Posture adaptive control method and system based on dynamic parameter optimization

Abstract

The present application belongs to the technical field of carrier aircraft attitude control, and provides an attitude adaptive control method and system based on dynamic parameter optimization, comprising the following steps: obtaining the maximum limit value of the drift amount and the drift amount early warning value of the corresponding parameters of the carrier aircraft in different flight state stages according to the historical flight database of the carrier aircraft; comparing the real-time parameters with the corresponding drift amount early warning value according to the current flight state stage of the carrier aircraft, extracting the parameters exceeding the drift amount early warning value as the parameters to be optimized and triggering the calculation and analysis process, and analyzing the change trend of each parameter to be optimized and calculating the predicted time period corresponding to the maximum limit value of the drift amount. The decision imbalance problem caused by using only a single indicator is optimized, the shortest prediction time period is set as the limited time, and all parameter optimization is required to be completed before the over-limit critical point of the most urgent parameter, thereby reducing the chain failure risk caused by the short board effect and increasing the safety.

Description

Technical Field

[0001] This invention belongs to the field of attitude control technology for launch vehicles, specifically an attitude adaptive control method and system based on dynamic parameter optimization. Background Technology

[0002] As a core technology unit of aerospace launch vehicle systems, the attitude control of launch vehicles directly determines key mission indicators such as orbit insertion accuracy and reentry attitude stability. It is an important technical foundation for ensuring that launch vehicles can accurately deliver payloads and that spacecraft can achieve orbital maneuvers. Throughout the entire flight process of a launch vehicle, the dynamic evolution characteristics of key parameters such as engine thrust vector parameters, aerodynamic servo parameters, and inertial navigation system measurement data affect the real-time calculation of attitude control torque and the precise adjustment of the actuator through coupling effects. For example, during the active phase of flight, small fluctuations in engine thrust, drift of the aerodynamic characteristics of the launch vehicle with changes in Mach number, and time-varying characteristics of the zero-bias stability of inertial devices can all significantly affect the closed-loop stability of the attitude control loop. However, the flight of a launch vehicle goes through multiple stages, including low dynamic, high dynamic and stable flight. The parameter fluctuation characteristics of each stage are significantly different. Furthermore, the risk of parameter drift is aggravated by the complex environment (such as atmospheric disturbance, sensor noise and actuator aging). Traditional attitude control methods usually rely on static threshold setting and fixed parameter adjustment strategies, which are difficult to adapt to the dynamic characteristics of multiple stages and abnormal parameter fluctuations in complex environments in real time. This results in insufficient system robustness and may also lead to the risk of attitude loss of control due to the failure to handle the out-of-limit critical parameters in a timely manner. To address this, the present invention provides an attitude adaptive control method and system based on dynamic parameter optimization. Summary of the Invention

