Brake performance enhancement method and system based on wheel control model

Abstract

This invention relates to a braking performance enhancement method and system based on a wheel control model, belonging to the field of aviation control. The method includes: identifying multiple sensor data sources; performing failure impact analysis to obtain multiple failure impact indicators; adjusting the initial failure probability based on failure sensitivity indicators to output multiple first failure probabilities; adjusting the multiple first failure probabilities using the multiple failure impact indicators to output multiple second failure probabilities; classifying sensor data sources into first-type and second-type identified sensor data sources; performing real-time data compensation on the first-type identified sensor data sources to output compensated sensor data; and generating optimized control parameters based on the compensated sensor data and the real-time sensor data corresponding to the second-type identified sensor data sources to achieve dynamic enhancement of braking performance. This invention solves the problem of existing aircraft braking performance enhancement methods' inability to accurately calculate and obtain the failure probabilities of sensor data sources under different braking mission scenarios.

Description

Technical Field

[0001] This invention belongs to the field of aviation control, specifically relating to a method and system for enhancing braking performance based on a wheel control model. Background Technology

[0002] With the increasing maturity of the aviation industry and the diversification of flight missions, the safety and reliability of aircraft, as an important part of modern transportation, have always been the focus of attention for the industry and the public. Among the many safety features of an aircraft, braking performance is particularly critical, as it directly relates to the aircraft's ability to decelerate and stop after landing, thereby affecting the safety of the aircraft and passengers. However, traditional aircraft braking systems are often limited by the limitations of the wheel control model, which makes it difficult to achieve ideal braking performance in certain situations. The aircraft's wheel control model will generate a failure probability of data source under different braking mission scenarios, and will be affected to varying degrees depending on the landing runway, thus affecting the quality of aircraft braking and increasing the risk of malfunctions and accidents.

[0003] Therefore, current technologies for enhancing aircraft braking performance suffer from technical problems such as the inability to accurately calculate and obtain the failure probability of sensor data sources under different braking mission scenarios, and the varying degrees of influence of different runway data on the failure probability, which in turn affects braking quality and leads to poor reliability and stability of braking performance. Summary of the Invention

[0004] The technical problem to be solved:

[0005] To overcome the shortcomings of existing technologies, this invention provides a braking performance enhancement method and system based on a wheel control model. Employing techniques such as multi-point parallel tracking, data dimensionality reduction, feature space construction, and information fusion modeling, it solves the technical problems of existing aircraft braking performance enhancement methods, including the inability to accurately calculate and obtain the failure probability of sensor data sources under different braking mission scenarios, and the varying degrees of influence of different runway data on the failure probability, which in turn affects braking quality and leads to poor reliability and stability of braking performance. This invention achieves intelligent monitoring and control of the braking system, thereby improving braking quality and the reliability and stability of braking performance.

[0006] The technical solution of this invention is: a braking performance enhancement method based on a wheel control model, the specific steps of which are as follows:

[0007] Connect the wheel control model and determine multiple sensor data sources for the wheel control model;

[0008] Acquire runway monitoring data, and perform failure impact analysis based on the runway monitoring data and the multiple sensor data sources to obtain multiple failure impact indicators;

[0009] Receive the first braking task, parse the task requirements of the first braking task, analyze the failure sensitivity indicators corresponding to the multiple sensor data sources according to the task requirements, calculate the initial failure probability of the multiple sensor data sources using a Bayesian network, adjust the initial failure probability in combination with the failure sensitivity indicators, and output multiple first failure probabilities.

[0010] The plurality of first failure probabilities are adjusted using the plurality of failure impact indicators to output a plurality of second failure probabilities;

[0011] Based on the plurality of second failure probabilities, the sensor data sources are divided into a first type of identification sensor data source and a second type of identification sensor data source, wherein the failure probability of the first type of identification sensor data source is greater than or equal to a preset threshold.

[0012] The first type of identification sensor data source is input into the enhancement compensation model for real-time data compensation, and the compensated sensor data is output.

[0013] Based on the compensation sensing data and the real-time sensing data corresponding to the second type of identification sensing data source, optimized control parameters are generated to achieve dynamic enhancement of braking performance.

[0014] A further technical solution of the present invention is: the analysis of the failure sensitivity index includes:

[0015] Define the task sensitivity matrix M m×n , of which M ij This represents the sensitivity of the i-th task sample to the j-th sensor data source, where i = 1, 2, 3, ..., m, j = 1, 2, 3, ..., n, m is the number of task type samples, and n is the number of sensor data sources;

[0016] Based on the task sensitivity matrix and failure sensitivity analysis model, a failure sensitivity vector is generated as multiple failure sensitivity indicators to adjust the initial failure probability.

[0017] A further technical solution of the present invention is: the construction of the failure sensitivity analysis model includes:

[0018] Perform a dot product or weighted sum operation on the task sensitivity matrix and the failure sensitivity vector to generate a total sensitivity index, which is the total sensitivity of each task to the failure of the sensor data source.

[0019] Based on the total sensitivity index, the weight distribution of the initial failure probability output by the Bayesian network is dynamically adjusted.

[0020] A further technical solution of the present invention is: the calculation of the initial failure probability of the multiple sensor data sources using a Bayesian network includes:

[0021] The historical data from the multiple sensor data sources are subjected to scale standardization to generate standardized sensor data samples.

[0022] A failure probability calculation model is established based on the standardized samples. The failure probability calculation model includes a prediction function with a mapping relationship and a Bernoulli probability distribution.

[0023] The multiple sensor data sources are input into the failure probability calculation model. The prediction outputs of multiple prediction functions are used as intermediate variables and input into multiple Bernoulli probability distributions to identify failure probabilities and obtain multiple initial failure probabilities.

[0024] A further technical solution of the present invention is: the expression of the failure probability calculation model is as follows:

[0025] P(F i (t)=1)=∫P(F i (t)=1|D i (t))P(D i (t))dD i (t);

[0026] In the formula, F i (t) represents the failure probability of the i-th sensor data source at time t, D i (t) represents the sensing data of the i-th sensing data source at time t, which follows a multivariate normal distribution. Let be the mean vector of a multivariate normal distribution. Let F be the covariance matrix of a multivariate normal distribution. i (t) Based on sensor data source D i The probability distribution of (t) is used to calculate the i-th sensor data source F under failure conditions. i The probability (t) = 1 is expressed by the integral representing the probability of all possible sensor data sources D. i (t) performs a weighted summation;

[0027] P(D i (t) is the prediction function, P(D) i (t))=∫∑ k∈S P(D i (t)|x k (t))P(x k This is used to calculate the sensor data source D at time t. i The probability distribution of (t), where S is the number of factors characterizing wheel control failure, P(x k ) represents the failure probability under the k-th factor.

