Solid engine shell dynamic loading self-adaptive control method and system

Abstract

The invention discloses a solid engine shell dynamic loading self-adaptive control method and system, and the method comprises the steps: building an element-level model and a mathematical analysis model of a dynamic loading device for a multi-cylinder combined hydraulic loading device and a shell test piece combination; introducing uncertain factors including friction, leakage, pressure fluctuation and the like into the element-level model to simulate a real loading device, taking the real loading device as a test platform, and establishing a joint simulation interface for signal interaction between the test platform and the controller; based on the mathematical analysis model, generating a reference input signal by using a nonlinear sliding mode surface introducing an error integral power function and an exponential reaching rate to limit error convergence time, and establishing a reference model; and a neural network sliding mode controller is used to generate a compensation signal for uncertain factors existing in the test platform so as to eliminate a force output error between the reference model and the test platform. According to the invention, a predetermined output force track can be rapidly tracked, the test cost is effectively reduced, and the overall control performance is improved.

Description

Technical Field

[0001] The invention belongs to the field of hydraulic system control simulation, and in particular relates to a method and system for adaptively controlling dynamic loading of a solid engine casing. Background Art

[0002] Solid engines are the power source of missile equipment. Their power characteristics are directly related to the flight performance, maneuverability, and high-speed penetration capability of weapons and equipment. They are the premise and foundation for the combat capability of weapons, and are the core technology to promote the upgrading of missile weapons. As the main load-bearing component of the engine, the composite shell not only needs to withstand the internal pressure of the combustion chamber of more than 10MPa, but also needs to withstand the load from the nozzle vector control force and various loads of the engine during launch and flight. Faced with the continuous improvement of the actual combat capability requirements of solid engines, the mechanical loads faced by the engine shell are more complex. Therefore, it is necessary to carry out dynamic loading test research simulating the real environment.

[0003] Although we now have a solid engine structural strength joint loading test system that can carry out solid engine structural strength tests, and have conventional engine structural performance assessment conditions and conventional assessment and verification capabilities for engine structural performance. In order to improve the dynamic loading simulation capability, the design of the control algorithm has become a key breakthrough point. During the dynamic loading process, the carefully designed control algorithm can not only effectively handle the complex nonlinear and uncertainty problems of the system, but also accurately predict and feedback control the system response under various load conditions. It can significantly enhance the functions of the simulation verification and analysis software platform, optimize the dynamic loading simulation system, and improve the accuracy and effectiveness of the loading device performance test, thereby ensuring the comprehensiveness and reliability of the test verification. Therefore, the innovative design of the control algorithm is of considerable theoretical and practical importance in improving the overall technical level and improving the dynamic loading simulation technology of solid engines. Summary of the invention

[0004] The purpose of the present invention is to address the above problems and propose a solid motor casing dynamic loading adaptive control method based on accurate modeling of the hydraulic loading device. The control method can quickly track the given output force trajectory, achieve high-precision force control feedback and output effect verification, and meet the dynamic loading requirements of the solid motor casing under different working conditions.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for adaptively controlling dynamic loading of a solid motor casing, comprising the following steps: For the combination of multi-cylinder combined hydraulic loading device and shell specimen, the component-level model and mathematical analytical model of the dynamic loading device are established; Uncertain factors including friction, leakage, and pressure fluctuations are introduced into the component-level model to simulate the real loading device, which is then used as a test platform. A joint simulation interface for signal interaction between the test platform and the control algorithm is established. Based on the mathematical analytical model, the reference input signal is generated by using the nonlinear sliding surface and exponential convergence rate of the error integral power function to limit the error convergence time and establish a reference model; A neural network sliding mode controller is used to generate compensation signals for the uncertainties existing in the test platform to eliminate the force output error between the reference model and the test platform.

[0006] As a further improvement of the present invention, the establishment of the component level model and the mathematical analysis model of the dynamic loading device includes: Based on the hydraulic loading device, a component-level model is established, and a two-channel force system is used to load the solid engine shell to apply a specified output force to the shell; then the simulation structure of the loading device under study is clarified, and the component-level model and mathematical analytical model of the dynamic loading device are obtained; The force output elements of the hydraulic loading device include hydraulic pumps, hydraulic cylinders, control valves, loads and pipelines.

[0007] As a further improvement of the present invention, a valve core with an irregularly shaped throttle hole is used to simulate a pressure reducing valve, and the key parameters of the valve core are determined including: flow area, delay characteristics; The flow area of ​​the valve core at different openings Calculate according to the relevant data of the actual pressure reducing valve:

[0008] is the mass flow rate, is the hydraulic oil density corresponding to the pressure after the pressure reducing valve, is the pressure drop of the pressure reducing valve, is the bulk modulus of hydraulic oil, is the orifice flow coefficient, which is determined by the orifice shape of the pressure reducing valve; The delay characteristic of the valve core is to approximate the valve core action process of the pressure reducing valve as a second-order system response. After a given input, the valve core reaches the specified position after a short oscillation process. The transfer function is:

[0009] in, are the damping ratio and response frequency of the pressure reducing valve, is the gain coefficient; in the experiment, a step signal is applied to the pressure reducing valve, and the step response of the pressure reducing valve is observed. and Calculated by the following formula:

[0010]

[0011] and They are the adjustment time and overshoot of the valve to the step signal respectively.

