Fault diagnosis methods, devices, equipment, and media for underwater vehicles based on physical data hybrid modeling strategy.

Abstract

This application relates to the field of fault diagnosis technology and discloses a fault diagnosis method for underwater vehicles based on a physical data hybrid modeling strategy. It employs at least a rudder effect prediction physical model and a rudder effect prediction neural network, including: inputting dynamic parameters into the rudder effect prediction physical model to predict rudder effect, obtaining a theoretically driven predicted rudder effect; inputting dynamic parameters into the rudder effect prediction neural network to predict rudder effect, obtaining a data-driven predicted rudder effect; performing fault diagnosis based on the theoretically driven predicted rudder effect, theoretical prediction confidence level, data-driven predicted rudder effect, data prediction confidence level, and dynamic parameters, obtaining a fault diagnosis result; and making fault emergency decisions based on the fault diagnosis result and the underwater vehicle's operating state parameters, outputting a fault-tolerant control strategy. Its beneficial effect is that by using multi-strategy rudder effect prediction and performing fault diagnosis based on the prediction results, it can adapt to different operating conditions, significantly improving the accuracy and reliability of the fault diagnosis results.

Description

Technical Field

[0001] This application relates to the field of fault diagnosis technology, and in particular to a method, apparatus, equipment and medium for fault diagnosis of underwater vehicles based on a physical data hybrid modeling strategy. Background Technology

[0002] An Autonomous Underwater Vehicle (AUV) is a robotic device capable of performing tasks underwater, playing a crucial role in various fields such as marine scientific research, resource exploration, and military reconnaissance. In real-world scenarios, the complex and ever-changing underwater environment presents significant challenges to AUVs, while navigation modes also influence mission execution. Under dynamic operating conditions, AUVs may experience mechanical failures during mission execution, not only affecting mission performance but also potentially leading to loss of control and serious damage. Therefore, fault diagnosis is essential for the normal navigation and operation of AUVs, but the accuracy of fault diagnosis methods in related technologies still needs improvement. Summary of the Invention

[0003] This application provides a method, apparatus, equipment, and medium for underwater vehicle fault diagnosis based on a physical data hybrid modeling strategy. By performing multi-strategy rudder effect prediction on the underwater vehicle and performing fault diagnosis and emergency decision-making based on the prediction results of different prediction strategies, the adaptability of the prediction results to different operating conditions is effectively improved, and the accuracy and reliability of the fault diagnosis results are significantly improved.

[0004] To achieve the above objectives, the main technical solutions adopted in this application include:

[0005] In a first aspect, embodiments of this application provide a fault diagnosis method for underwater vehicles based on a physical data hybrid modeling strategy, wherein the physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network; the method includes:

[0006] The dynamic parameters of the underwater vehicle are input into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle, thereby obtaining the theoretically driven predicted rudder effect of the underwater vehicle; the dynamic parameters are input into the rudder effect prediction neural network to predict the rudder effect of the underwater vehicle, thereby obtaining the data-driven predicted rudder effect of the underwater vehicle; wherein, the theoretically driven predicted rudder effect corresponds to a theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to a data prediction confidence level.

[0007] The underwater vehicle is diagnosed based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and the dynamic parameters, and the fault diagnosis results of the underwater vehicle are obtained.

[0008] Based on the fault diagnosis results and the operating status parameters of the underwater vehicle, emergency fault decisions are made, and a fault-tolerant control strategy for handling the corresponding fault is output.

[0009] The underwater vehicle fault diagnosis method proposed in this application, based on a physical data hybrid modeling strategy, predicts the underwater vehicle's rudder effect using multiple strategies based on its dynamic parameters. This is achieved through a rudder effect prediction physical model and a rudder effect prediction neural network, respectively. The theoretically driven predicted rudder effect corresponding to the physical model and the data-driven predicted rudder effect corresponding to the neural network are obtained. Based on this, fault diagnosis is performed on the underwater vehicle according to the predicted rudder effects and their corresponding prediction confidence levels for different prediction strategies. Finally, a fault-tolerant control strategy for handling the corresponding fault is output based on the fault diagnosis results. Compared with related technologies, this application utilizes a physical data hybrid modeling strategy that includes both a rudder effect prediction physical model and a rudder effect prediction neural network to predict the underwater vehicle's rudder effect using multiple strategies. It then performs comprehensive fault diagnosis based on each predicted rudder effect and its corresponding confidence level. This allows the fault diagnosis results to adapt to the actual operating conditions of the underwater vehicle, comprehensively considering the impact of actual operating conditions on the underwater vehicle's behavior and effectively improving the accuracy and reliability of the fault diagnosis results.