[0028] A further technical solution of the present invention is: the expression of the enhanced compensation model is as follows:

[0029]

[0030] in, To compensate, α i To compensate for the weight parameters of the compensation strategy corresponding to the i-th data source in the first type of identification sensing data source based on time t, α i Optimization is achieved through gradient descent; D i (t) represents the real-time sensing data of the i-th data source in the first type of identification sensing data source based on time t. The predicted sensing data for the i-th data source in the first type of identification sensing data source is based on time t.

[0031] A further technical solution of the present invention is: the weighting parameter α i The optimizations include:

[0032] A braking performance evaluation model is established, which includes braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate.

[0033] Connect the input of the braking performance evaluation model to the output of the enhancement compensation model;

[0034] The compensation strategy for each round of optimization is evaluated based on the braking performance evaluation model to obtain the performance enhancement rate.

[0035] Based on the performance enhancement rate and the preset performance enhancement rate, α is iteratively optimized using the gradient descent algorithm. i Until the preset performance enhancement rate is met.

[0036] A braking performance enhancement system based on a wheel control model includes:

[0037] The sensor data source determination module is used to connect to the wheel control model and determine multiple sensor data sources;

[0038] The failure impact index acquisition module is used to acquire runway monitoring data, analyze the failure impact relationship between runway monitoring data and sensor data sources, and generate failure impact indexes.

[0039] The first failure probability output module is used to calculate and adjust the initial failure probability based on the Bayesian network and the task sensitivity matrix, and output the first failure probability.

[0040] The second failure probability output module is used to dynamically correct the first failure probability by combining the failure impact index and output the second failure probability.

[0041] The identification sensor data source acquisition module is used to classify the first type and the second type of sensor data sources according to the second failure probability.

[0042] The compensation sensor data output module is used to compensate the first type of sensor data in real time through the enhancement compensation model;

[0043] The optimized control parameter acquisition module is used to fuse compensation data and second-type data to generate optimized control parameters.

[0044] A further technical solution of the present invention is: the first failure probability output module includes:

[0045] The standardization processing unit is used to standardize the historical data from the sensor data source.

[0046] The failure probability calculation unit establishes a mapping model between the prediction function and the Bernoulli probability distribution based on standardized data, and outputs the initial failure probability.

[0047] A further technical solution of the present invention is: the optimized control parameter acquisition module quantifies the performance enhancement rate through a braking performance evaluation model, the braking performance evaluation model including braking distance enhancement rate, response time enhancement rate and braking stability enhancement rate, and iteratively optimizes the control parameters through a gradient descent algorithm.

[0048] Beneficial effects

[0049] The beneficial effects of this invention are as follows:

[0050] 1. This invention improves the overall performance of the braking system from 80% to 92% using dynamic failure probability calculation and enhanced compensation mechanism, effectively shortening the braking distance and increasing the response speed.

[0051] 2. This invention combines runway monitoring data with Bayesian networks to accurately quantify the impact of different runway conditions (such as humidity and friction coefficient) on sensor failure, thereby reducing the risk of brake failure caused by environmental changes.

[0052] 3. This invention uses gradient descent-optimized weight parameters to perform real-time compensation for sensor data with high failure probability, reducing the interference of single sensor failure on the system and improving data robustness.

[0053] 4. This invention introduces a multi-index evaluation model that includes braking distance enhancement rate, response time enhancement rate, and stability enhancement rate, to achieve synergistic optimization of the braking system in terms of safety, efficiency, and stability.

[0054] 5. This invention quantifies the dependence of different braking tasks on sensors through a task sensitivity matrix and dynamically adjusts the failure probability threshold, enabling the system to flexibly adapt to various scenario requirements such as emergency braking and coasting deceleration.

[0055] 6. Based on the standardized processing of historical data and failure probability prediction, this invention can identify potential faulty sensors in advance, reducing maintenance downtime and costs caused by sudden failures.

[0056] 7. This invention optimizes the control parameter generation module to enable the braking system to make autonomous decisions and adjust in real time, reducing the need for manual intervention and improving operational efficiency and safety. Attached Figure Description

[0057] Figure 1 A schematic flowchart of a braking performance enhancement method based on a wheel control model provided in this application embodiment;

[0058] Figure 2 A schematic diagram of the braking performance enhancement system based on the wheel control model provided in this application embodiment;

[0059] Figure 3 This is a test curve of the control rate of the traditional braking control method;

[0060] Figure 4 The control rate test curve is shown after adopting this method. Detailed Implementation

[0061] The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0062] Current technologies for enhancing aircraft braking performance suffer from several drawbacks. These include the inability to accurately calculate the failure probability of sensor data sources under different braking scenarios, and the varying degrees of influence of different runway data on the failure probability. These issues, which affect braking quality and lead to poor reliability and stability, necessitate a braking performance enhancement method based on a wheel control model. The specific steps are as follows:

[0063] Step S100: Connect the wheel control model and determine multiple sensor data sources of the wheel control model;

[0064] Step S200: Obtain runway monitoring data, and perform failure impact analysis based on the runway monitoring data and the multiple sensor data sources to obtain multiple failure impact indicators;

[0065] Step S300: Receive the first braking task, parse the task requirements of the first braking task, analyze the failure sensitivity indicators corresponding to the multiple sensor data sources according to the task requirements, calculate the initial failure probability of the multiple sensor data sources using a Bayesian network, adjust the initial failure probability in combination with the failure sensitivity indicators, and output multiple first failure probabilities.

[0066] Step S400: Adjust the plurality of first failure probabilities using the plurality of failure impact indicators, and output a plurality of second failure probabilities;

[0067] Step S500: Based on the plurality of second failure probabilities, the sensor data source is divided into a first type of identification sensor data source and a second type of identification sensor data source, wherein the failure probability of the first type of identification sensor data source is greater than or equal to a preset threshold.

[0068] Step S600: Input the first type of identification sensor data source into the enhancement compensation model for real-time data compensation, and output the compensated sensor data;

[0069] Step S700: Based on the compensation sensing data and the real-time sensing data corresponding to the second type of identification sensing data source, generate optimized control parameters to achieve dynamic enhancement of braking performance.

[0070] In a possible implementation, the analysis of the failure sensitivity index includes: collecting task type samples; and defining a task sensitivity matrix M based on the task type samples, where M... ij The sensitivity of the i-th task sample to the j-th sensor data source is included. The size of M is m×n, where m is the number of samples of the task type and n is the number of sensor data sources. A failure sensitivity analysis model is established based on the task sensitivity matrix M and the vector that identifies the magnitude of failure sensitivity. The task requirements are input into the failure sensitivity analysis model to identify multiple failure sensitivity vectors, which are output as multiple failure sensitivity indicators.