[0012] As a further improvement of the present invention, the nonlinear sliding surface is determined by the following formula:

[0013] in is the error between the output force of the reference model and the expected value, All are positive hyperparameters; The convergence rate is determined by the following formula:

[0014] In the formula, is the exponential approach rate, is the linear approach rate, is a symbolic function.

[0015] As a further improvement of the present invention, before establishing the reference model, the method includes: The displacement, speed and output force of the hydraulic cylinder piston rod of the loading device are selected as the three state variables: , the simulation signal flow process is used to analytically model the loading device, and the state space expression is:

[0016] in is the damping coefficient, stiffness and mass after load equivalence; The calculation formula is: When u>0

[0017] When u<0

[0018] The calculation formula is:

[0019] in, is the flow area of ​​the pressure reducing valve, is the valve core displacement, is the gain coefficient between the input signal and the valve core displacement, is the flow area of ​​the reversing valve oil outlet, is the average density of hydraulic oil before and after the valve, is the cross-sectional area of ​​the high-pressure chamber and the low-pressure chamber, is the elastic modulus of hydraulic oil, and is the orifice flow coefficient corresponding to the pressure reducing valve and the reversing valve, is the pump pressure, It is divided into the pressure of the chamber directly connected to the pump source, is the pressure of the other chamber, For the corresponding volume.

[0020] As a further improvement of the present invention, after the state space, the following control algorithm is used to complete the design of the reference model: The input signal u needs to make the output converge within a finite time. Based on the nonlinear sliding surface of the PI controller, the power function of the error integral is introduced on the basis of the sliding surface design, and its expression is:

[0021] in, is the error between the reference model output force and the expected value, ; After entering the sliding mode, if the system state is far away from zero, the convergence time is mainly determined by the attractor If the system state is close to zero, the convergence time is determined by the sliding mode control law: The exponential reaching law is used to ensure that s converges to 0 in a finite time, and the system enters the sliding mode:

[0022] In the formula, is the exponential approach rate, is the linear approach rate, is a symbolic function.

[0023] System status point reached After that, from this current state Converge to The time is:

[0024] By reasonably selecting the values ​​of each parameter and converging within the required time, the expression of the reference control signal u for the pressure reducing valve is obtained as follows:

[0025] in is the expected output force.

[0026] As a further improvement of the present invention, a neural network sliding mode controller is used to generate a compensation signal for the uncertain factors existing in the test platform to eliminate the force output error between the reference model and the test platform, including: The errors between the high-pressure chamber and the low-pressure chamber of the reference model and the test platform are used as the input of the neural network, and the radial basis function is used as the activation function of the hidden layer neurons. The adaptive rate of the neural network parameters is determined by the second Lyapunov method, and the uncertainty of the output force error derivative is obtained. The uncertainty is used as the output of the neural network. The output of the neural network is substituted into the exponential reaching law of the sliding mode control to obtain the compensation signal calculation formula, so as to achieve the purpose of eliminating the tracking error between the test platform and the reference model.

[0027] As a further improvement of the present invention, the neural network sliding mode controller is used to generate a compensation signal for the uncertain factors existing in the test platform to eliminate the force output error between the reference model and the test platform, specifically including: Based on sliding mode control and combined with neural network adaptive control theory, a compensation signal is calculated to eliminate the error between the test platform and the reference model, so that the test platform is loaded according to the set trajectory of the reference model; including: S1, clarify the system uncertainty; through the analytical modeling of the hydraulic loading device, the loading device state space with the hydraulic cylinder output force as the observation index is obtained; in the reference model, the output force change rate is calculated by the following formula:

[0028] Incorporating the influence of uncertainty, the output force change rate of the test platform is rewritten as

[0029] To compensate the signal, define the tracking error:

[0030] Derivative of the tracking error:

[0031] The uncertainty of the error F is written as:

[0032] The input of the neural network is the pressure error between the high-pressure chamber and the low-pressure chamber between the test platform and the reference model, and the input is recorded as ; Theoretical optimal weight Adaptively update weights Instead, we get the neural network's estimate of the error F:

[0033] The error between the estimated value and the true value:

[0034] is the activation function, The error between the estimated weights and the ideal weights for a neural network ， , is a high-order uncertainty with an upper bound, that is ; S2, calculate the relevant parameters of the neural network; use the RBF neural network to approximate the uncertainty in S1, and the activation function of each hidden layer neuron is:

[0035] In the formula c j and b j For the j The central node and radiation width of the hidden layer neurons are selected as fixed values; To obtain the adaptive rate of the neural network related parameters, first construct the integral sliding surface

[0036] in is the error integral gain.

[0037] The Lyapunov function is designed as:

[0038] For The estimated value of is, according to Lyapunov's second method, the sufficient condition for the stability of the system is that the derivative of V is negative definite, and the calculation formula is:

[0039] Since in the above formula Uncertain, to eliminate its influence, set its coefficient equal to 0, and we get , the adaptive rate of the neural network related parameters at this time is:

[0040]

[0041]

[0042] S3, generate the final sliding mode control signal; obtain the neural network approximation value of the uncertainty term F , take the exponential approach law to increase the adjustment rate of the system when the error is large:

[0043] in is the exponential approach rate, is a sign function, designed as a hyperbolic tangent function to reduce the jitter of the input signal:

[0044] Is a positive constant representing the smoothing width.

[0045] Obtain the compensation control signal The expression is:

[0046] The compensation signal With reference signal Superimposed as the input signal of the pressure reducing valve of the test platform.