[0010] Optionally, the dynamic parameters include the heading speed and rudder angle of the underwater vehicle; the step of inputting the dynamic parameters of the underwater vehicle into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle and obtain the theoretical driving prediction rudder effect of the underwater vehicle includes:

[0011] Based on the rudder angle of each rudder on the underwater vehicle and its corresponding hydrodynamic coefficient, the torque exerted by each rudder on the underwater vehicle is calculated to obtain the static rudder effect of the underwater vehicle.

[0012] The static rudder effect is nonlinearly corrected based on the heading speed to obtain the theoretically predicted rudder effect.

[0013] Optionally, the dynamic parameters include the underwater vehicle's speed and angular velocity; the step of inputting the dynamic parameters into the rudder effect prediction neural network to predict the rudder effect of the underwater vehicle and obtain the data-driven predicted rudder effect of the underwater vehicle includes:

[0014] Acceleration is calculated at set time intervals based on the sailing speed to obtain the sailing acceleration; angular acceleration is calculated at set time intervals based on the sailing angular velocity to obtain the sailing angular acceleration.

[0015] The navigation acceleration and the navigation angular acceleration are input into the rudder effect prediction neural network, and the navigation acceleration and the navigation angular acceleration are linearly combined using basis functions with different weights to obtain parameter feature mapping;

[0016] The rudder effect is predicted based on the parameter feature mapping to obtain the data-driven predicted rudder effect.

[0017] Optionally, the step of performing fault diagnosis on the underwater vehicle based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and the dynamic parameters to obtain the fault diagnosis result of the underwater vehicle includes:

[0018] The theoretically predicted rudder effect is corrected based on the theoretical prediction confidence level to obtain the adaptive theoretically predicted rudder effect; the data-driven predicted rudder effect is corrected based on the data prediction confidence level to obtain the adaptive data-driven predicted rudder effect.

[0019] Based on the adaptive theory-predicted rudder effect, the adaptive data-predicted rudder effect, and the dynamic parameters, a fault diagnosis network is used to diagnose the underwater vehicle and obtain the fault diagnosis results of the underwater vehicle.

[0020] Optionally, the theoretical prediction confidence level and the data prediction confidence level are obtained in the following ways:

[0021] The theoretically driven predicted rudder effect and the data-driven predicted rudder effect are input into the fault diagnosis network for confidence prediction, so as to obtain the initial theoretical confidence level corresponding to the theoretically driven predicted rudder effect and the initial data confidence level corresponding to the data-driven predicted rudder effect.

[0022] The theoretically driven predicted rudder effect is weighted and adjusted based on the initial theoretical confidence level to obtain the iterative theoretical predicted rudder effect; the data-driven predicted rudder effect is weighted and adjusted based on the initial data confidence level to obtain the iterative data predicted rudder effect.

[0023] The iterative theoretical prediction of rudder effect and the iterative data prediction of rudder effect are input into the fault diagnosis network for confidence prediction, so as to update the initial theoretical confidence and the initial data confidence; the above confidence prediction and weight allocation adjustment steps are repeated until the preset stopping condition is reached;

[0024] The initial theoretical confidence level is used as the theoretical prediction confidence level, and the initial data confidence level is used as the data prediction confidence level.

[0025] Optionally, the fault diagnosis network can be trained in the following manner:

[0026] The initial network parameters of the fault diagnosis network are searched using the ant colony algorithm to obtain the optimized network parameters of the fault diagnosis network.

[0027] The fault diagnosis network, configured according to the optimized network parameters, is trained based on the training dataset to update the optimized network parameters and obtain the target network parameters; the fault diagnosis network is then configured with parameters based on the target network parameters to complete the training.

[0028] Optionally, the fault diagnosis result includes the fault type and fault severity, and the operating status parameters include the remaining energy status of the underwater vehicle, the navigation mission type, and the environmental disturbance level; the step of making fault emergency decisions based on the fault diagnosis result and the operating status parameters of the underwater vehicle, and outputting a fault-tolerant control strategy for handling the corresponding fault, includes:

[0029] Fuzzy variables are obtained by fuzzy partitioning the fault type, the fault degree, the remaining energy status, the navigation mission type, and the environmental disturbance level;

[0030] The fault-tolerant control strategy is obtained by performing fuzzy inference on the fuzzy variables based on pre-defined fuzzy rules.

[0031] Secondly, embodiments of this application provide an underwater vehicle fault diagnosis device based on a physical data hybrid modeling strategy, wherein the physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network; the device includes:

[0032] A multi-strategy rudder effect prediction module is used to input the dynamic parameters of the underwater vehicle into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle and obtain the theoretically driven predicted rudder effect of the underwater vehicle; and to input the dynamic parameters into the rudder effect prediction neural network to predict the rudder effect of the underwater vehicle and obtain the data-driven predicted rudder effect of the underwater vehicle; wherein, the theoretically driven predicted rudder effect corresponds to a theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to a data prediction confidence level.