[0071] In a possible implementation, a Bayesian network is used to calculate multiple initial failure probabilities corresponding to the multiple sensor data sources. The following processing is also performed: historical data from the multiple sensor data sources are scaled to obtain multiple standardized sensor data samples; a failure probability calculation model is established based on the multiple standardized sensor data samples, the failure probability calculation model including multiple prediction functions and multiple Bernoulli probability distributions, the multiple prediction functions and multiple Bernoulli probability distributions having a mapping relationship; the multiple sensor data sources are input into the failure probability calculation model, and the prediction outputs of the multiple prediction functions are used as intermediate variables, input into the multiple Bernoulli probability distributions for failure probability identification, thereby obtaining multiple initial failure probabilities.

[0072] In possible implementations, the failure probability calculation model includes:

[0073] P(F i (t)=1)=∫P(F i (t)=1|D i (t))P(D i (t))dD i (t);

[0074] Among them, F i (t) represents the failure probability of the i-th sensor data source at time t, D i (t) represents the sensing data of the i-th sensing data source at time t, which follows a multivariate normal distribution. Let σ be the mean vector of a multivariate normal distribution. Di Let F be the covariance matrix of a multivariate normal distribution. i (t) Based on sensor data source D i The probability distribution of (t) is used to calculate the i-th sensor data source F under failure conditions. i The probability (t) = 1 is expressed by the integral representing the probability of all possible sensor data sources D. i (t) performs a weighted summation;

[0075] P(D i (t) is the prediction function, P(D) i (t))=∫∑ k∈S P(D i (t)|x k (t))P(x k This is used to calculate the sensor data source D at time t. i The probability distribution of (t), where S is the number of factors characterizing wheel control failure, P(x k ) represents the failure probability under the k-th factor.

[0076] In a possible implementation, the expression for the enhanced compensation model is as follows:

[0077]

[0078] in, To compensate, α i To compensate for the weight parameters of the compensation strategy corresponding to the i-th data source in the first type of identification sensing data source based on time t, α i Optimization is achieved through gradient descent;

[0079] D i (t) represents the real-time sensing data of the i-th data source in the first type of identification sensing data source based on time t. The predicted sensing data for the i-th data source in the first type of identification sensing data source is based on time t.

[0080] In a possible implementation, the optimized control parameters of the wheel control model are obtained using the compensation sensing data corresponding to the first type of identified sensing data source and the real-time sensing data corresponding to the second type of identified sensing data source. The following processing is also performed: establishing a braking performance evaluation model, the input of which is connected to the output of the enhanced compensation model, the braking performance evaluation model including braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate; evaluating the compensation strategy for each round of optimization based on the braking performance evaluation model to obtain the performance enhancement rate; and, based on the performance enhancement rate and a preset performance enhancement rate, adjusting the weight parameter α in the enhanced compensation model using gradient descent. i To find the best option.

[0081] This application also provides a braking performance enhancement system based on a wheel control model, including:

[0082] A sensor data source determination module is used to connect to the wheel control model and determine multiple sensor data sources of the wheel control model.

[0083] A failure impact index acquisition module is used to acquire runway monitoring data, perform failure impact analysis based on the runway monitoring data and the multiple sensor data sources, and acquire multiple failure impact indices.

[0084] The first failure probability output module is used by the wheel control model to receive a first braking task, calculate the failure probability of the multiple sensor data sources according to the first braking task, and output multiple first failure probabilities.

[0085] The second failure probability output module is used to adjust the plurality of first failure probabilities with the plurality of failure impact indicators and output a plurality of second failure probabilities.

[0086] The identification sensor data source acquisition module acquires a first type of identification sensor data source and a second type of identification sensor data source based on the plurality of second failure probabilities, wherein the first type of identification sensor data source is a sensor data source with a failure probability greater than or equal to a preset failure probability, and the second type of identification sensor data source is a sensor data source with a failure probability less than the preset failure probability.

[0087] The compensation sensor data output module is used to input the first type of identification sensor data source into the enhancement compensation model to enhance and compensate the real-time sensor data corresponding to the first type of identification sensor data source, and output the compensation sensor data corresponding to the first type of identification sensor data source.

[0088] An optimized control parameter acquisition module is used to acquire the optimized control parameters of the turbine control model using compensation sensing data corresponding to a first type of identified sensing data source and real-time sensing data corresponding to a second type of identified sensing data source.

[0089] Specifically, the first failure probability output module further includes:

[0090] The standardization processing unit is used to standardize the historical data from the sensor data source.

[0091] The failure probability calculation unit establishes a mapping model between the prediction function and the Bernoulli probability distribution based on standardized data, and outputs the initial failure probability.

[0092] Specifically, the optimized control parameter acquisition module quantifies the performance enhancement rate through a braking performance evaluation model, which includes braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate, and iteratively optimizes the control parameters through a gradient descent algorithm.

[0093] This application proposes a braking performance enhancement method and system based on a wheel control model. The method involves identifying multiple sensor data sources for the wheel control model; performing failure impact analysis based on runway monitoring data and these data sources to obtain multiple failure impact indicators; calculating failure probabilities based on a first braking task to output multiple first failure probabilities; adjusting these first failure probabilities using the multiple failure impact indicators to output multiple second failure probabilities; acquiring first and second type-specific identification sensor data sources; inputting the first type-specific identification sensor data sources into the enhancement compensation model for enhancement compensation, outputting the compensated sensor data corresponding to the first type-specific identification sensor data sources; and obtaining the optimized control parameters of the wheel control model. This method solves the technical problems of existing aircraft braking performance enhancement methods, such as the inability to accurately calculate and obtain the failure probabilities of sensor data sources under different braking task scenarios, and the varying degrees of influence of different runway data on the failure probabilities, which in turn affects braking quality and leads to poor reliability and stability of braking performance. It achieves intelligent monitoring and control of the braking system, improving braking performance from 80% to 92%.