[0003] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved. The technical solution adopted by this invention to solve its technical problem is: an attitude adaptive control method based on dynamic parameter optimization, comprising the following steps: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight states of the launch vehicle. Based on the current flight status of the launch vehicle, the real-time parameters are compared with the corresponding drift warning values. Parameters exceeding the drift warning values ​​are extracted as parameters to be optimized and the calculation and analysis process is triggered. The changing trend of each parameter to be optimized is analyzed and the predicted time period corresponding to the maximum drift limit is calculated. The influence coefficient of each parameter to be optimized on the attitude of the launch vehicle is obtained, and a comprehensive analysis is performed in combination with the prediction time period corresponding to each parameter to be optimized to obtain the adjustment priority of the parameter to be optimized. Based on the adjustment priority of the parameter to be optimized, the parameter to be optimized is optimized and compensated within a limited time. A launch vehicle attitude adaptive control system based on dynamic parameter optimization, comprising: Parameter benchmark definition module: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight state stages of the launch vehicle; Real-time monitoring and prediction module: Based on the current flight status stage of the launch vehicle, the real-time parameters are compared with the corresponding drift warning values. Parameters exceeding the drift warning values ​​are extracted as parameters to be optimized and the calculation and analysis process is triggered. The changing trend of each parameter to be optimized is analyzed and the predicted time period corresponding to the maximum drift limit is calculated. Multi-dimensional decision compensation module: This module acquires the influence coefficient of each parameter to be optimized on the attitude of the launch vehicle, and performs a comprehensive analysis based on the prediction time period corresponding to each parameter to determine the adjustment priority. Then, based on this priority, it optimizes and compensates for the parameters within a specified timeframe. The beneficial effects of this invention are as follows: 1. By using a dual triggering mechanism of warning value and maximum limit approach value, a graded warning of parameter drift is achieved, which can capture parameters that are about to approach the safety critical state in advance, reduce the response lag problem caused by relying on a single critical value. Combining the trend analysis of linear fitting and random forest model, it can not only adapt to the stable change law of linear drift scenario, but also deal with the risk of nonlinear mutation or accelerated drift through historical maximum change rate and dynamic prediction model, and then calculate the prediction time period for the parameter to reach the maximum limit. This optimizes the defects of static evaluation that assumes constant drift speed, and can perceive the dynamic characteristics of parameter drift in complex environment in real time, leaving sufficient adjustment time for subsequent decision making. 2. Based on the dual dimensions of the degree of influence of parameters on attitude and the urgency of time, an adjustment priority evaluation system is constructed to optimize the decision imbalance problem caused by using only a single indicator. By setting the shortest prediction time period as a time limit and requiring all parameter optimization to be completed before the most urgent parameter exceeds the limit, the risk of chain failure caused by the weakest link effect is reduced and safety is increased. 3. Based on priority, the control cycle and compensation amount are dynamically allocated, which reduces resource waste and ineffective adjustments. Compensation is achieved within a unified time window, which reduces the system oscillation that may be caused by one-time adjustments and improves the continuity and efficiency of control actions, ultimately achieving the optimal balance between risk control and resource utilization. Attached Figure Description The invention will now be further described with reference to the accompanying drawings. Figure 1 This is a flowchart of the attitude adaptive control method based on dynamic parameter optimization of the present invention; Figure 2 This is a diagram of the attitude adaptive control system architecture based on dynamic parameter optimization of this invention. Detailed Implementation To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments. Example 1 Please see Figure 1 As shown in the embodiment of the present invention, the attitude adaptive control method based on dynamic parameter optimization includes the following steps: Step S10: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight state stages of the launch vehicle; In step S10, the first thing to define is that, according to dynamic characteristics, the flight state phases of the launch vehicle can be divided into low-dynamic flight phase, high-dynamic flight phase, and stable flight phase. Obtain the historical flight database of launch vehicles. The methods for obtaining the historical flight database of launch vehicles include, but are not limited to: aerospace data service providers, third-party websites, and aerospace institutions. The maximum drift limit and the drift warning value of the corresponding parameters for different state stages of the launch vehicle are determined based on statistical analysis methods. The drift warning value is less than the maximum drift limit. The drift warning value is used to provide warnings to people in the field and to conduct subsequent predictive analysis based on the warning. Optionally, the confidence interval method is used to process the parameter data of each flight state stage in the historical flight database of the launch vehicle, calculate the mean and standard deviation of each parameter, take the sum of the mean and k1 times the standard deviation as the maximum value of the normal fluctuation range of the corresponding parameter, and take the difference between the mean and k2 times the standard deviation as the minimum value of the normal fluctuation range of the corresponding parameter. It needs to be defined that the drift of a parameter is the absolute value of the difference between the parameter value and the corresponding mean, which represents the mean of each parameter calculated based on the parameter data in the historical flight database of the launch vehicle. The maximum drift limit and the drift warning value are obtained based on the