[0033] The vehicle fault diagnosis module is used to diagnose the underwater vehicle based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and the dynamic parameters, and to obtain the fault diagnosis result of the underwater vehicle.

[0034] The vehicle emergency decision module is used to make emergency decisions based on the fault diagnosis results and the operating status parameters of the underwater vehicle, and output a fault-tolerant control strategy for handling the corresponding fault.

[0035] Thirdly, embodiments of this application provide a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in any of the above embodiments.

[0036] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the method described in any one of the above embodiments.

[0037] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which are used to cause a computer to perform the method described in any of the above embodiments. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0039] Figure 1 A step diagram illustrating the underwater vehicle fault diagnosis method based on a physical data hybrid modeling strategy provided in this application embodiment;

[0040] Figure 2 This is a diagram illustrating the steps involved in obtaining the theoretically driven predicted rudder effect in the embodiments of this application.

[0041] Figure 3 This is a diagram illustrating the steps involved in obtaining data-driven prediction of rudder effectiveness in the embodiments of this application.

[0042] Figure 4 This is a flowchart illustrating the steps for fault diagnosis of an underwater vehicle in an embodiment of this application.

[0043] Figure 5 This is a flowchart illustrating the steps involved in obtaining the theoretical prediction confidence level and the data prediction confidence level in the embodiments of this application.

[0044] Figure 6 This is a diagram illustrating the steps of training the fault diagnosis network in an embodiment of this application.

[0045] Figure 7 This is a flowchart of the training of the fault diagnosis network in the embodiments of this application;

[0046] Figure 8 This is a flowchart illustrating the steps of outputting the fault-tolerant control strategy in an embodiment of this application.

[0047] Figure 9 This is an overall framework diagram of the underwater vehicle fault diagnosis method based on a physical data hybrid modeling strategy provided in the embodiments of this application.

[0048] Figure 10 A block diagram of an underwater vehicle fault diagnosis device based on a physical data hybrid modeling strategy provided in this application embodiment;

[0049] Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] An Autonomous Underwater Vehicle (AUV) is a robotic device capable of performing tasks underwater, playing a crucial role in various fields such as marine scientific research, resource exploration, and military reconnaissance. In real-world scenarios, the complex and ever-changing underwater environment presents significant challenges to AUVs, while navigation modes also influence mission execution. Under dynamic operating conditions, AUVs may experience mechanical failures during mission execution, not only affecting mission performance but also potentially leading to loss of control and serious damage. Therefore, fault diagnosis is essential for the normal navigation and operation of AUVs.

[0052] In related technologies, methods for fault diagnosis of underwater vehicles typically rely on a single model. For example, they may calculate the underwater vehicle's operating state solely based on a physical model, or utilize neural networks to learn the underwater vehicle's operating modes to predict potential faults. While physical models, employing fundamental physical principles, offer strong interpretability, they have limitations when dealing with complex and dynamically changing environmental conditions. Neural networks, on the other hand, learn from large-scale data, effectively handling fault prediction under complex conditions, but the interpretability of predictions obtained from neural networks is poor. Therefore, fault diagnosis methods based on a single model are limited by the dynamic changes in the underwater vehicle's operating conditions and often fail to provide accurate and reliable fault prediction results.

[0053] To address the aforementioned issues, this application provides a fault diagnosis method for underwater vehicles based on a physical data hybrid modeling strategy. This method employs at least a rudder effect prediction physical model and a rudder effect prediction neural network, comprising: inputting dynamic parameters into the rudder effect prediction physical model to predict rudder effect, obtaining a theoretically driven predicted rudder effect; inputting dynamic parameters into the rudder effect prediction neural network to predict rudder effect, obtaining a data-driven predicted rudder effect; the theoretically driven predicted rudder effect corresponds to a theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to a data prediction confidence level; fault diagnosis is performed based on the theoretically driven predicted rudder effect, theoretical prediction confidence level, data-driven predicted rudder effect, data prediction confidence level, and dynamic parameters to obtain a fault diagnosis result; and fault emergency decision-making is performed based on the fault diagnosis result and the underwater vehicle's operating state parameters, outputting a fault-tolerant control strategy.

[0054] The underwater vehicle fault diagnosis method based on the physical data hybrid modeling strategy provided in this application, based on the dynamic parameters of the underwater vehicle, performs multi-strategy rudder effect prediction on the underwater vehicle through a rudder effect prediction physical model and a rudder effect prediction neural network, respectively, to obtain the theoretically driven predicted rudder effect corresponding to the rudder effect prediction physical model and the data-driven predicted rudder effect corresponding to the rudder effect prediction neural network; on this basis, the underwater vehicle is diagnosed according to the predicted rudder effect corresponding to different prediction strategies and their corresponding prediction confidence, and the fault-tolerant control strategy for handling the corresponding fault is output based on the fault diagnosis results.