[0094] The above technical solution will be further explained below with reference to the accompanying drawings:

[0095] In one embodiment, refer to Figure 1 As shown, this embodiment provides a braking performance enhancement method based on a wheel control model, including the following steps:

[0096] Step S100: Connect the wheel control model and determine multiple sensor data sources for the wheel control model. The wheel control model is a model used in the aircraft braking system to control the wheel state (such as speed, acceleration, etc.), describing the dynamic behavior of the wheels during braking. The sensor data sources refer to various real-time data provided to the wheel control model by multiple sensors installed on the aircraft, including various data related to braking performance. These sensors are connected to the wheel control model via wired (e.g., RS232, 4-20mA, etc.) or wireless (e.g., WiFi, Bluetooth, etc.). Specifically, the multiple sensor data sources may include, but are not limited to, wheel speed sensors, wheel temperature sensors, pressure sensors, etc. The wheel speed sensor is used for real-time monitoring. The rotational speed of the wheel is one of the key parameters in brake control. Through the wheel speed sensor, the wheel control model can accurately understand the real-time speed of the wheel, and thus adjust the magnitude and distribution of braking force according to speed changes. The wheel temperature sensor is used to monitor the temperature change of the wheel during braking. Excessive wheel temperature may lead to a decrease in braking performance or even failure. Therefore, temperature data is crucial for the control and protection of braking performance. The pressure sensor is used to monitor the pressure changes inside the braking system to ensure that the braking system is in normal working condition. Pressure data can help the wheel control model determine whether there are any abnormalities in the braking system and take measures in advance to avoid failure.

[0097] Step S200: Obtain runway monitoring data, and perform failure impact analysis based on the runway monitoring data and the multiple sensor data sources to obtain multiple failure impact indicators. Runway monitoring data is collected through specialized sensors, cameras, weather stations, and other equipment installed on the runway. This data includes key information such as runway humidity, runway temperature, friction coefficient, and pollution levels. Runway monitoring data directly impacts braking performance and effectiveness because different runway conditions significantly affect braking performance. Specifically, failure impact analysis based on runway monitoring data and multiple sensor data sources assesses the potential impact of different factors (such as runway conditions and sensor data) on braking system performance. The analysis considers various failure modes (such as sensor failure and changes in runway conditions) and their specific impact on braking performance, thereby obtaining multiple failure impact indicators. These indicators quantify the degree of impact of different failure modes on braking performance. Failure impact indicators may include the increase in braking distance (the additional braking distance due to a certain failure mode compared to normal conditions), the increase in braking time (the increase in braking time due to a certain failure mode compared to normal conditions), and the decrease in braking efficiency (the percentage decrease in braking efficiency due to a certain failure mode compared to normal conditions), used to assess the specific impact of different failure modes on braking performance. In summary, by analyzing the failure impact of runway monitoring data and multiple sensor data sources, multiple failure impact indicators are obtained, the influence of different factors on braking performance is assessed, and ground contact factor data is integrated into the wheel control model. This allows for real-time understanding of runway conditions, thereby enabling dynamic adjustment of braking strategies.

[0098] In step S300, the wheel control model receives a first braking task, calculates the failure probability of the multiple sensor data sources based on the first braking task, and outputs multiple first failure probabilities. When the aircraft performs a braking action, the wheel control model receives a braking task, such as emergency braking or deceleration taxiing, which is the first braking task. Calculating the failure probability of the multiple sensor data sources means that before performing a braking task, the wheel control model evaluates the reliability of multiple sensor data sources (such as wheel speed sensors, temperature sensors, etc.). The failure probability refers to the likelihood that a certain sensor data source will fail (i.e., malfunction or provide inaccurate data) when the aircraft performs a braking task. The calculation of the failure probability is usually based on historical data. For example, for a wheel speed sensor, the failure probability of the current wheel speed sensor can be estimated based on its performance in similar braking tasks in the past (number of failures, error range, etc.). For each sensor... According to the source, the wheel control model calculates a failure probability corresponding to the first braking task, i.e., the first failure probability. These first failure probabilities will serve as important indicators for evaluating the reliability of the sensor data source. For example, assuming the wheel control model is connected to 5 sensor data sources (A, B, C, D, E), and the historical failure probabilities of each sensor data source in a similar first braking task are 0.01 (A), 0.02 (B), 0.005 (C), 0.015 (D), and 0.03 (E), respectively, when the wheel control model receives the first braking task, it will recalculate the failure probability of each sensor data source based on these historical data and the current operating environment (such as temperature, humidity, etc.) and output a new first failure probability.

[0099] In one possible implementation, step S300 further includes step S310, parsing the task requirements of the first braking task and analyzing multiple failure sensitivity indicators corresponding to the multiple sensor data sources based on the task requirements. The task requirement of the first braking task is to ensure that the aircraft can safely and effectively decelerate and stop during emergency braking or deceleration. Based on the parsed task requirements of the first braking task, multiple failure sensitivity indicators corresponding to the multiple sensor data sources of the first braking task are analyzed. It also includes step S320, using a Bayesian network to calculate multiple initial failure probabilities corresponding to the multiple sensor data sources. A Bayesian network (also known as a directed acyclic graph model) consists of nodes representing random variables and directed edges connecting these nodes (representing the relationships between nodes, i.e., a parent node pointing to its child node). Conditional probabilities are used to express the strength of these relationships. In a braking system, the state of the sensor data source is random and may be affected by many factors. A Bayesian network is used to represent the state of these sensor data sources and the relationships between their influencing factors. Specifically, the state of the sensor data source (normal / failed) is treated as a random variable, and factors that may cause the sensor data source to fail (such as temperature, humidity, mechanical wear, etc.) are treated as influencing factors. A directed acyclic graph is used to represent the state of the sensor data source and the relationships between its influencing factors. Each node in the graph represents a random variable or influencing factor, and directed edges represent the dependencies between them. For each node, its conditional probability distribution is determined, describing the probability of the child node's state given the state of the parent node. Using the conditional probability distribution of the Bayesian network, combined with observed data (such as the current state of the sensor data source, environmental factors, etc.), the initial failure probability of the sensor data source can be calculated.

[0100] Step S300 further includes step S330, adjusting the multiple initial failure probabilities based on the multiple failure-sensitive indicators to obtain multiple first failure probabilities. If the data output by the sensor data source deviates significantly from the expected value, its failure probability may increase. During adjustment, the initial failure probability can be increased or decreased according to the magnitude and direction of the deviation. For example, a threshold can be set for each failure-sensitive indicator. When the performance indicators of the sensor data source exceed these thresholds, the adjustment of the initial failure probability will be triggered. Alternatively, a weight can be assigned to each sensitive indicator to represent its importance in the failure risk assessment. Then, the initial failure probability is adjusted by weighting according to the actual performance indicators of the sensor data source and the corresponding weights, ultimately obtaining multiple first failure probabilities.