normal fluctuation range of the parameters. Therefore, the maximum drift limit is k1 times the standard deviation, and the drift warning value is k2 times the standard deviation. Among them, the value of k1 is usually 3, and the value of k2 is usually 2. The essence of their value is the optimal choice under the combination of statistical confidence and engineering safety requirements. In some extreme scenarios, the value of k1 can be 2.5, which is summarized and set by those skilled in the art based on the characteristics of the scenario. It is important to note that before calculating the mean and standard deviation of each parameter, the parameter data needs to be filtered and outliers removed so that the calculated mean and standard deviation represent the normal fluctuation range of the parameters of each launch vehicle flight state stage, making the maximum drift limit and warning value more reasonable. Specifically, the filtering and removal of outliers includes: removing abnormal parameter values ​​for each flight stage of the launch vehicle; Compare the value of each parameter with the threshold range of the corresponding parameter; If a parameter value is not within the threshold range of the corresponding parameter, it is considered an abnormal parameter value; if a parameter value is within the threshold range of the corresponding parameter, it is considered a normal parameter value; abnormal parameter values ​​are removed before calculating the mean and standard deviation of each parameter. This step involves first dividing the flight state of the launch vehicle into low-dynamic, high-dynamic, and stable flight stages; then acquiring the historical flight database of the launch vehicle; next, filtering and removing outliers from the parameter data in the database to eliminate abnormal parameter values ​​for each flight state stage; and then using the confidence interval method to calculate the mean and standard deviation of each parameter to determine the normal fluctuation range of the parameter. Based on this, the sum of the mean and k1 times the standard deviation is taken as the maximum value of the normal fluctuation range, and the difference between the mean and k2 times the standard deviation is taken as the minimum value. The parameter drift is the absolute value of the difference between the parameter value and the corresponding mean, thus deriving the maximum drift limit (k1 times the standard deviation) and the drift warning value (k2 times the standard deviation). The optimization addresses the issue of inaccurate judgment and untimely warning of parameter drift caused by unreasonable definition of the normal fluctuation range of parameters in complex environments. It makes the set maximum drift limit and warning value more consistent with the actual flight state of the launch vehicle, and improves the adaptability and reliability of the launch vehicle attitude control system to parameter changes in different flight stages. It has the following effects: Through statistical analysis of historical data and screening of abnormal data, the maximum drift limit and warning value can more accurately represent the normal fluctuation range of parameters in each flight state stage, providing scientific and reasonable benchmark data for subsequent real-time parameter monitoring and optimization compensation, and improving the system's accuracy in identifying parameter anomalies and its early warning capability. Step S20: Based on the current flight status stage of the launch vehicle, compare the real-time parameters with the corresponding drift warning values, extract the parameters that exceed the drift warning values ​​as parameters to be optimized, analyze the changing trend of each parameter to be optimized, and calculate the predicted time period corresponding to the maximum drift limit. In step S20, the current flight status stage of the launch vehicle is obtained, and based on the current flight status stage of the launch vehicle, the real-time parameter drift amount is compared with the drift amount warning value; If the real-time parameter drift is less than the drift warning value, a normal signal is generated; if the real-time parameter drift is greater than or equal to the drift warning value, a warning signal is generated. Based on the warning signal, a maximum drift limit is set, which is set by those skilled in the art based on experience; the real-time parameter drift is compared with the maximum drift limit. It should be explained that the maximum drift limit is close to the value. When the real-time drift reaches or exceeds this value, it is considered that the parameter is about to approach the safety critical state. It should be immediately extracted as a "parameter to be optimized" and the predictive analysis process should be started. This allows sufficient time to adjust the parameter before it reaches the maximum limit, reduces response lag caused by directly relying on the maximum limit as a trigger condition, and ensures that the system has enough time to implement optimization compensation to prevent risks caused by parameter exceeding the limit. If the real-time parameter drift is less than the maximum drift limit, then continue with subsequent real-time comparisons. If the real-time parameter drift is greater than or equal to the maximum drift limit, then the parameter corresponding to the real-time parameter drift being greater than or equal to the maximum drift limit is extracted as the parameter to be optimized to trigger the calculation and analysis process. In step S20, the execution process that triggers the calculation and analysis flow is as follows: For each parameter to be optimized, the time point corresponding to the generation of the early warning signal and the time point corresponding to the triggering of the calculation and analysis process are extracted as data windows; Use the parameter drift data from the data window as the dataset for trend analysis; The data in the dataset is linearly fitted, and the fitting formula is: The slope a and intercept b are calculated using the least squares method; The goodness of fit is used to determine whether a dataset exhibits linear variation. The goodness of fit is quantified by the coefficient of determination, reflecting the degree to which the independent variable (time t) explains the dependent variable (drift d(t)). The calculation formula is as follows: ;in, This represents the absolute coefficient, and n represents the number of values ​​in the dataset. Describe the i-th actual drift value in the dataset. This represents the i-th fitted drift value. This represents the average actual drift. Set a minimum acceptable goodness of fit. If the calculated coefficient of determination is greater than or equal to the minimum acceptable goodness of fit, it