[0055] Compared with related technologies, this application utilizes a physical data hybrid modeling strategy that includes a rudder effect prediction physical model and a rudder effect prediction neural network to perform multi-strategy rudder effect prediction for underwater vehicles. Based on each predicted rudder effect and its corresponding confidence level, a comprehensive fault diagnosis is performed, enabling the fault diagnosis results to adapt to the actual operating conditions of the underwater vehicle. This comprehensively considers the impact of actual operating conditions on the behavior of the underwater vehicle, effectively improving the accuracy and reliability of the fault diagnosis results.

[0056] The underwater vehicle fault diagnosis method based on a physical data hybrid modeling strategy provided in this specification can be applied to the fault diagnosis of underwater vehicles. Underwater vehicles may include autonomous underwater vehicles (XAUVs), remotely operated underwater vehicles (ROVs), hybrid underwater vehicles, or underwater robots, etc. The physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network. It is understood that, after adaptive modifications, this method can also be used for fault monitoring of working equipment other than underwater vehicles, or for rudder effect prediction using strategies other than physical models and neural networks.

[0057] According to an embodiment of this application, an embodiment of an underwater vehicle fault diagnosis method based on a physical data hybrid modeling strategy is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0058] This embodiment provides a fault diagnosis method for underwater vehicles based on a physical data hybrid modeling strategy, which can be used to diagnose faults in the aforementioned underwater vehicles. (Refer to...) Figure 1 As shown, the physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network; the method includes:

[0059] S100. Input the dynamic parameters of the underwater vehicle into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle and obtain the theoretically driven predicted rudder effect of the underwater vehicle; input the dynamic parameters into the rudder effect prediction neural network to predict the rudder effect of the underwater vehicle and obtain the data-driven predicted rudder effect of the underwater vehicle; wherein, the theoretically driven predicted rudder effect corresponds to the theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to the data prediction confidence level.

[0060] S200. Based on theoretically driven predicted rudder effect, theoretical prediction confidence level, data-driven predicted rudder effect, data prediction confidence level, and dynamic parameters, fault diagnosis of the underwater vehicle is performed to obtain the fault diagnosis results of the underwater vehicle.

[0061] S300. Make emergency fault decisions based on fault diagnosis results and the operating status parameters of the underwater vehicle, and output fault-tolerant control strategies for handling the corresponding faults.

[0062] The rudder effect prediction physical model can be used to deduce and simulate the actual physical behavior of underwater vehicles during operation based on their dynamic equations and related physical principles, thereby obtaining the theoretically driven predicted rudder effect. The physical principles utilized in the rudder effect prediction physical model include, but are not limited to, rudder surface mechanics models, hydrodynamic models, and physical transport equations. It is understandable that the rudder effect prediction physical model simulates based on known physical laws, and under stable conditions, it can accurately characterize the working process of underwater vehicles, exhibiting high interpretability.

[0063] The rudder effect prediction neural network can learn from the large amount of operating condition data generated by underwater vehicles during operation, extract the complex nonlinear relationships between the operating condition data, and predict the trend of rudder effect changes of underwater vehicles, thus obtaining a data-driven prediction of rudder effect. For example, the neural network type used in the rudder effect prediction neural network can be a radial basis function neural network, a multilayer perceptron, a support vector machine, or K-nearest neighbor regression. It can be understood that the rudder effect prediction neural network can automatically model based on the operating condition data of underwater vehicles, and through an adaptive mechanism, adjust in real time according to changes in operating conditions, achieving flexible prediction of the rudder effect of underwater vehicles.

[0064] Specifically, sensors installed within the underwater vehicle collect dynamic parameters generated during its operation. After obtaining these dynamic parameters, a physical data hybrid modeling strategy is employed to establish multiple prediction models for multi-strategy rudder effect prediction. It should be noted that the physical data hybrid modeling strategy includes at least a rudder effect prediction physical model and a rudder effect prediction neural network. The rudder effect prediction physical model predicts the rudder effect of the underwater vehicle based on physical principles, while the rudder effect prediction neural network predicts the rudder effect based on data relationships, thus obtaining predicted rudder effect from different perspectives and effectively improving the comprehensiveness of the prediction results. In some embodiments, the number of rudder effect prediction physical models or rudder effect prediction neural networks can be one or more.