[0101] In one possible implementation, step S310 further includes step S311, collecting task type samples. The task types include braking tasks under various conditions. It also includes step S312, defining a task sensitivity matrix M based on the task type samples, where M... ijThe sensitivity matrix M includes the sensitivity of the i-th task sample to the j-th sensor data source. M is of size m×n, where m is the number of samples of the task type and n is the number of sensor data sources. The task sensitivity matrix M represents the sensitivity between different braking task type samples (m samples) and sensor data sources (n data sources). The number of rows m represents the number of task type samples, and the number of columns n represents the number of sensor data sources. Sensitivity represents the degree to which a task type depends on a specific sensor data source or the importance of that sensor data source to the task. Sensitivity can be represented numerically, for example, as a floating-point number between 0 and 1, where 0 indicates insensitivity (i.e., the sensor data source has no impact on the task) and 1 indicates high sensitivity (i.e., the sensor data source is crucial to the task).

[0102] Step S310 further includes step S313, establishing a failure sensitivity analysis model based on the task sensitivity matrix M and the vector indicating the magnitude of failure sensitivity. The vector indicating the magnitude of failure sensitivity is the same as the number n of sensor data sources, and each element represents the magnitude of failure sensitivity of the corresponding sensor data source. The model parameters are initialized, including the task sensitivity matrix M and the failure sensitivity vector. The total sensitivity of each task to sensor data source failure is calculated, which can be achieved by performing a dot product (or weighted sum) between each row in M ​​(representing a task) and the failure sensitivity vector, representing the sensitivity of the task to failure of all sensor data sources. The total sensitivity is sorted as needed to more intuitively compare the sensitivity differences between different tasks. The model result is output, i.e., the total sensitivity of each task to sensor data source failure. This failure sensitivity analysis model is used to evaluate the sensitivity of different task types to sensor data source failure. Step S314 further includes inputting the task requirements into the failure sensitivity analysis model, identifying multiple failure sensitivity vectors, and outputting them as multiple failure sensitivity indicators. The specific task requirements are input into the previously established failure sensitivity analysis model. Based on the task requirements, the corresponding row (i.e., the task sample) in the task sensitivity matrix M is found. The selected row (task sample) is then multiplied by the vector that indicates the magnitude of the failure sensitivity to obtain the total sensitivity of the task to the failure of all sensor data sources. The calculated total sensitivity is used as an element in the vector. The above steps are performed for each sensor data source, ultimately forming a failure sensitivity vector equal to the number of sensor data sources. This vector represents the sensitivity of a specific task to the failure of each sensor data source. For each input task requirement, a corresponding failure sensitivity vector is obtained, which consists of multiple failure sensitivity indicators that quantitatively represent the sensitivity of different tasks to the failure of sensor data sources.

[0103] Step S320 further includes step S321, which involves performing scale standardization on the historical data from the multiple sensor data sources to obtain multiple standardized sensor data samples. Obtaining historical data from multiple sensor data sources includes a sufficient time span and data points to reflect the changing trends and characteristics of the data sources. Scale standardization refers to using standardization methods such as Z-score standardization (standard deviation standardization) or Min-Max standardization (linear transformation) to convert the original data (historical data from the sensor data sources) into a distribution with a mean of 0 and a standard deviation of 1, or a specified range (usually 0 to 1), to obtain the corresponding multiple standardized sensor data samples. Step S322 further includes establishing a failure probability calculation model based on the multiple standardized sensor data samples. The failure probability calculation model includes multiple prediction functions and multiple Bernoulli probability distributions, and the multiple prediction functions and multiple Bernoulli probability distributions have a mapping relationship. A failure probability calculation model is established based on multiple standardized sensor data samples. This model includes multiple prediction functions and multiple Bernoulli probability distributions, which are mapped to each other. Specifically, the prediction function is a mathematical function used to predict the future state (e.g., failure) of a system or component based on input data (i.e., standardized sensor data samples). The Bernoulli probability distribution is a discrete probability distribution used to describe random trials with only two possible outcomes (success or failure), i.e., failure or no failure. The Bernoulli distribution has only one parameter, representing the probability of no failure. In the failure probability calculation model, the Bernoulli distribution is used to quantify the failure probability output by the prediction function. The prediction function outputs a value between 0 and 1, which can be interpreted as the probability of component failure (or 1 minus this value, i.e., the probability of component no failure). The model also includes step S323, where the multiple sensor data sources are input into the failure probability calculation model, and the prediction outputs of the multiple prediction functions are used as intermediate variables to input into the multiple Bernoulli probability distributions for failure probability identification, thereby obtaining multiple initial failure probabilities. Multiple sensor data sources (after standardization) are used as input data and fed into the failure probability calculation model. The output of the prediction function is regarded as an intermediate variable. The output of the prediction function (i.e., the intermediate variable) is used as a parameter and input into the associated Bernoulli probability distribution. The Bernoulli probability distribution is used to map the continuous output value of the prediction function (probability value between 0 and 1) to discrete failure states (usually represented as 1 for failure and 0 for normal). The output value of the prediction function (e.g., 0.3) is used as the probability of success in the Bernoulli distribution (i.e., the probability of no failure). Then, the probability of failure (i.e., 1 minus the probability of no failure) is calculated according to the Bernoulli distribution. Finally, the failure probability associated with each prediction function is obtained, which is the initial failure probability.

[0104] In one possible implementation, step S322 further includes the failure probability calculation model comprising:

[0105] P(F i (t)=1)=∫P(F i (t)=1|D i (t))P(D i (t))dD i (t);

[0106] F i (t) represents the failure probability of the i-th sensor data source at time t, D i (t) represents the sensing data of the i-th sensing data source at time t, which follows a multivariate normal distribution. Let be the mean vector of a multivariate normal distribution. Let F be the covariance matrix of a multivariate normal distribution. i (t) Based on sensor data source D i The probability distribution of (t) is used to calculate the i-th sensor data source F under failure conditions. i The probability (t) = 1 is expressed by the integral representing the probability of all possible sensor data sources D. i (t) is used for weighted summation. Where F i (t) is a binary variable representing whether the corresponding sensor data source is faulty or not, F i The probability of (t) = 1 refers to the failure probability.

[0107] P(D i (t) is the prediction function, P(D) i (t))=∫∑ k∈S P(D i (t)|x k (t))P(x k This is used to calculate the sensor data source D at time t. i The probability distribution of (t), where S is the number of factors characterizing wheel control failure, P(x k Let be the failure probability under the k-th factor. The set of factors for wheel control failure may include wheel speed, tire pressure, temperature, and friction coefficient, etc.