means that the dataset fits the linear relationship well and conforms to linear change. Then, the slope 'a' value is used as the rate of change of the steady drift of the corresponding parameter. Based on the rate of change of the stable drift, the difference between the real-time drift and the maximum drift limit is calculated, and then the ratio of the difference to the rate of change of the stable drift is calculated to obtain the prediction time period of the corresponding parameter. If the calculated absolute coefficient is less than the minimum acceptable goodness of fit, it indicates that the dataset has a poor linear fit and does not conform to linear change. In this case, the maximum rate of change of the drift in the dataset is taken as the pre-calculated rate of change of the drift. The process of extracting the pre-calculated drift rate of change is as follows: calculate the change of each data point in the dataset compared to the previous data point, and use the ratio of the change to the time interval between the two data points as the change ratio, and extract the value with the largest change ratio as the pre-calculated drift rate of change. Based on the pre-calculated drift rate of change, the difference between the real-time drift and the maximum drift limit is calculated. Then, the ratio of the difference to the pre-calculated drift rate of change is calculated to obtain the corresponding pre-analysis and prediction time period. Based on the pre-analysis and prediction time period, the pre-analysis time point for reaching the maximum drift limit is calculated in combination with the current time point; Furthermore, as the number of drift values ​​increases in real time over subsequent periods, the number of values ​​in the dataset will increase, and the maximum rate of change of drift may change. Therefore, this example also includes a drift rate of change prediction process, the specific process of which is as follows: The drift rate of change for each data point is integrated into a drift rate of change sequence, and then divided into a training set and a validation set in a 7:3 ratio. The drift rate of change prediction model is trained using the training set, and the model can be a random forest model. The training set is used to fit a random forest regression model, and the root mean square error (RMSE) and mean absolute error (MAE) are calculated on the validation set. If the set conditions are not met, the feature dimension is increased or the hyperparameters are adjusted until the model accuracy meets the engineering requirements. The drift rate prediction model, obtained through training, is used to calculate and output a sequence of predicted drift rates from the current time point to the pre-analysis time point; the maximum value in the predicted drift rate sequence is then extracted. The pre-calculated drift rate of change is compared with the maximum value in the predicted drift rate of change sequence. If the pre-calculated drift rate of change is greater than or equal to the maximum value in the predicted drift rate of change sequence, then the pre-calculated drift rate of change is taken as the unstable drift rate of change. The aforementioned calculation and analysis prediction period is the prediction period of the corresponding parameter. If the pre-calculated drift rate is less than the maximum value in the predicted drift rate sequence, then the maximum value in the predicted drift rate sequence is taken as the unstable drift rate. Based on the unstable drift rate, the difference between the real-time drift and the maximum drift limit is calculated. Then, the difference is compared with the unstable drift rate to obtain the prediction time period for the corresponding parameter. Obtain the predicted time period for all parameters to be optimized; This step first compares the drift of real-time parameters with the corresponding drift warning values ​​based on the current flight status of the launch vehicle. If the drift exceeds the warning value, a warning signal is generated. The drift is then compared with the nearest value to the maximum drift limit (an empirical threshold between the warning value and the maximum limit). Parameters exceeding the nearest value are extracted as parameters to be optimized. Subsequently, the time point when the warning signal is generated and the analysis process is triggered is used as a data window. The drift data within the window is linearly fitted, and the goodness of fit is determined by the coefficient of determination. If the goodness of fit is satisfactory, the linear slope is used as the stable drift rate. If it is not satisfactory, the maximum rate of change in the dataset is used as the pre-calculated drift rate. A random forest model is then used to predict the future drift rate. Combining the historical maximum rate of change with the predicted value, the unstable drift rate is finally determined. Based on this, the predicted time period for the parameter to be optimized to reach the maximum drift limit is calculated, providing a time reference for subsequent parameter optimization. The following issues are optimized through a progressive processing approach: early warning, near-value triggering, multi-mode trend analysis, and dynamic prediction. This approach reduces the need for adjustments to be triggered only when parameters approach their maximum limits, addressing the problem of insufficient adjustment time near critical values ​​by intervening early. It also compensates for the shortcomings of linear fitting in scenarios involving parameter abrupt changes or nonlinear drift by using the maximum rate of change and prediction models, improving prediction accuracy under complex conditions. Furthermore, by combining historical data with real-time predictions, the drift rate is updated in real-time, addressing the shortcomings of static assessments that assume a constant drift rate and enhancing the system's dynamic perception of parameter drift risks. It has the following effects: By using both warning and near-warning values, it achieves hierarchical identification of parameter anomalies, reduces response lag caused by direct reliance on maximum limits, and allows sufficient time for parameter adjustment; by combining linear fitting and nonlinear change in comprehensive judgment, it adapts to parameter drift characteristics under different scenarios and improves the accuracy of trend analysis; by introducing a random forest model to predict the rate of change of drift, it dynamically captures possible future acceleration drift risks and enhances adaptability to complex flight environments. Step S20: Based on the current flight status stage of the launch vehicle, compare the real-time parameters with the corresponding drift warning values, extract the parameters that