[0065] Furthermore, the rudder effect prediction physical model simulates based on known physical laws, and the resulting theory-driven prediction of rudder effect is more accurate in stable environments. Meanwhile, the rudder effect prediction neural network learns complex relationships between data to predict rudder effect, and the resulting data-driven prediction can adjust in real time according to changes in operating conditions, making it more suitable for complex and variable environments. Since different prediction strategies are based on different principles and use different operating conditions, after obtaining multi-strategy rudder effect prediction results for underwater vehicles, the prediction confidence of different prediction results under the current operating conditions of the underwater vehicle is analyzed to obtain the theoretical prediction confidence corresponding to theory-driven rudder effect prediction and the data prediction confidence corresponding to data-driven rudder effect prediction. These prediction confidences are used to fuse the multi-strategy rudder effect prediction results, enabling the rudder effect prediction results based on different prediction strategies to complement each other, significantly improving the accuracy of rudder effect prediction results, and thus improving the accuracy and reliability of fault diagnosis results.

[0066] Furthermore, after obtaining the theoretically driven predicted rudder effect, theoretical prediction confidence level, data-driven predicted rudder effect, and data prediction confidence level, the results of the theoretically driven predicted rudder effect and data-driven predicted rudder effect are fused based on the theoretical prediction confidence level and data prediction confidence level. This allows for fault diagnosis of the underwater vehicle based on the fused results and its dynamic parameters, yielding fault diagnosis results. For example, the fault diagnosis results may include fault types, including but not limited to rudder jamming, multi-faceted damage, and centering faults, each corresponding to a specific fault severity. Based on this, fault emergency decisions are made according to the obtained fault diagnosis results and the underwater vehicle's operating state parameters, outputting a fault-tolerant control strategy to handle the corresponding fault types and ensure the normal operation and navigation of the underwater vehicle. It is understood that the underwater vehicle's operating state parameters can be directly obtained from sensors within the underwater vehicle, representing its current state.

[0067] The underwater vehicle fault diagnosis method based on the physical data hybrid modeling strategy provided in this embodiment predicts the underwater vehicle's rudder effect using a multi-strategy rudder effect prediction physical model and a rudder effect prediction neural network based on the underwater vehicle's dynamic parameters. This yields a theoretically driven predicted rudder effect corresponding to the rudder effect prediction physical model and a data-driven predicted rudder effect corresponding to the rudder effect prediction neural network. Based on this, the underwater vehicle is diagnosed for faults according to the predicted rudder effect corresponding to different prediction strategies and their corresponding prediction confidence levels. Finally, a fault-tolerant control strategy for handling the corresponding faults is output based on the fault diagnosis results.

[0068] Compared with related technologies, this application utilizes a physical data hybrid modeling strategy that includes a rudder effect prediction physical model and a rudder effect prediction neural network to perform multi-strategy rudder effect prediction for underwater vehicles. Based on each predicted rudder effect and its corresponding confidence level, a comprehensive fault diagnosis is performed, enabling the fault diagnosis results to adapt to the actual operating conditions of the underwater vehicle. This comprehensively considers the impact of actual operating conditions on the behavior of the underwater vehicle, effectively improving the accuracy and reliability of the fault diagnosis results.

[0069] Reference Figure 2 As shown in one embodiment of this application, the dynamic parameters include the heading speed and rudder angle of the underwater vehicle; the dynamic parameters of the underwater vehicle are input into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle, and the theoretical driving prediction rudder effect of the underwater vehicle is obtained, including:

[0070] S110. Based on the rudder angle of each rudder on the underwater vehicle and its corresponding hydrodynamic coefficient, calculate the torque exerted by each rudder on the underwater vehicle to obtain the static rudder effect of the underwater vehicle.

[0071] S120. The static rudder effect is nonlinearly corrected based on the bow velocity to obtain the theoretically predicted rudder effect.

[0072] The heading speed of an underwater vehicle can be considered its forward speed relative to the underwater environment. The rudder angle can be the relative angle between any rudder surface and the vehicle's longitudinal axis. Underwater vehicles can be equipped with different types of rudders, including but not limited to stern rudders and bow rudders. Stern rudders can be X-shaped, and one or more of each type can be installed. Different types of rudders play different roles during navigation, exerting torque on the underwater vehicle in different ways; therefore, each type of rudder has its own hydrodynamic coefficient. The heading speed and rudder angle determine the underwater vehicle's trajectory and attitude, jointly affecting its maneuverability and stability.

[0073] Specifically, each rudder on an underwater vehicle interacts with the water flow to provide torque to the underwater vehicle. The static rudder effect of the underwater vehicle is obtained by calculating the torque exerted by each rudder on the underwater vehicle based on the rudder angle and its corresponding hydrodynamic coefficient.