[0108] Step S400: Adjust the multiple first failure probabilities using the multiple failure impact indicators to output multiple second failure probabilities. For each sensor data source's corresponding first failure probability, the wheel control model adjusts it with reference to its corresponding failure impact indicators (such as the increase in braking distance, the extension of braking time, the degree of reduction in braking efficiency, etc.). For example, assuming that the failure of a certain sensor data source (such as a wheel speed sensor) has historically resulted in a 20% increase in braking distance, when the first failure probability of this sensor data source is 0.02, the wheel control model may combine this first failure probability with the failure impact indicators, that is, considering the serious consequences of the failure, and output a higher second failure probability, such as 0.03. By combining the failure impact indicators to adjust the multiple first failure probabilities and outputting the corresponding multiple second failure probabilities, the likelihood and consequences of each sensor data source failure during aircraft braking operations can be more accurately reflected.

[0109] Step S500: Based on the plurality of second failure probabilities, obtain a first type of identification sensor data source and a second type of identification sensor data source, wherein the first type of identification sensor data source is a sensor data source with a failure probability greater than or equal to a preset failure probability, and the second type of identification sensor data source is a sensor data source with a failure probability less than the preset failure probability. The preset failure probability is a threshold set based on historical data to determine whether a sensor data source is likely to fail during the upcoming braking task. Based on a comparison of the second failure probability of each sensor data source with the preset failure probability, the wheel control model categorizes sensor data sources into two types: a first-class identified sensor data source and a second-class identified sensor data source. Specifically, if the second failure probability of a first-class identified sensor data source is greater than or equal to the preset failure probability, it is considered a sensor data source that may fail or provide inaccurate data during the upcoming braking task. For first-class identified sensor data sources, the wheel control model may take some compensatory measures or temporarily suspend the use of their sensor data to ensure the safety and reliability of the braking system. If the second failure probability of a second-class identified sensor data source is less than the preset failure probability, it is considered a relatively reliable sensor data source that is unlikely to fail or provide inaccurate sensor data during the upcoming braking task. The wheel control model can place greater trust in the data provided by these sensor data sources for braking decisions and control.

[0110] In one possible implementation, step S500 further includes step S510, where the expression for the enhanced compensation model is as follows:

[0111]

[0112] in, To compensate, α iTo compensate for the weight parameters of the compensation strategy corresponding to the i-th data source in the first type of identification sensing data source based on time t, α i Optimization is achieved through gradient descent; D i (t) represents the real-time sensing data of the i-th data source in the first type of identification sensing data source based on time t. The predicted sensing data for the i-th data source in the first type of identification sensing data source is based on time t.

[0113] In one possible implementation, step S500 further includes step S520, establishing a braking performance evaluation model, wherein the input of the braking performance evaluation model is connected to the output of the enhancement compensation model, and the braking performance evaluation model includes braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate. The braking performance evaluation model is a model used to quantitatively evaluate the performance of a car's braking system. It evaluates the braking system's performance under different conditions through a series of parameters and indicators. The input of the braking performance evaluation model is connected to the output of the enhancement compensation model, meaning that the input data of the evaluation model comes from the output of the enhancement compensation model. Specifically, the braking distance enhancement rate reflects the proportion by which the braking distance is reduced compared to the original state after the braking system has been enhanced or improved under the same conditions. For example, if the original braking distance is 18 meters and the braking distance is 15 meters after enhancement, then the braking distance enhancement rate is (18-15) / 18 = 16.7%. Braking response time refers to the time required from when the pilot starts braking to when the aircraft begins to decelerate. The response time enhancement rate reflects the proportion by which the braking system's response time is shortened. A shorter response time means that the braking system is more sensitive. The braking stability enhancement rate reflects the proportion by which the braking system's ability to maintain the aircraft's attitude stability during braking is improved after enhancement.

[0114] Step S500 further includes step S530, evaluating the compensation strategy optimized in each round according to the braking performance evaluation model to obtain the performance enhancement rate. The compensation strategy obtained in each round of optimization is applied to the braking system. The evaluation model quantifies the effect of the compensation strategy based on parameters such as braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate. By comparing the braking distance before and after applying the compensation strategy, the reduction ratio of braking distance and the shortening ratio of response time are calculated. Simulation tests are used to evaluate the degree of vehicle attitude change during braking and calculate the stability improvement ratio. The combined results obtain the performance enhancement rate, reflecting the enhancement and improvement effect of the current compensation strategy on the braking system performance. Step S540 further includes, based on the performance enhancement rate and a preset performance enhancement rate, using gradient descent to adjust the weight parameter α in the enhancement compensation model. iOptimization is performed. The preset performance enhancement rate is a pre-defined target value for the expected performance improvement of the braking system. The performance enhancement rate is compared with the preset performance enhancement rate, and the error (such as mean squared error, MSE) is calculated. Gradient descent is used to calculate the gradient (i.e., the partial derivative of the loss function with respect to the weight parameters) based on the error. The gradient shows how to adjust the weight parameters to reduce the error. Then, based on the learning rate (the step size of gradient descent, a positive number) and the calculated gradient, the weight parameters are updated to obtain the optimal weight parameters in the enhancement compensation model.

[0115] Step S600: The first type of identification sensor data source is input into the enhancement compensation model. Enhancement compensation is performed on the real-time sensing data corresponding to the first type of identification sensor data source, and the compensated sensing data corresponding to the first type of identification sensor data source is output. The enhancement compensation model is a pre-trained model that processes the input real-time sensing data to eliminate or reduce data errors caused by sensor failure or other noise and other interference factors. Specifically, during the operation of the wheel control model, the first type of identification sensor data source continuously generates real-time sensing data, reflecting the real-time status of the wheel, braking system, or other related components. Sensor data sources identified as first type (i.e., sensor data sources whose second failure probability is greater than or equal to a preset failure probability) are used as input into the enhancement compensation model. The enhancement compensation model enhances and compensates the real-time sensing data corresponding to the first type of identification sensor data source, and outputs the compensated sensing data corresponding to the first type of identification sensor data source. This is the result of the enhancement compensation model's enhancement compensation of the real-time sensing data, which has higher accuracy and reliability compared to the original real-time sensing data.

[0116] Step S700: Using the compensated sensing data corresponding to the first type of identification sensing data source and the real-time sensing data corresponding to the second type of identification sensing data source, the optimized control parameters of the wheel control model are obtained. In the wheel control model, the compensated sensing data corresponding to the first type of identification sensing data source (sensor data source with a failure probability greater than or equal to a preset value) and the real-time sensing data corresponding to the second type of identification sensing data source (sensor data source with a failure probability less than a preset value) are used to obtain the optimized control parameters. Specifically, the compensated data from the first type of sensing data source and the real-time sensing data from the second type of sensing data source are integrated. Based on the integrated data, the wheel control model begins to calculate the optimized control parameters, for example, using gradient descent, root locus method, etc., to calculate the control parameters that enable the wheel control model to achieve optimal performance, best stability, and minimum error. After the optimized control parameter calculation, the wheel control model outputs the optimized control parameters to guide the actual control process of the wheel, ensuring that the wheel can exhibit excellent performance under various operating conditions.