exceed the drift warning values ​​as parameters to be optimized, analyze the changing trend of each parameter to be optimized, and calculate the predicted time period corresponding to the maximum drift limit. In step S30, for each parameter to be optimized, the correlation coefficient between the drift data of the parameter to be optimized and the attitude error of the launch vehicle is calculated, and the absolute value of the calculated correlation coefficient is used as the influence coefficient. The correlation coefficient can be calculated using the Pearson correlation coefficient. Obtain the influence coefficient of each parameter to be optimized, and perform a comprehensive analysis in conjunction with the prediction time period corresponding to each parameter to be optimized. The process is as follows: For each parameter to be optimized, the ratio of the corresponding prediction time period to the sum of the prediction time periods for all parameters to be optimized is taken as the time ratio; the difference between the influence coefficient and the time ratio is taken as the adjustment priority coefficient. It should be noted that the smaller the time ratio, the closer the parameter to be optimized is to the maximum drift limit and the more urgent the situation (the shorter the remaining time), and the higher the priority should be; the larger the influence coefficient, the greater the influence of the parameter to be optimized on the attitude of the launch vehicle, and the higher the priority should be; and the larger the adjustment priority coefficient, the higher the corresponding adjustment priority. Sort the parameters to be optimized according to their adjustment priority coefficients from largest to smallest. Based on the adjustment priority of the parameters to be optimized, optimize and compensate for the parameters to be optimized within a limited time. It needs to be explained that the time limit represents the minimum value within the predicted time period corresponding to all parameters to be optimized. The reason for using the time limit is as follows: When calculating the adjustment priority, both time and impact are considered. The attitude stability of the launch vehicle depends on the weakest link (i.e., the parameter that exceeds the limit first). If a longer predicted time period is used as the time limit, parameters that exceed the limit in a short period of time may not be processed in time, leading to a chain of failures. For example, parameter A is predicted to exceed the limit in 20 seconds (high impact coefficient, priority 1), and parameter B is predicted to exceed the limit in 30 seconds (low impact coefficient, priority 2). If the time limit is 30 seconds, parameter A may have already exceeded the limit in 20 seconds. At this time, it is meaningless to process parameter B. The system safety depends entirely on whether parameter A is adjusted within 20 seconds. Taking the minimum value of 20 seconds as the time limit forces the optimization of all parameters (including the low-priority parameter B) to be completed within 20 seconds, ensuring that the safety boundary of the most urgent parameter is not breached. Furthermore, setting the time limit to the minimum of all forecast time periods is a dual choice for risk control and resource optimization; Risk control: Using the remaining time of the most urgent parameter as the bottom line, ensuring that the safety limit of any parameter is not exceeded; Resource optimization: Allocating control actions according to priority within a unified time window to reduce ineffective adjustments caused by time misalignment; The optimization and compensation process is as follows: calculate the total compensation amount for each parameter to be optimized. The formula for calculating the total compensation amount is: Where BC represents the compensation amount, This represents the control gain, which is determined by the system design and can be set to 0.8. This indicates an adjustment of priority. This represents the current drift amount, and PX is the maximum drift limit. It should be explained that the calculation of the total compensation amount integrates three dimensions: control gain, adjustment priority, and remaining adjustment space. If the parameter adjustment priority is high and the remaining space is small, the total compensation amount will increase accordingly. If the remaining space is large, even if the priority is high, the compensation amount will be limited to reduce the possibility of over-adjustment. The control gain is set to reduce the possibility of system oscillation caused by excessive compensation amount. The ratio of the adjustment priority coefficient corresponding to each parameter to be optimized to the sum of the adjustment priority coefficients of all parameters to be optimized is calculated to obtain the control resource allocation ratio of each parameter to be optimized. The time limit is divided into several control cycles, and the total number of control cycles is obtained according to the proportion of control resources allocated to each parameter to be optimized. The single compensation amount for each control cycle is calculated based on the total compensation amount for each parameter to be optimized and the total number of control cycles. This step first calculates the correlation between the drift of the parameter to be optimized and the attitude error of the launch vehicle using the Pearson correlation coefficient. The absolute value of the correlation coefficient is used as the influence coefficient to measure the degree of influence of the parameter on the attitude. Then, the time ratio is calculated by combining the prediction time period of each parameter. The difference between the influence coefficient and the time ratio is used as the adjustment priority coefficient. The parameter adjustment priority is determined according to the coefficient from large to small. Then, the minimum value in all prediction time periods is used as the time limit. The total compensation amount of each parameter is calculated based on the adjustment priority. The time limit is divided into multiple control cycles according to the proportion of control time. The single compensation amount of each cycle is allocated to complete the optimization compensation of the parameter within the time limit. The optimization addresses the issue of parameters with high impact but long remaining time being over-competing for resources, or parameters with low impact but about to exceed limits being ignored, by sorting parameters solely based on their impact or remaining time. It also optimizes the accuracy of compensation adjustment by controlling the resource allocation ratio and periodic compensation to reasonably distribute the total compensation amount to each control