[0047] As a further improvement of the present invention, the central node and the radiation width are determined using a K-means clustering algorithm, specifically including: First, initialize each hidden layer neuron; Each input vector is then classified into the class that is closest to it:

[0048] Then update the center point:

[0049] is the learning rate, and n is the number of iterations.

[0050] The radiation width calculation formula is:

[0051] is the maximum distance between the center points, is the number of neurons.

[0052] In a second aspect, the present invention provides a solid motor casing dynamic loading adaptive control system, comprising: Model building module, used to build component-level model and mathematical analysis model of dynamic loading device for multi-cylinder combined hydraulic loading device and shell specimen combination; The simulation module is used to introduce uncertain factors such as friction, leakage, and pressure fluctuations into the component-level model to simulate the real loading device, and then use it as a test platform, and establish a joint simulation interface for signal interaction between the test platform and the control algorithm; A reference model building module is used to generate a reference input signal based on a mathematical analytical model by using a nonlinear sliding surface and an exponential approach rate that introduce an error integral power function to limit an error convergence time and to build a reference model; The error elimination module is used to generate compensation signals for uncertain factors existing in the test platform using a neural network sliding mode controller to eliminate the force output error between the reference model and the test platform.

[0053] The beneficial effects of the present invention are embodied in: The present invention designs a method for adaptive control of dynamic loading of a solid engine casing, which mainly includes comprehensive modeling of a hydraulic dynamic loading device and the design of an adaptive control algorithm. The comprehensive modeling method combines component-level modeling and analytical modeling, and introduces the error factors in the actual loading process into the component-level modeling, which serves as a test platform for the performance of the adaptive control algorithm, thereby improving the accuracy and predictability of the system model; the adaptive control algorithm adopts a model reference adaptive structure, introduces a nonlinear sliding surface on the basis of analytical modeling, establishes a reference model, generates a reference input signal, and then uses a neural network sliding mode controller to compensate for the error between the reference model and the test platform caused by uncertain factors, thereby improving the precise control of the output force by the loading device. The invention can quickly track the predetermined output force trajectory, meet the dynamic loading requirements of the engine casing under complex working conditions, effectively reduce the test cost and improve the overall control performance. Obtain more detailed parameters: The reference model can obtain the performance characteristics and response behaviors of each component in the system, provide a scientific basis for the design of the control algorithm, and improve the intuitiveness and interpretability of the control strategy. It has the following advantages: Strong flexibility: The reference model independently generates control signals and output forces, is not affected by the test platform, and can be combined with other control algorithms to enhance adaptability and scalability.

[0054] Reduce experimental costs: Parameter tuning can be performed in the reference model to generate simulation results and select the best parameter combination, which can significantly reduce the consumption of specimens and equipment and improve optimization efficiency and accuracy.

[0055] Loading synchronization requirements: In a dual-cylinder system, the expected output signal generated by the reference model can control the two hydraulic cylinders at the same time to achieve master-slave control, effectively meet the synchronization requirements during the loading process, and ensure coordinated operation under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the embodiments of the present invention or the drawings of related technical solutions in the prior art are introduced below. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0057] Figure 1 It is a schematic diagram of the overall design structure of model reference adaptive control; Figure 2 It is a structural schematic diagram of a real loading device device referred to in the embodiment of the present invention; Figure 3 This is a schematic diagram of the pressure reducing valve modeling structure in AMEsim; Figure 4 This is a schematic diagram of the AMEsim / Simulink joint simulation structure; Figure 5 It is a schematic diagram of the simulation signal flow process; Figure 6 is a reference model schematic; Figure 7 It is a schematic diagram of the adopted RBF neural network structure and related parameters; Figure 8 It is a structural diagram of the control algorithm implementation method. DETAILED DESCRIPTION

[0058] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be construed as limitations of the present invention. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0059] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0060] The first part of the present invention provides a comprehensive modeling method for a hydraulic dynamic loading device, which combines component-level modeling and analytical modeling techniques to improve the accuracy and feasibility of system modeling, and includes the following steps: For the combination of multi-cylinder combined hydraulic loading device and shell specimen, the component-level model and mathematical analytical model of the dynamic loading device are established; Uncertain factors including friction, leakage, and pressure fluctuations are introduced into the component-level model to simulate the real loading device, which is then used as a test platform for the performance of the adaptive control algorithm. A joint simulation interface for signal interaction between the test platform and the control algorithm is established. Based on the mathematical analytical model, the reference input signal is generated by using the nonlinear sliding surface and exponential convergence rate of the error integral power function to limit the error convergence time and establish a reference model; A neural network sliding mode controller is used to generate compensation signals for the uncertainties existing in the test platform to eliminate the force output error between the reference model and the test platform.

[0061] The principle of the comprehensive modeling method for a hydraulic dynamic loading device proposed in the first part of the present invention is as follows: Combination of component-level modeling and mathematical analytical model: Component-level model: Use professional hydraulic system simulation software (such as AMEsim) and predefined hydraulic component design library (HCD library) to build a model of the actual loading device. The model is connected according to the topological structure of the actual system, including core components such as hydraulic pumps, hydraulic cylinders, control valves, loads and pipelines, to truly reflect the physical layout and workflow of the actual equipment.

[0062] Mathematical analytical model: Based on the physical characteristics and working principle of the system, mathematical equations are established to describe the dynamic behavior of the system. This model can be used to generate reference input signals to limit the error convergence time and serve as the basis for control algorithm design and verification.