[0117] In the above text, refer to Figure 1A braking performance enhancement method based on a wheel control model according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 A braking performance enhancement system based on a wheel control model according to an embodiment of the present invention is described.

[0118] The braking performance enhancement system based on a wheel control model according to an embodiment of the present invention addresses the technical problems of existing aircraft braking performance enhancement methods, such as the inability to accurately calculate and obtain the failure probability of sensor data sources under different braking mission scenarios, and the varying degrees of influence of different runway data on the failure probability, which in turn affects braking quality and leads to poor reliability and stability of braking performance. The system achieves intelligent monitoring and control of the braking system, thereby improving braking quality and the reliability and stability of braking performance. The braking performance enhancement system based on a wheel control model includes: a sensor data source determination module 10, a failure impact index acquisition module 20, a first failure probability output module 30, a second failure probability output module 40, an identification sensor data source acquisition module 50, a compensation sensor data output module 60, and an optimized control parameter acquisition module 70.

[0119] The sensor data source determination module 10 is used to connect to the wheel control model and determine multiple sensor data sources of the wheel control model.

[0120] Failure impact index acquisition module 20 is used to acquire runway monitoring data, perform failure impact analysis based on the runway monitoring data and the multiple sensor data sources, and acquire multiple failure impact indices.

[0121] The first failure probability output module 30 is used for the wheel control model to receive a first braking task, calculate the failure probability of the multiple sensor data sources according to the first braking task, and output multiple first failure probabilities.

[0122] The second failure probability output module 40 is used to adjust the plurality of first failure probabilities with the plurality of failure impact indicators and output a plurality of second failure probabilities.

[0123] The identification sensor data source acquisition module 50 acquires a first type of identification sensor data source and a second type of identification sensor data source based on the plurality of second failure probabilities. The first type of identification sensor data source is a sensor data source with a failure probability greater than or equal to a preset failure probability, and the second type of identification sensor data source is a sensor data source with a failure probability less than the preset failure probability.

[0124] The compensation sensor data output module 60 is used to input the first type of identification sensor data source into the enhancement compensation model to enhance and compensate the real-time sensor data corresponding to the first type of identification sensor data source, and output the compensation sensor data corresponding to the first type of identification sensor data source.

[0125] The optimized control parameter acquisition module 70 is used to acquire the optimized control parameters of the turbine control model using the compensation sensing data corresponding to the first type of identified sensing data source and the real-time sensing data corresponding to the second type of identified sensing data source.

[0126] The specific configuration of the first failure probability output module 30 will be described in detail below. The first failure probability output module 30 may further include: parsing the task requirements of the first braking task; analyzing multiple failure sensitivity indicators corresponding to the multiple sensor data sources based on the task requirements; calculating multiple initial failure probabilities corresponding to the multiple sensor data sources using a Bayesian network; and adjusting the multiple initial failure probabilities with the multiple failure sensitivity indicators to obtain multiple first failure probabilities.

[0127] The specific configuration of the first failure probability output module 30 will be described in detail below. The first failure probability output module 30 further includes: collecting task type samples; and defining a task sensitivity matrix M based on the task type samples, where M... ij The sensitivity of the i-th task sample to the j-th sensor data source is included. The size of M is m×n, where m is the number of samples of the task type and n is the number of sensor data sources. A failure sensitivity analysis model is established based on the task sensitivity matrix M and the vector that identifies the magnitude of failure sensitivity. The task requirements are input into the failure sensitivity analysis model to identify multiple failure sensitivity vectors, which are output as multiple failure sensitivity indicators.

[0128] The specific configuration of the first failure probability output module 30 will be described in detail below. The first failure probability output module 30 further includes: performing scale standardization processing on historical data from the multiple sensor data sources to obtain multiple standardized sensor data samples; establishing a failure probability calculation model based on the multiple standardized sensor data samples, the failure probability calculation model including multiple prediction functions and multiple Bernoulli probability distributions, the multiple prediction functions and multiple Bernoulli probability distributions having a mapping relationship; inputting the multiple sensor data sources into the failure probability calculation model, using the prediction outputs of the multiple prediction functions as intermediate variables, inputting them into the multiple Bernoulli probability distributions for failure probability identification, and obtaining multiple initial failure probabilities.

[0129] The specific configuration of the first failure probability output module 30 will be described in detail below. The first failure probability output module 30 further includes: the failure probability calculation model includes:

[0130] P(F i (t)=1)=∫P(F i (t)=1|D i (t))P(D i (t))dD i (t);

[0131] F i (t) represents the failure probability of the i-th sensor data source at time t, D i (t) represents the sensing data of the i-th sensing data source at time t, which follows a multivariate normal distribution. Let be the mean vector of a multivariate normal distribution. Let F be the covariance matrix of a multivariate normal distribution. i (t) Based on sensor data source D i The probability distribution of (t) is used to calculate the i-th sensor data source F under failure conditions. i The probability (t) = 1 is expressed by the integral representing the probability of all possible sensor data sources D. i (t) performs a weighted summation;

[0132] P(D i (t) is the prediction function, P(D) i (t))=∫∑ k∈S P(D i (t)|x k (t))P(x k This is used to calculate the sensor data source D at time t. i The probability distribution of (t), where S is the number of factors characterizing wheel control failure, P(x k ) represents the failure probability under the k-th factor.

[0133] The specific configuration of the compensation sensor data output module 60 will be described in detail below. The compensation sensor data output module 60 further includes the following: the expression for the enhanced compensation model is as follows:

[0134]

[0135] in, To compensate, α i To compensate for the weight parameters of the compensation strategy corresponding to the i-th data source in the first type of identification sensing data source based on time t, α i Optimization is achieved through gradient descent; D i(t) represents the real-time sensing data of the i-th data source in the first type of identification sensing data source based on time t. The predicted sensing data for the i-th data source in the first type of identification sensing data source is based on time t.

[0136] The specific configuration of the optimization control parameter acquisition module 70 will be described in detail below. The optimization control parameter acquisition module 70 may further include: establishing a braking performance evaluation model, the input of which is connected to the output of the enhancement compensation model, the braking performance evaluation model including braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate; evaluating the compensation strategy for each round of optimization based on the braking performance evaluation model to obtain the performance enhancement rate; and, based on the performance enhancement rate and a preset performance enhancement rate, adjusting the weight parameter α in the enhancement compensation model using gradient descent. i To find the best option.