cycle, reducing system oscillations caused by one-time adjustments, while ensuring the continuity and accuracy of compensation actions and optimizing the control effect. It has the following effects: By combining the degree of influence of parameters on attitude with the urgency of time, it achieves priority ranking based on both risk level and remaining time, reducing decision-making bias caused by using a single indicator; by using the shortest prediction period as the time limit, it requires all parameter adjustments to be completed before the critical time of the most urgent parameter exceeds the limit, ensuring that the system does not fail due to the premature exceedance of individual parameters, thus improving overall safety; and by dynamically allocating control cycles and compensation amounts according to priority, it efficiently executes optimization actions within a unified time window, reducing ineffective adjustments and improving the utilization efficiency of control resources. Example 2 Based on the same inventive concept as the attitude adaptive control method based on dynamic parameter optimization in the foregoing embodiments, such as Figure 2 As shown, this application provides an attitude adaptive control system based on dynamic parameter optimization, wherein the system specifically includes: Parameter benchmark definition module: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight state stages of the launch vehicle; This module is used to perform the following process: First, the flight status phases of the launch vehicle are divided into low-dynamic, high-dynamic, and stable flight phases. Then, the historical flight database of the launch vehicle is obtained. Next, the parameter data in the database is filtered to remove outliers and abnormal parameter values ​​for each flight status phase. Then, the mean and standard deviation of each parameter are calculated using the confidence interval method to determine the normal fluctuation range of the parameter. Based on this, the sum of the mean and k1 times the standard deviation is taken as the maximum value of the normal fluctuation range, and the difference between the mean and k2 times the standard deviation is taken as the minimum value. The parameter drift is the absolute value of the difference between the parameter value and the corresponding mean. Thus, the maximum drift limit and the drift warning value are obtained. Real-time monitoring and prediction module: Based on the current flight status stage of the launch vehicle, the real-time parameters are compared with the corresponding drift warning values, the parameters exceeding the drift warning values ​​are extracted as parameters to be optimized, the changing trend of each parameter to be optimized is analyzed and the prediction time period corresponding to the maximum drift limit is calculated; This module performs the following process: First, based on the current flight state of the launch vehicle, the drift of real-time parameters is compared with the corresponding drift warning value. If the drift exceeds the warning value, a warning signal is generated. Then, it is compared with the maximum drift limit value. Parameters exceeding the maximum drift limit value are extracted as parameters to be optimized. Subsequently, the time point when the warning signal is generated and the analysis process is triggered is used as the data window. The drift data within the window is linearly fitted, and the goodness of fit is judged by the coefficient of determination. If the goodness of fit is up to standard, the linear slope is used as the stable drift rate. If it is not up to standard, the maximum rate of change in the dataset is taken as the pre-calculated drift rate. The random forest model is then used to predict the future drift rate. Combining the historical maximum rate of change with the predicted value, the unstable drift rate is finally determined. Based on this, the predicted time period for the parameter to be optimized to reach the maximum drift limit value is calculated, providing a time reference for subsequent parameter optimization. Multi-dimensional decision compensation module: Obtain the influence coefficient of each parameter to be optimized on the attitude of the launch vehicle, and perform comprehensive analysis in combination with the prediction time period corresponding to each parameter to be optimized to obtain the adjustment priority of the parameter to be optimized. Based on the adjustment priority of the parameter to be optimized, optimize and compensate the parameter to be optimized within a limited time. This module performs the following process: First, it calculates the correlation between the drift of the parameter to be optimized and the attitude error of the launch vehicle using the Pearson correlation coefficient, and uses the absolute value of the correlation coefficient as the influence coefficient. Then, it calculates the time ratio based on the prediction time period of each parameter, and uses the difference between the influence coefficient and the time ratio as the adjustment priority coefficient. The parameter adjustment priority is determined according to the coefficient from large to small. Subsequently, the minimum value in all prediction time periods is used as the time limit. Based on the adjustment priority, the total compensation amount of each parameter is calculated, and the time limit is divided into multiple control cycles according to the proportion of control time. The single compensation amount of each cycle is allocated, and the optimization compensation of the parameters is completed within the time limit. The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An attitude adaptive control method based on dynamic parameter optimization, characterized in that: Includes the following steps: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight states of the launch vehicle. Based on the current flight status of the launch vehicle, the real-time parameters are compared with the corresponding drift warning values. Parameters exceeding the drift warning values ​​are extracted as parameters to be optimized and the calculation and analysis process is triggered. The changing trend of each parameter to be optimized is analyzed and the predicted time period corresponding to the maximum drift limit is calculated. The influence coefficient of each parameter to be optimized on the attitude of the launch vehicle is obtained, and a comprehensive analysis is performed in combination with the prediction time period corresponding to each parameter to be optimized to obtain the adjustment priority of the parameter to be optimized. Based on the adjustment priority of the parameter to be optimized, the parameter to be optimized is optimized and compensated within a limited time.