[0074] Furthermore, when an underwater vehicle is traveling at high speed, the nonlinear effects of water flow on the underwater vehicle become more significant, thus affecting the relationship between rudder effectiveness and heading speed. Therefore, after obtaining the static rudder effectiveness of the underwater vehicle, it is necessary to perform nonlinear correction based on the heading speed of the underwater vehicle to predict the rudder effectiveness of the underwater vehicle at the current heading speed, thereby obtaining the theoretically driven predicted rudder effectiveness.

[0075] For example, the theoretically predicted rudder effect of an underwater vehicle includes the theoretically predicted pitch rudder effect and the theoretically predicted steering rudder effect; correspondingly, the hydrodynamic coefficients include the pitch dynamics coefficient and the steering dynamics coefficient. When the underwater vehicle is equipped with four sets of X-shaped stern and bow rudders, the theoretically predicted pitch rudder effect and the theoretically predicted steering rudder effect can be expressed by the following formula:

[0076]

[0077]

[0078] in, This indicates a theoretical prediction of pitch rudder effectiveness; This indicates a theoretical prediction of steering rudder effectiveness; , , , and These are the pitch dynamics coefficients of each rudder on the underwater vehicle; , , and These are the steering dynamics coefficients of each rudder on the underwater vehicle; The forward speed of the underwater vehicle; , , and These are the rudder angles, respectively. This refers to the rudder angle at the bow.

[0079] Reference Figure 3 As shown, in one embodiment of this application, the dynamic parameters include the underwater vehicle's speed and angular velocity; the dynamic parameters are input into a rudder effect prediction neural network to predict the rudder effect of the underwater vehicle, obtaining a data-driven predicted rudder effect, including:

[0080] S130. Calculate the acceleration based on the sailing speed at set time intervals to obtain the sailing acceleration; calculate the acceleration based on the sailing angular velocity at set time intervals to obtain the sailing angular acceleration.

[0081] S140. Input the navigation acceleration and angular acceleration into the rudder effect prediction neural network, and use the basis functions with different weights to linearly combine the navigation acceleration and angular acceleration to obtain the parameter feature map.

[0082] S150. Based on the parameter feature mapping, predict the rudder effect to obtain the data-driven predicted rudder effect.

[0083] The underwater vehicle's speed includes three degrees of freedom, with the speeds in different directions being independent of each other. The underwater vehicle's angular velocity includes three degrees of freedom, with the rotational speeds in different directions being independent of each other.

[0084] Specifically, the six-degree-of-freedom acceleration of the underwater vehicle is calculated and used as input to the rudder effect prediction neural network. The six-degree-of-freedom acceleration of the underwater vehicle is expressed by the following formula:

[0085]

[0086] in, , and These represent the accelerations of the underwater vehicle in different directions. , and These represent the speeds of the underwater vehicle in different directions. , and These represent the angular accelerations of the underwater vehicle in different directions. , and These represent the angular velocities of the underwater vehicle in different directions. The current moment for predicting rudder effectiveness; To set the time interval, it is usually set to 0.1s.

[0087] Furthermore, the rudder effect prediction neural network can be a pre-constructed and trained neural network. For example, the rudder effect prediction neural network can be a radial basis function (RBF) neural network, constructing a three-layer neural network consisting of an input layer, a hidden layer, and an output layer. In the rudder effect prediction neural network, an adaptive algorithm of the radial basis function is used for real-time adjustment to gradually approach the nonlinear function that serves as the prediction result. The rudder effect prediction neural network receives the six-degree-of-freedom acceleration of the underwater vehicle as input, and uses basis functions with corresponding weights in the hidden layer to linearly combine the navigation acceleration and the navigation angular acceleration to obtain a parameter feature map, expressed by the following formula:

[0088]

[0089] in, Represents the parameter feature mapping; The input to the rudder effect prediction neural network; This is the weight matrix; For Gaussian functions; This represents the approximate error.

[0090] Furthermore, the output layer receives the parameter feature map and adjusts the weights of each Gaussian function in the parameter feature map to predict the rudder effect of the underwater vehicle based on the parameter feature map, thus obtaining the data-driven predicted rudder effect. The data-driven predicted rudder effect includes data-predicted pitch rudder effect and data-predicted steering rudder effect, where the data-predicted pitch rudder effect is represented as... Data predicts steering effectiveness as .

[0091] Reference Figure 4 As shown, in one embodiment of this application, fault diagnosis of an underwater vehicle is performed based on theoretically driven predicted rudder effectiveness, theoretical prediction confidence level, data-driven predicted rudder effectiveness, data prediction confidence level, and dynamic parameters to obtain fault diagnosis results for the underwater vehicle, including:

[0092] S210. The theoretically driven predicted rudder effect is corrected based on the confidence level of the theoretical prediction to obtain the adaptive theoretically predicted rudder effect; the data-driven predicted rudder effect is corrected based on the confidence level of the data prediction to obtain the adaptive data-driven predicted rudder effect.