[0063] Introducing uncertain factors: In the component-level model, uncertain factors such as friction, leakage, and pressure fluctuation are introduced to simulate the complexity and nonlinear characteristics of the real loading device. This helps to more accurately evaluate the performance of the system and the effectiveness of the control algorithm.

[0064] Signal interaction and joint simulation: Establish a signal interaction interface between the test platform and the control algorithm to achieve real-time data exchange between the test platform and the control system. This helps to verify the performance of the control algorithm in the simulation environment and make necessary adjustments and optimizations.

[0065] Nonlinear sliding mode control and neural network compensation: The nonlinear sliding mode surface and exponential convergence rate are used to generate reference input signals to limit the error convergence time. At the same time, the neural network sliding mode controller is used to generate compensation signals for the uncertain factors existing in the test platform to eliminate the force output error between the reference model and the test platform. This method improves the robustness and control accuracy of the system.

[0066] As an example, using the component-level modeling model with introduced errors as a test platform to simulate the real hydraulic device, the control algorithm can be developed and verified in the simulation environment, saving the test cost. Based on the mathematical analytical model, the parameters that cannot be directly measured by the test are fully utilized, and the nonlinear sliding surface and exponential approach rate of the error integral power function are used to establish a reference model, generate a reference input signal, and limit the error convergence time; the neural network sliding mode controller is used to generate compensation signals for the uncertain factors existing in the test platform to eliminate the force output error between the reference model and the test platform, so that the test platform can accurately output the loading force according to the predetermined trajectory.

[0067] In component-level modeling, the professional hydraulic system simulation software AMEsim is used to build a real hydraulic loading device model through the predefined hydraulic component design library (HCD) library. During the modeling process, the connection is made according to the topological structure of the real loading device to ensure that the simulation model can reflect the physical layout and workflow of the real equipment, including core components such as hydraulic pumps, hydraulic cylinders, control valves, loads and pipelines. This component-level modeling method is intuitive and highly interactive. It can adjust and optimize system parameters in real time, providing a reliable simulation test environment for the control algorithm. Through component-level modeling, multiple tests and optimizations can be performed in a virtual environment, which significantly reduces the cost and time of actual tests and improves development efficiency.

[0068] In analytical modeling, detailed mathematical models are established for key components in hydraulic dynamic loading devices, such as hydraulic cylinders and servo valves. This includes the use of nonlinear dynamic equations, basic laws of fluid mechanics, and dynamic equations that take into account compressible fluid characteristics. After obtaining the nonlinear dynamic equations of these key components, a state space model of the force closed-loop control system is constructed. The state space model describes the dynamic behavior of the system through state variables, and can accurately reflect the topological structure of the system and the coupling and interaction between its core components, making the system more controllable and predictable at the mathematical level, and is the basis for subsequent precise control strategy design and stability analysis.

[0069] The second part of the present invention provides a model reference adaptive control algorithm, which specifically includes: a reference model control algorithm and an adaptive feedback control algorithm. The overall structural diagram is shown in FIG. Figure 1 Its implementation includes the following steps: S1, based on the analytical modeling in the first part, according to the given expected output trajectory, the reference input signal is generated using the nonlinear sliding surface and exponential approach rate that introduces the error integral power function to limit the error convergence time and complete the design of the reference model.

[0070] S2. In this application, all steps are completed in a simulation environment. In order to distinguish it from the reference model, the component-level modeling model in the first part is called a "test platform". Its function is to simulate the real loading environment and provide support for the design and verification of the control algorithm.

[0071] Furthermore, the role of the test platform of the present application is to simulate real hydraulic actuators, and the role of mathematical analytical modeling is to design control algorithms. Whether the control algorithm is effective is based on the output of the component-level modeling model.

[0072] Since there are many uncertain factors in the real loading process, such as the influence of the volume effect of the hydraulic cylinder, the influence of the dead zone characteristics of the valve, and the influence of the time-varying nature of various soft parameters, when the same control signal is applied to the valve, there will be a certain error in the force output between the test platform and the reference model. Therefore, a neural network sliding mode controller is further designed in MATLAB / Simulink to generate a compensation signal and superimpose it on the reference input signal in S1 to act on the test platform. The error between the high-pressure chamber and the low-pressure chamber of the reference model and the test platform is used as the input of the neural network, and the radial basis function is used as the activation function of the hidden layer neuron. The adaptive rate of the neural network parameters is determined by the second Lyapunov method, and the uncertainty of the output force error derivative is obtained. This uncertainty is used as the output of the neural network. The output of the neural network is substituted into the exponential reaching law of the sliding mode control to obtain the calculation formula of the compensation signal, so as to eliminate the tracking error between the test platform and the reference model.

[0073] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0074] Example 1 This embodiment refers to Figure 2 The component-level model of the real hydraulic loading device shown is established. Its application background is to use a two-channel force system to load the solid engine casing, with the purpose of quickly and accurately applying a specified output force to the casing.

[0075] The purpose of this embodiment is to clarify the simulation structure of the loading device under study and provide a basis for the subsequent description of the control algorithm embodiment. Therefore, the generalized modeling process of other components is not described in detail. The pressure reducing valve, as the core component for controlling the force output of the loading device, has a key impact on the simulation accuracy. The pressure reducing valve design in the AMEsim hydraulic component design library is different from the actual pressure reducing valve. In order to ensure that the characteristics of this key component are as consistent as possible with the actual object, this embodiment adopts the following method: 1) Determination of basic valve parameters The valve core with irregular throttle hole in the AMEsim hydraulic component design library is used to simulate the pressure reducing valve. Its structure is as follows: Figure 3 As shown in the figure, the According to the relevant data of the actual pressure reducing valve, it is determined by the following formula:

[0076] is the mass flow rate, is the hydraulic oil density corresponding to the pressure after the pressure reducing valve, is the pressure drop of the pressure reducing valve, is the bulk modulus of hydraulic oil, is the orifice flow coefficient, which is determined by the orifice shape of the pressure reducing valve.