2. The attitude adaptive control method based on dynamic parameter optimization according to claim 1, characterized in that: The process for obtaining the maximum drift limit and the drift warning value is as follows: The confidence interval method is used to process the parameter data of each flight state stage in the historical flight database of the launch vehicle, calculate the mean and standard deviation of each parameter, take the sum of the mean and k1 times the standard deviation as the maximum value of the normal fluctuation range of the corresponding parameter, and take the difference between the mean and k2 times the standard deviation as the minimum value of the normal fluctuation range of the corresponding parameter. The maximum drift value and the drift warning value are obtained based on the normal fluctuation range of the parameters. The maximum drift value is k1 times the standard deviation, and the drift warning value is k2 times the standard deviation.

3. The attitude adaptive control method based on dynamic parameter optimization according to claim 1, characterized in that: The process of obtaining the parameters to be optimized is as follows: If the real-time parameter drift is greater than or equal to the drift warning value, a warning signal is generated; Based on the early warning signal, the real-time parameter drift is compared with the value close to the maximum drift limit; If the real-time parameter drift is greater than or equal to the maximum drift limit, then the parameter corresponding to the real-time parameter drift being greater than or equal to the maximum drift limit is extracted as the parameter to be optimized.

4. The attitude adaptive control method based on dynamic parameter optimization according to claim 1, characterized in that: The calculation and analysis process includes: Extract the time point corresponding to the generation of the early warning signal and the time point corresponding to the triggering of the calculation and analysis process as a data window; The parameter drift data in the data window is used as the dataset for trend analysis; the slope is calculated by linear fitting of the data in the dataset using the least squares method. The determination coefficient is used to determine whether the dataset conforms to linear change. If the determination coefficient is greater than or equal to the minimum acceptable goodness of fit, the slope is used as the rate of change of the steady drift of the corresponding parameter. Based on the rate of change of the steady drift, the prediction time period of the corresponding parameter is calculated by combining the real-time drift and the maximum drift limit. If the absolute coefficient is less than the minimum acceptable goodness of fit, the maximum rate of change of the drift in the dataset is taken as the pre-calculated rate of change of the drift; based on the pre-calculated rate of change of the drift, the corresponding pre-analysis and prediction time period is calculated by combining the real-time drift and the maximum drift limit. Based on the pre-analysis prediction time period, the pre-analysis time point for reaching the maximum drift limit is calculated in combination with the current time point.