[0093] S220. Based on the adaptive theory to predict the rudder effect, the adaptive data to predict the rudder effect and dynamic parameters, the fault diagnosis network is used to perform fault diagnosis on the underwater vehicle, and the fault diagnosis results of the underwater vehicle are obtained.

[0094] Specifically, the prediction strategies in the physical data hybrid modeling strategy are based on different principles, and different prediction strategies are applicable to different operating conditions of underwater vehicles. Among them, the rudder effect prediction physical model is based on known physical laws for simulation, and the theoretically driven prediction of rudder effect is more accurate in stable environments. The rudder effect prediction neural network, on the other hand, learns the complex relationships between data to predict rudder effect. The data-driven prediction of rudder effect can be adjusted in real time according to changes in operating conditions, making it more suitable for complex and variable environments. Therefore, when performing fault diagnosis on underwater vehicles based on multi-strategy rudder effect prediction results, it is necessary to fuse the multi-strategy rudder effect prediction results so that the fault diagnosis results can adapt to the actual operating conditions of the underwater vehicle, thereby comprehensively considering the impact of actual operating conditions on the behavior of the underwater vehicle and effectively improving the accuracy and reliability of the fault diagnosis results.

[0095] Furthermore, theoretical prediction confidence and data prediction confidence are used to correct the confidence of their respective rudder effect prediction results, thereby fusing the multi-strategy rudder effect prediction results. This allows rudder effect prediction results based on different prediction strategies to complement each other, significantly improving the accuracy of rudder effect prediction results, and consequently improving the accuracy and reliability of fault diagnosis results. It can be understood that adaptive theoretical prediction rudder effect includes adaptive theoretical pitch rudder effect and adaptive theoretical steering rudder effect, and adaptive data prediction rudder effect includes adaptive data pitch rudder effect and adaptive data steering rudder effect. The confidence correction is expressed by the following formula:

[0096]

[0097] in, This indicates adaptive data pitch control effect. This indicates adaptive data steering rudder effect; This indicates the pitch control effect of adaptive theory. This indicates the adaptive theory steering effect; To predict confidence levels for the data, The confidence level for theoretical predictions.

[0098] Furthermore, after confidence correction, the above data, including the obtained adaptive theoretical pitch rudder effect, adaptive theoretical steering rudder effect, adaptive data pitch rudder effect, and adaptive data steering rudder effect, as well as the dynamic parameters of the underwater vehicle, are input into the fault diagnosis network to diagnose the faults of the underwater vehicle and obtain the fault diagnosis results.

[0099] It should be noted that the fault diagnosis network can be used to perform feature learning and data classification on input data to achieve accurate identification of various fault types in the underwater vehicle's steering system. For example, the fault diagnosis network can be a backpropagation neural network, with an overall structure of a three-layer feedforward network, including an input layer, hidden layers, and an output layer. The input layer receives input data including adaptive data pitch rudder effect, adaptive theoretical steering rudder effect, adaptive data pitch rudder effect, adaptive data steering rudder effect, and various dynamic parameters of the underwater vehicle. The number of neurons in the input layer is the same as the number of input data points and corresponds to these input data points.

[0100] The output layer outputs fault diagnosis results and can use One-Hot encoding to represent three types of faults, further distinguishing the specific fault severity under each fault type through fault category sub-labels. In some embodiments, fault types may include rudder jamming faults, multi-faceted damage faults, and off-center faults, where rudder jamming faults correspond to 7 specific fault severity levels, including corresponding rudder jamming angles of -30°, -20°, -10°, 0°, 10°, 20°, and 30°. Multi-faceted damage faults correspond to 3 specific fault severity levels, including corresponding damage ratios of 30%, 50%, and 80%. Off-center faults correspond to 2 specific fault severity levels, including corresponding offsets of -10 and 10. The number of neurons in the output layer is determined based on the fault type and the number of corresponding specific fault severity levels. For example, when the total number of specific fault severity levels is 12, the number of neurons in the output layer can be 60.

[0101] Hidden layers are used for feature learning, and the number of neurons in them can be determined by an empirical formula, expressed as follows:

[0102]

[0103] in, This indicates the number of neurons in the hidden layer; The number of neurons in the input layer; This represents the number of neurons in the output layer. The value of the adjustment factor is between [1, 10]. In some embodiments, with 15 neurons in the input layer and 60 neurons in the output layer, the optimal value of the adjustment factor is determined to be 35 through comprehensive experiments.

[0104] Reference Figure 5 As shown, in one embodiment of this application, the theoretical prediction confidence level and the data prediction confidence level are obtained in the following manner:

[0105] S160. Input the theoretically driven predicted rudder effect and the data-driven predicted rudder effect into the fault diagnosis network to predict the confidence level, and obtain the initial theoretical confidence level corresponding to the theoretically driven predicted rudder effect and the initial data confidence level corresponding to the data-driven predicted rudder effect.

