next moment; the training uses the typical fault model of liquid rocket engines to generate several groups of training samples of the flow regulator valve core opening α and the fuel main valve opening τ near the rated state of the main stage working condition. f And the combustion chamber pressure P c , mixing ratio K c Normalized dataset of; Step B2), construct the neural network optimization solver equation as u(t)=g(x(t),r(t),W) Among them, r(t) is the system command input value at time t, and W is the solver weight matrix; the solver adopts a single hidden layer feedforward BP neural network structure, with the input layer being x(t) and r(t), and the output layer being u(t), which is used to recursively calculate the predicted control quantity u' and the predicted output quantity x' in the prediction time domain; For error vector and control vector quadratic performance indicators: Where, e(t) = x1(t) - r(t), P is the controller prediction time domain, e(t1+P) T Ze(t1+P) is the terminal error vector performance index, Q, R, and Z are the error vector, control vector, and terminal value error vector weighting coefficient matrices in the quadratic performance index functional, respectively, and are taken as diagonal matrices; Let e(t) T Qe(t)+u(t) T Ru(t)=L(t), considering that the initial value x(t1) of the rolling optimization state is equal to the final value x'(t1'+P) of the rolling optimization state in the previous stage, the performance index augmentation function is obtained by introducing λ and q according to the Lagrange multiplier method: Among them, λ and q are Lagrange multipliers; In order to find the control sequence that makes J as small as possible, it is necessary to recursively solve λ and q: For t = t1 + P: For t = t1+P-1,…,t1+2,t1+1: Continuously update the weights of the neural network optimization solver until ΔW = 0: W'=W+ΔW Among them, α is the learning rate, and W' is the modified weight matrix. After the weight update is completed, the controller uses the neural network optimization solver equation calculation to only apply the actual control amount at the current moment to the liquid rocket engine, forming a rolling optimization. Step B3), in the controller solution, the corresponding maximum and minimum constraint penalty function terms are introduced into the error vector and the control vector quadratic performance index, and the performance index becomes: At this time, L(t) becomes: L(t)=e(t) T Qe(t)+u(t) T Ru(t)+σ1[max(x2(t)-x max ,0)] T [max(x2(t)-x max ,0)]+σ2[-min(x2(t)-x min ,0)] T [-min(x2(t)-x min ,0)] Among them, σ1 and σ2 represent the penalty factors exceeding the upper threshold and the lower threshold respectively; x max 、x min Respectively represent the mixing ratio K c The upper and lower limit thresholds.

5. A multivariable fault-tolerant control method for liquid rocket engine thrust maintenance as claimed in claim 1, characterized in that: The specific steps of step C) are as follows: Step C1), based on the proposed liquid rocket engine thrust maintenance multivariable predictive optimization controller, combined with the established typical fault model, a thrust maintenance multivariable predictive optimization fault-tolerant control structure for typical faults is designed; The control structure specifically includes three parts: the establishment of a typical engine fault model, a thrust-maintaining multivariable predictive optimization controller, and a liquid rocket engine object. The thrust-maintaining multivariable predictive optimization controller mainly includes a thrust-maintaining optimization solver, a prediction model, and a feedback correction module. By giving the thrust command of the liquid rocket engine, the thrust maintenance optimization solver solves the corresponding combustion chamber pressure command. According to the predicted output x'(t+1), x'(t+2), ..., x'(t+P) within P steps given by the prediction model, the error of the penalty function and the quadratic index J of the control vector are added to optimize the solution. The optimal control quantity sequence u'(t), u'(t+1), ..., u'(t+P-1) within P steps is calculated and input into the prediction model. Only the first control quantity u(t) in the control quantity sequence, i.e. the current moment, is used. The liquid rocket engine actuator controls the engine propellant flow rate according to the flow regulator valve core opening and fuel main valve opening signals given by the controller. Under the access of the typical fault model, the engine output state quantity x is fed back to the thrust maintenance optimization solver and the feedback correction module in real time. The feedback correction module corrects the prediction model by calculating the difference between the actual output x(t+1) and the predicted output x'(t+1), and then performs a new optimization solution, so that the optimization result is not only based on the model, but also utilizes the feedback information of the engine. Step C2), under the rated state of the main stage of the liquid rocket engine, based on the established fault model, the injection of typical faults such as oxygen pre-compression turbine blade erosion, oxygen main pump cavitation, and combustion chamber gas leakage are simulated respectively, and the loop switching PI thrust control method is used as a reference to study and verify the thrust maintenance fault-tolerant control architecture and control performance.

Images