5. The attitude adaptive control method based on dynamic parameter optimization according to claim 4, characterized in that: The calculation and analysis process also includes: The drift rate of change for each data point is integrated into a drift rate of change sequence, and a drift rate of change prediction model is trained. Using the drift rate of change prediction model, calculate and output the predicted drift rate of change sequence from the current time point to the pre-analysis time point; extract the maximum value in the predicted drift rate of change sequence; If the pre-calculated drift rate of change is greater than or equal to the maximum value in the predicted drift rate of change sequence, then the pre-calculated drift rate of change is taken as the unstable drift rate of change, and the analysis and prediction period is the prediction period of the corresponding parameter. If the pre-calculated drift rate is less than the maximum value in the predicted drift rate sequence, then the maximum value in the predicted drift rate sequence is taken as the unstable drift rate. Based on the unstable drift rate, the predicted time period for the corresponding parameters is calculated by combining the real-time drift and the maximum drift limit.

6. The attitude adaptive control method based on dynamic parameter optimization according to claim 1, characterized in that: The calculation process for the influence coefficient is as follows: Calculate the correlation coefficient between the drift data of the parameters to be optimized and the attitude error of the launch vehicle, and use the absolute value of the calculated correlation coefficient as the influence coefficient.

7. The attitude adaptive control method based on dynamic parameter optimization according to claim 5, characterized in that: The process for obtaining the adjustment priority of the parameter to be optimized is as follows: For each parameter to be optimized, the ratio of the corresponding prediction time period to the sum of the prediction time periods for all parameters to be optimized is taken as the time ratio; the difference between the influence coefficient and the time ratio is taken as the adjustment priority coefficient. The parameters to be optimized are sorted according to their adjustment priority coefficients from largest to smallest.

8. The attitude adaptive control method based on dynamic parameter optimization according to claim 5, characterized in that: The time limit is: The time limit represents the minimum value of the prediction time period corresponding to all parameters to be optimized.

9. The attitude adaptive control method based on dynamic parameter optimization according to claim 1, characterized in that: The optimization and compensation process is as follows: Calculate the total compensation for each parameter to be optimized; The ratio of the adjustment priority coefficient corresponding to each parameter to be optimized to the sum of the adjustment priority coefficients of all parameters to be optimized is calculated to obtain the control resource allocation ratio of each parameter to be optimized. The time limit is divided into several control cycles, and the total number of control cycles is obtained according to the proportion of control resources allocated to each parameter to be optimized. The single compensation amount for each control cycle is calculated based on the total compensation amount for each parameter to be optimized and the total number of control cycles.

10. An attitude adaptive control system based on dynamic parameter optimization, characterized in that, The system is used to perform the method according to any one of claims 1-9, the system comprising: Parameter benchmark definition module: Based on the historical flight database of the launch vehicle, obtain the maximum drift limit and drift warning value of the parameters corresponding to different flight state stages of the launch vehicle; Real-time monitoring and prediction module: Based on the current flight status stage of the launch vehicle, the real-time parameters are compared with the corresponding drift warning values. Parameters exceeding the drift warning values ​​are extracted as parameters to be optimized and the calculation and analysis process is triggered. The changing trend of each parameter to be optimized is analyzed and the predicted time period corresponding to the maximum drift limit is calculated. Multidimensional decision compensation module: Obtain the influence coefficient of each parameter to be optimized on the attitude of the launch vehicle, and perform comprehensive analysis in combination with the prediction time period corresponding to each parameter to be optimized to obtain the adjustment priority of the parameter to be optimized. Based on the adjustment priority of the parameter to be optimized, optimize and compensate the parameter to be optimized within a limited time.

Images