[0077] Its piston diameter With piston rod diameter The following formula should be satisfied:

[0078] in is the maximum travel of the valve core, It is the maximum volume of the pressure reducing valve chamber.

[0079] 2) Valve delay characteristics The response delay of moving parts caused by mechanical inertia, the dynamic sluggishness caused by hydraulic oil flow and compression effect, the buffer introduced by valve damping and spring effect, and the electrical signal delay caused by the inductance and resistance of the electromagnetic coil. These factors work together to cause a time delay between the pressure reducing valve receiving the control signal and the actual response. The valve core action process of the pressure reducing valve can be approximately regarded as a second-order system. After a given input, the valve core reaches the specified position after a short oscillation process. Its transfer function can be written as:

[0080] in are the damping ratio and response frequency of the pressure reducing valve, is the gain factor.

[0081] The valve core with irregular shaped throttle hole mentioned in step 1) is actuated immediately without delay according to the input signal, which is different from the action process of the actual valve, so it needs to be corrected.

[0082] right The impulse response function of the pressure reducing valve is obtained by performing Laplace inverse transform:

[0083] In order to determine the damping ratio and response frequency of the pressure reducing valve, a step signal can be applied to the pressure reducing valve in the experiment to observe the step response of the pressure reducing valve. and Calculated by the following formula:

[0084]

[0085] and They are the adjustment time and overshoot of the valve to the step signal respectively.

[0086] After obtaining the damping ratio and response frequency of the valve, the second-order transfer function of equation (3) is substituted to process the input signal to simulate the delay characteristics of the valve.

[0087] Since AMEsim has advantages in modeling hydraulic loading devices but is slightly insufficient in generating control signals, and MATLAB has powerful data processing capabilities, an AMEsim / Simulink joint simulation platform was built to couple the test platform with the controller designed in MATLAB. The overall structure is as follows: Figure 4 As shown. Pressure sensors are set at the oil inlet and outlet of the hydraulic cylinder, and force sensors are set at the connection between the hydraulic cylinder piston rod and the load. The output of the sensor is imported into the Simucosim interface, and the controller in MATLAB generates a control signal to control the action of the hydraulic components in the AMEsim test platform, completing the design and verification of the control algorithm. In this way, the analysis and design efficiency of the hydraulic loading device can be significantly improved.

[0088] Example 2 The purpose of this implementation is to perform mathematical analytical modeling on the loading device and establish a reference model based on this, which serves to provide a reference input signal for the test platform of Example 1 through the interface of the joint simulation platform.

[0089] Since the control goal of the loading device is to achieve closed-loop force control and focus on the change of system force, the displacement, speed and output force of the hydraulic cylinder piston rod of the loading device are selected as the three state variables, namely ,according to Figure 5 The simulation signal flow process shown in the figure is used to perform mathematical analytical modeling of the loading device in MATLAB / Simulink, and the expression is:

[0090] in are the damping coefficient, stiffness and mass after load equivalence. The calculation formula is: When u>0

[0091] When u<0

[0092] The calculation formula is:

[0093] in, is the flow area of ​​the pressure reducing valve, is the valve core displacement, is the gain coefficient between the input signal and the valve core displacement, is the flow area of ​​the reversing valve oil outlet, is the average density of hydraulic oil before and after the valve, is the cross-sectional area of ​​the high-pressure chamber and the low-pressure chamber, is the elastic modulus of hydraulic oil, and is the orifice flow coefficient corresponding to the pressure reducing valve and the reversing valve, is the pump pressure, It is divided into the pressure of the chamber directly connected to the pump source, is the pressure of the other chamber, For the corresponding volume.

[0094] After establishing the mathematical analytical model, the following control algorithm was designed on this basis to complete the design of the reference model: Since the electro-hydraulic servo system has inherent nonlinear characteristics, parameter uncertainty, and high requirements for response speed and accuracy, conventional PID control often cannot meet the dynamic performance requirements. However, sliding mode control has the advantages of robustness and fast response, and can effectively deal with parameter changes and external interference, providing strong anti-interference ability, which is specifically manifested in the ability to compensate for friction and dead zone in the electro-hydraulic servo system and time-varying soft parameters. Therefore, sliding mode control has a high degree of adaptability for such systems.

[0095] For the dynamic loading test of the solid motor casing, it is required to generate a large output force within a specified time. Therefore, the input signal u needs to make the output converge within a limited time. The traditional sliding surface does not limit the error convergence time and cannot meet the requirements. In addition, since the expected output force is a constant value rather than a rapid change, the demand for differential control is not high. Therefore, this application designs a nonlinear sliding surface based on the PI controller concept and introduces the power function of the error integral on the basis of the traditional sliding surface design. The expression is:

[0096] in is the error between the reference model output force and the expected value, calculated by formula (7), .

[0097] This sliding mode control inherits the advantages of traditional sliding mode control. After the system enters the sliding mode, if the system state is far away from zero, the convergence time is mainly determined by the attractor. If the system state is close to zero, the convergence time is determined by the traditional sliding mode control law.