[0137] Figure 3 The control rate test curve for the traditional braking control method is shown. Figure 4 The control rate test curve obtained after adopting this method shows that the braking performance has achieved greater reliability and stability, improved the anti-interference ability of braking control, and realized intelligent monitoring and control of the braking system.

[0138] The braking performance enhancement system based on the wheel control model provided in this embodiment of the invention can execute the braking performance enhancement method based on the wheel control model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0139] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0140] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A method for brake performance enhancement based on a wheel control model, characterized by The specific steps are as follows: Connect the wheel control model and determine multiple sensor data sources for the wheel control model; Acquire runway monitoring data, and perform failure impact analysis based on the runway monitoring data and multiple sensor data sources to obtain multiple failure impact indicators; Receive the first braking task, parse the task requirements of the first braking task, analyze the failure sensitivity indicators corresponding to the multiple sensor data sources according to the task requirements, calculate the initial failure probability of the multiple sensor data sources using a Bayesian network, adjust the initial failure probability in combination with the failure sensitivity indicators, and output multiple first failure probabilities. Multiple first failure probabilities are adjusted using the aforementioned multiple failure impact indicators to output multiple second failure probabilities; Based on the plurality of second failure probabilities, the sensor data sources are divided into a first type of identification sensor data source and a second type of identification sensor data source, wherein the failure probability of the first type of identification sensor data source is greater than or equal to a preset threshold. The first type of identification sensor data source is input into the enhancement compensation model for real-time data compensation, and the compensated sensor data is output. Based on the compensation sensing data and the real-time sensing data corresponding to the second type of identification sensing data source, optimized control parameters are generated to achieve dynamic enhancement of braking performance.

2. The method of claim 1, wherein: The analysis of the failure sensitivity indicators includes: Define the task sensitivity matrix M m×n , of which M ij This represents the sensitivity of the i-th task sample to the j-th sensor data source, where i = 1, 2, 3, ..., m, j = 1, 2, 3, ..., n, m is the number of task type samples, and n is the number of sensor data sources; Based on the task sensitivity matrix and failure sensitivity analysis model, a failure sensitivity vector is generated as multiple failure sensitivity indicators to adjust the initial failure probability.

3. The method of claim 2, wherein: The construction of the failure sensitivity analysis model includes: Perform a dot product or weighted sum operation on the task sensitivity matrix and the failure sensitivity vector to generate a total sensitivity index, which is the total sensitivity of each task to the failure of the sensor data source. Based on the total sensitivity index, the weight distribution of the initial failure probability output by the Bayesian network is dynamically adjusted.

4. The method of claim 3, wherein: The calculation of the initial failure probability of the multiple sensor data sources using Bayesian networks includes: The historical data from the multiple sensor data sources are subjected to scale standardization to generate standardized sensor data samples. A failure probability calculation model is established based on the standardized sensor data samples. The failure probability calculation model includes a prediction function with a mapping relationship and a Bernoulli probability distribution. The multiple sensor data sources are input into the failure probability calculation model. The prediction outputs of multiple prediction functions are used as intermediate variables and input into multiple Bernoulli probability distributions to identify failure probabilities and obtain multiple initial failure probabilities.

5. The method of claim 4, wherein: The failure probability calculation model is expressed as follows: P(F i (t)=1)=∫P(F i (t)=1|D i (t))P(D i (t))dD i (t); In the formula, F i (t) represents the failure probability of the i-th sensor data source at time t, D i (t) represents the sensing data of the i-th sensing data source at time t, which follows a multivariate normal distribution. Let be the mean vector of a multivariate normal distribution. Let F be the covariance matrix of a multivariate normal distribution. i (t) Based on sensor data source D i The probability distribution of (t) is used to calculate the i-th sensor data source F under failure conditions. i The probability (t) = 1 is expressed by the integral representing the probability of all possible sensor data sources D. i (t) performs a weighted summation; P(D i (t) is the prediction function, P(D) i (t))=∫∑ k∈S P(D i (t)|x k (t))P(x k This is used to calculate the sensor data source D at time t. i The probability distribution of (t), where S is the number of factors characterizing wheel control failure, P(x k ) represents the failure probability under the k-th factor.

6. The method of claim 5, wherein: The expression for the enhanced compensation model is as follows: in, To compensate, α i To compensate for the weight parameters of the compensation strategy corresponding to the i-th data source in the first type of identification sensing data source based on time t, α i Optimization is achieved through gradient descent; D i (t) represents the real-time sensing data of the i-th data source in the first type of identification sensing data source based on time t. The predicted sensing data for the i-th data source in the first type of identification sensing data source is based on time t.

7. The method of claim 6, wherein the method further comprises: The weight parameter a i The optimization includes: A braking performance evaluation model is established, which includes braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate. Connect the input of the braking performance evaluation model to the output of the enhancement compensation model; The compensation strategy for each round of optimization is evaluated based on the braking performance evaluation model to obtain the performance enhancement rate. According to the performance enhancement rate and a preset performance enhancement rate, iteratively optimize a by a gradient descent algorithm until the preset performance enhancement rate is met. i , until the preset performance enhancement rate is met.

8. A brake performance enhancement system based on a wheelie control model for implementing the brake performance enhancement method based on a wheelie control model according to any one of claims 1 to 7; characterized in that include: The sensor data source determination module is used to connect to the wheel control model and determine multiple sensor data sources; The failure impact index acquisition module is used to acquire runway monitoring data, analyze the failure impact relationship between runway monitoring data and sensor data sources, and generate failure impact indexes. The first failure probability output module is used to calculate and adjust the initial failure probability based on the Bayesian network and the task sensitivity matrix, and output the first failure probability. The second failure probability output module is used to dynamically correct the first failure probability by combining the failure impact index and output the second failure probability. The identification sensor data source acquisition module is used to classify the first type and the second type of sensor data sources according to the second failure probability. The compensation sensor data output module is used to compensate the first type of sensor data in real time through the enhancement compensation model; The optimized control parameter acquisition module is used to fuse compensation data and second-type data to generate optimized control parameters.

9. The braking performance enhancement system based on the wheel control model according to claim 8, characterized in that: The first failure probability output module includes: The standardization processing unit is used to standardize the historical data from the sensor data source. The failure probability calculation unit establishes a mapping model between the prediction function and the Bernoulli probability distribution based on standardized data, and outputs the initial failure probability.

10. The brake performance enhancement system based on a model of a vehicle wheel as defined in claim 8, wherein: The optimized control parameter acquisition module quantifies the performance enhancement rate through a braking performance evaluation model, which includes braking distance enhancement rate, response time enhancement rate, and braking stability enhancement rate, and iteratively optimizes the control parameters through a gradient descent algorithm.

Images