[0098] First, the exponential reaching law is used to ensure that s converges to 0 in a finite time, and the system enters the sliding mode:

[0099] In the formula is the exponential approach rate, is the linear approach rate, is a symbolic function.

[0100] System status point reached After that, from this current state Converge to The time is:

[0101] By reasonably selecting the values ​​of each parameter, the system can converge within the required time, and the expression of the reference control signal u for the pressure reducing valve is obtained as follows:

[0102] in is the expected output force.

[0103] The design of the reference model is now complete, and its overall schematic diagram is as follows: Figure 6 shown.

[0104] Example 3 There are many uncertain factors in the actual loading process, such as the influence of the hydraulic cylinder volume effect, the influence of the proportional valve dead zone characteristics, and the influence of the time-varying nature of various soft parameters. Therefore, if the same control signal shown in formula (14) is applied to the valve, there will be a certain error in the force output between the test platform and the reference model that introduces uncertain factors.

[0105] In order to eliminate the force output error between the reference model and the test platform, this embodiment further describes the construction process of a neural network adaptive control algorithm. The algorithm is based on the improvement of sliding mode control and combined with the neural network adaptive control theory. The purpose is to calculate a compensation signal and a reference input signal to superimpose the error between the test platform and the reference model, so that the test platform is loaded according to the set trajectory of the reference model. The specific implementation steps include: S1, clarify the system uncertainty. Based on Example 2, by analytically modeling the hydraulic loading device, the loading device state space with the hydraulic cylinder output force as the observation index is obtained. In the reference model, the loading process is idealized, and its output force change rate is calculated by the following formula:

[0106] Although there is uncertainty in the output force change rate during the actual loading process and it is difficult to obtain an accurate mathematical expression, the uncertainty does not affect the overall structure of the control equation. Therefore, referring to equation (15), the output force change rate of the test platform is rewritten by adding the influence of uncertainty:

[0107] In the formula To compensate the signal, define the tracking error:

[0108] Derivative of the tracking error:

[0109] The uncertainty F of the error can be written as:

[0110] Although Simulink contains a built-in differential module, which can theoretically directly calculate the derivative value of the tracking error, this design still regards the derivative of the tracking error as an uncertain quantity and then uses a neural network approximation. This is because the built-in derivative module is easily affected by noise in the joint simulation. Numerical differentiation will amplify high-frequency noise during the discretization process, especially when the signal contains tiny high-frequency fluctuations. The derivative module may introduce unnecessary oscillations, resulting in unstable simulation results. This is very disadvantageous in feedback control.

[0111] Since the pressure of the high-pressure chamber and the low-pressure chamber directly affects the change of the output force, the input of the neural network is the pressure error of the high-pressure chamber and the low-pressure chamber between the test platform and the reference model, and the input is recorded as The theoretical optimal weight is Unknown, so the weights are updated adaptively Instead, we get the neural network's estimate of the error F:

[0112] The error between the estimated value and the true value:

[0113] is the activation function, The error between the estimated weights and the ideal weights for a neural network ， , is a high-order uncertainty with an upper bound, that is .

[0114] S2, calculate the neural network related parameters. This control algorithm uses RBF neural network to approximate the uncertainty in S1, and its structure is as follows: Figure 7 As shown, the activation function of each hidden layer neuron is:

[0115] In the formula c j and b j For the j The central node and radiation width of each hidden layer neuron can generally be selected as a certain value based on engineering experience.

[0116] As an optional solution, in order to enhance the adaptability of the neural network, this embodiment uses a K-means clustering algorithm to determine its value, and the steps are: First, each hidden layer neuron is initialized, and then each input vector is classified, and the class it belongs to is the center point closest to it:

[0117] Then update the center point:

[0118] is the learning rate, and n is the number of iterations.

[0119] The radiation width calculation formula is:

[0120] is the maximum distance between the center points, is the number of neurons.

[0121] In order to obtain the adaptive rate of neural network related parameters, we first construct the integral sliding surface

[0122] The Lyapunov function is designed as:

[0123] From Lyapunov's second method, we know that a sufficient condition for the stability of the system is that the derivative of V is negative definite, and the calculation formula is:

[0124] Since in the above formula To eliminate its influence, let its coefficient equal to 0, and we can get , the adaptive rate of the neural network related parameters at this time is:

[0125]

[0126]

[0127] S3, generate the final sliding mode control signal. The above obtains the neural network approximation value of the uncertainty term F in formula (19): ,The last step of controller design is to obtain the compensation signal. ,Since the loading device has high requirements on the dynamic ,performance, the exponential approach law is still adopted to increase the ,adjustment rate of the system when the error is large.

[0128]

[0129] in is the exponential approach rate, is a sign function, designed as a hyperbolic tangent function to reduce the jitter of the input signal:

[0130] In the formula Is a positive constant representing the smoothing width.

[0131] Substituting equation (26) into equation (32) yields the compensation control signal The expression is:

[0132] The compensation signal The reference input signal in Example 2 The superposition is used as the final input signal of the pressure reducing valve of the test platform to achieve the purpose of loading according to the expected trajectory. So far, the design of the model reference adaptive control algorithm has been completed. The overall implementation diagram is shown in Figure 8 shown.

[0133] To this end, the present invention also provides a solid motor casing dynamic loading adaptive control system, comprising: Model building module, used to build component-level model and mathematical analysis model of dynamic loading device for multi-cylinder combined hydraulic loading device and shell specimen combination; The simulation module is used to introduce uncertain factors such as friction, leakage, and pressure fluctuations into the component-level model to simulate the real loading device, and then use it as a test platform, and establish a joint simulation interface for signal interaction between the test platform and the control algorithm; A reference model building module is used to generate a reference input signal based on a mathematical analytical model by using a nonlinear sliding surface and an exponential approach rate that introduce an error integral power function to limit an error convergence time and to build a reference model; The error elimination module is used to generate compensation signals for uncertain factors existing in the test platform using a neural network sliding mode controller to eliminate the force output error between the reference model and the test platform.

[0134] The third object of the embodiment of the present invention is to provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for adaptively controlling the dynamic loading of a solid motor casing when executing the computer program. The electronic device also includes a communication interface and a bus.

[0135] A fourth objective of an embodiment of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for adaptively controlling the dynamic loading of the solid motor casing is implemented.

[0136] A fifth objective of an embodiment of the present invention is to provide a computer program product, wherein the computer program product comprises computer instructions, wherein the computer instructions instruct a computer to execute the method for adaptively controlling dynamic loading of a solid motor casing.

[0137] In summary, the present invention has the following advantages: Combining component-level modeling and analytical modeling techniques can more comprehensively describe the dynamic behavior of the system and improve the accuracy and feasibility of modeling. By performing multiple tests and optimizations in a virtual environment, the cost and time of actual tests are significantly reduced, and development efficiency is improved. The introduction of nonlinear sliding mode control and neural network compensation technology can effectively deal with uncertainties in the system and improve the robustness and control accuracy of the system. The component-level modeling method is intuitive and highly interactive, and can adjust and optimize system parameters in real time, providing a reliable simulation test environment for the control algorithm.

[0138] In summary, the comprehensive modeling method of the present invention significantly improves the modeling accuracy, feasibility, robustness and control precision of the hydraulic dynamic loading device by combining the two technologies of component-level modeling and analytical modeling, and introducing nonlinear sliding mode control and neural network compensation technology, providing strong support for the design and verification of complex systems.

[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0141] The present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, readable storage media, optical storage, etc.) containing computer-usable program codes.

[0142] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0143] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for adaptive control of dynamic loading of a solid motor casing, characterized in that: The following steps are involved: For the combination of multi-cylinder combined hydraulic loading device and shell specimen, the component-level model and mathematical analytical model of the dynamic loading device are established; Uncertain factors including friction, leakage, and pressure fluctuations are introduced into the component-level model to simulate the real loading device, which is then used as a test platform, and a joint simulation interface for signal interaction between the test platform and the control algorithm is established; Based on the mathematical analytical model, the reference input signal is generated by using the nonlinear sliding surface and exponential convergence rate of the error integral power function to limit the error convergence time and establish a reference model; A neural network sliding mode controller is used to generate compensation signals for the uncertainties existing in the test platform to eliminate the force output error between the reference model and the test platform.

2. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 1, characterized in that: The establishment of the component level model and mathematical analysis model of the dynamic loading device includes: Based on the hydraulic loading device, a component-level model is established, and a two-channel force system is used to load the solid engine shell to apply a specified output force to the shell; then the simulation structure of the loading device under study is clarified, and the component-level model and mathematical analytical model of the dynamic loading device are obtained; The force output elements of the hydraulic loading device include hydraulic pumps, hydraulic cylinders, control valves, loads and pipelines.

3. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 2, characterized in that: A valve core with an irregular throttle hole is used to simulate a pressure reducing valve. The key parameters of the valve core include: flow area, delay characteristics; The flow area of ​​the valve core at different openings Calculate according to the relevant data of the actual pressure reducing valve: is the mass flow rate, is the hydraulic oil density corresponding to the pressure after the pressure reducing valve, is the pressure drop of the pressure reducing valve, is the bulk modulus of hydraulic oil, is the orifice flow coefficient, which is determined by the orifice shape of the pressure reducing valve; The delay characteristic of the valve core is to approximate the valve core action process of the pressure reducing valve as a second-order system response. After a given input, the valve core reaches the specified position after a short oscillation process. The transfer function is: in, are the damping ratio and response frequency of the pressure reducing valve, is the gain coefficient; in the experiment, a step signal is applied to the pressure reducing valve, and the step response of the pressure reducing valve is observed. and Calculated by the following formula: and They are the adjustment time and overshoot of the valve to the step signal respectively.

4. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 1, characterized in that: The nonlinear sliding surface is determined by the following formula: in is the error between the reference model output force and the expected value, All are positive hyperparameters; The convergence rate is determined by the following formula: In the formula, is the exponential approach rate, is the linear approach rate, is a symbolic function.

5. The method for adaptively controlling dynamic loading of a solid motor casing according to claim 1, characterized in that: Before establishing the reference model, the following steps are included: The displacement, speed and output force of the hydraulic cylinder piston rod of the loading device are selected as the three state variables: , the simulation signal flow process is used to analytically model the loading device, and the state space expression is: in is the damping coefficient, stiffness and mass after load equivalence; The calculation formula is: When u>0 When u<0 The calculation formula is: in, is the flow area of ​​the pressure reducing valve, is the valve core displacement, is the gain coefficient between the input signal and the valve core displacement, is the flow area of ​​the reversing valve oil outlet, is the average density of hydraulic oil before and after the valve, is the cross-sectional area of ​​the high-pressure chamber and the low-pressure chamber, is the elastic modulus of hydraulic oil, and is the orifice flow coefficient corresponding to the pressure reducing valve and the reversing valve, is the pump pressure, It is divided into the pressure of the chamber directly connected to the pump source, is the pressure in the other chamber, For the corresponding volume.

6. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 5, characterized in that: After the state space is described, the following control algorithm is used to complete the design of the reference model: The input signal u needs to make the output converge within a finite time. Based on the nonlinear sliding surface of the PI controller, the power function of the error integral is introduced on the basis of the sliding surface design, and its expression is: in, is the error between the reference model output force and the expected value, ; After entering the sliding mode, if the system state is far away from zero, the convergence time is mainly determined by the attractor If the system state is close to zero, the convergence time is determined by the sliding mode control law: The exponential reaching law is used to ensure that s converges to 0 in a finite time, and the system enters the sliding mode: In the formula, is the exponential approach rate, is the linear approach rate, is a symbolic function; System status point reached After that, from this current state Converge to The time is: By reasonably selecting the values ​​of each parameter and converging within the required time, the expression of the reference control signal u for the pressure reducing valve is obtained as follows: in is the expected output force.

7. A method for adaptively controlling dynamic loading of a solid rocket motor casing according to claim 1, characterized in that: A neural network sliding mode controller is used to generate compensation signals for the uncertainties of the test platform to eliminate the force output error between the reference model and the test platform, including: The errors between the high-pressure chamber and the low-pressure chamber of the reference model and the test platform are used as the input of the neural network, and the radial basis function is used as the activation function of the hidden layer neurons. The adaptive rate of the neural network parameters is determined by Lyapunov's second method, and the uncertainty of the output force error derivative is obtained. The uncertainty is used as the output of the neural network. The output of the neural network is substituted into the exponential reaching law of the sliding mode control to obtain the compensation signal calculation formula, so as to achieve the purpose of eliminating the tracking error between the test platform and the reference model.

8. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 7, characterized in that: The method of using a neural network sliding mode controller to generate a compensation signal for uncertain factors existing in the test platform to eliminate the force output error between the reference model and the test platform specifically includes: Based on sliding mode control and combined with neural network adaptive control theory, a compensation signal is calculated to eliminate the error between the test platform and the reference model, so that the test platform is loaded according to the set trajectory of the reference model; including: S1, clarify the system uncertainty; through the analytical modeling of the hydraulic loading device, the loading device state space with the hydraulic cylinder output force as the observation index is obtained; in the reference model, the output force change rate is calculated by the following formula: Incorporating the influence of uncertainty, the output force change rate of the test platform is rewritten as To compensate the signal, define the tracking error: Derivative of the tracking error: The uncertainty of the error F is written as: The input of the neural network is the pressure error between the high-pressure chamber and the low-pressure chamber between the test platform and the reference model, and the input is recorded as ; Theoretical optimal weight Adaptively update weights Instead, we get the neural network's estimate of the error F: The error between the estimated value and the true value: is the activation function, The error between the estimated weights and the ideal weights for a neural network ， , is a high-order uncertainty with an upper bound, that is ; S2, calculate the relevant parameters of the neural network; use the RBF neural network to approximate the uncertainty in S1, and the activation function of each hidden layer neuron is: In the formula c j and b j For the j The central node and radiation width of the hidden layer neurons are selected as fixed values; To obtain the adaptive rate of the neural network related parameters, first construct the integral sliding surface in is the error integral gain; The Lyapunov function is designed as: For The estimated value of is, according to Lyapunov's second method, the sufficient condition for the stability of the system is that the derivative of V is negative definite, and the calculation formula is: Since in the above formula Uncertain, to eliminate its influence, set its coefficient equal to 0, and we get , the adaptive rate of the neural network related parameters at this time is: S3, generate the final sliding mode control signal; obtain the neural network approximation value of the uncertainty term F , take the exponential approach law to increase the adjustment rate of the system when the error is large: in is the exponential approach rate, is a sign function, designed as a hyperbolic tangent function to reduce the jitter of the input signal: is a positive constant, representing the smoothing width; Obtain the compensation control signal The expression is: The compensation signal With reference signal Superimposed as the input signal of the pressure reducing valve of the test platform.

9. A method for adaptively controlling dynamic loading of a solid motor casing according to claim 8, characterized in that: The central node and the radiation width are determined by using the K-means clustering algorithm, which specifically includes: First, initialize each hidden layer neuron; Each input vector is then classified into the class that is closest to it: Then update the center point: is the learning rate, n is the number of iterations; The radiation width calculation formula is: is the maximum distance between the center points, is the number of neurons.

10. A solid rocket engine casing dynamic loading adaptive control system, characterized in that: include: Model building module, used to build component-level model and mathematical analysis model of dynamic loading device for multi-cylinder combined hydraulic loading device and shell specimen combination; The simulation module is used to introduce uncertain factors such as friction, leakage, and pressure fluctuations into the component-level model to simulate the real loading device, and then use it as a test platform, and establish a joint simulation interface for signal interaction between the test platform and the control algorithm; A reference model building module is used to generate a reference input signal based on a mathematical analytical model by using a nonlinear sliding surface and an exponential approach rate that introduce an error integral power function to limit an error convergence time and to build a reference model; The error elimination module is used to generate compensation signals for uncertain factors existing in the test platform using a neural network sliding mode controller to eliminate the force output error between the reference model and the test platform.

Images