[0106] S170. Adjust the weight allocation of the theoretically driven predicted rudder effect according to the initial theoretical confidence level to obtain the iterative theoretical predicted rudder effect; adjust the weight allocation of the data-driven predicted rudder effect according to the initial data confidence level to obtain the iterative data predicted rudder effect.

[0107] S180. Input the iterative theoretical prediction of rudder effect and the iterative data prediction of rudder effect into the fault diagnosis network to predict the confidence level, so as to update the initial theoretical confidence level and the initial data confidence level; repeat the above steps of confidence level prediction and weight allocation adjustment until the preset stopping condition is reached.

[0108] S190. Use the initial theoretical confidence level as the theoretical prediction confidence level, and use the initial data confidence level as the data prediction confidence level.

[0109] Specifically, the fault diagnosis network outputs not only the fault diagnosis results but also the prediction confidence scores corresponding to different rudder effect prediction results, representing the degree of adaptation of different prediction strategies to the current operating conditions of the underwater vehicle. Therefore, after obtaining the multi-strategy rudder effect prediction results for the underwater vehicle, the fault diagnosis network predicts the confidence scores of different prediction results based on the current operating conditions of the underwater vehicle, obtaining the initial theoretical confidence scores corresponding to the theoretically driven predicted rudder effect and the initial data confidence scores corresponding to the data-driven predicted rudder effect. It is understandable that the initial theoretical confidence scores and initial data confidence scores are obtained by the fault diagnosis network through preliminary learning based on the theoretically driven and data-driven predicted rudder effects, and there are still errors between them and the actual confidence scores of each predicted rudder effect, which cannot fully represent the degree of adaptation of different prediction strategies to the current operating conditions.

[0110] Furthermore, the weight allocation of the theoretically driven predicted rudder effect is adjusted based on the initial theoretical confidence level to obtain the iterative theoretical predicted rudder effect. The weight allocation of the data-driven predicted rudder effect is adjusted based on the initial data confidence level to obtain the iterative data predicted rudder effect. The iterative theoretical predicted rudder effect and the iterative data predicted rudder effect are then input into the fault diagnosis network for confidence prediction to update the initial theoretical confidence level and the initial data confidence level. It should be noted that in the above process, the fault diagnosis network further learns from the initial theoretical confidence level and the initial data confidence level to predict different rudder effects, thereby adjusting the operational adaptability of the rudder effect prediction results based on the initial confidence level of the previous iteration, improving the accuracy of the adjusted initial confidence level.

[0111] Furthermore, the steps of confidence prediction and weight allocation adjustment described above are repeated until a preset stopping condition is met. The initial theoretical confidence level of the current iteration is used as the theoretical predicted confidence level, and the initial data confidence level of the current iteration is used as the data predicted confidence level. It is understood that the preset stopping condition includes, but is not limited to, an iteration count condition, a confidence convergence condition, and a confidence threshold condition. The iteration count condition can be that the number of iterations described above reaches the predicted iteration count. The confidence convergence condition can be that the rate of change of the initial confidence level does not exceed a preset rate of change threshold within a certain number of iterations. The confidence threshold condition can be that the initial confidence level value is less than a preset value threshold.

[0112] Reference Figure 6 As shown, in one embodiment of this application, the fault diagnosis network is trained in the following manner:

[0113] S230. Use the ant colony algorithm to search for the initial network parameters of the fault diagnosis network and obtain the optimized network parameters of the fault diagnosis network.

[0114] S240. Train the fault diagnosis network model based on the training dataset and the optimized network parameters to update the optimized network parameters and obtain the target network parameters; set the parameters of the fault diagnosis network according to the target network parameters to complete the training.

[0115] Specifically, the Ant Colony Optimization (ACO) algorithm is a heuristic random search algorithm that solves combinatorial optimization problems by simulating the pathfinding process of ants in nature. In this embodiment, the ACO algorithm is used to search for and optimize the initial network parameters of the fault diagnosis network. These initial network parameters include the weights and thresholds of the fault diagnosis network. (Refer to...) Figure 7 As shown, during the search process, the ant colony and the pheromone distribution in the search space are first initialized, and the corresponding search space is constructed. Next, in the current iteration, the ants are moved, and the pheromone distribution in the search space is updated based on the movement. Then, based on the ant's movement, the ant colony state transition probability, pheromone concentration, and fitness of the current iteration are calculated sequentially. If the fitness meets the preset fitness condition, the search result at this point is output as the optimized network parameters for the fault diagnosis network; otherwise, the next iteration continues.

[0116] In some embodiments, to further improve the search efficiency of the ant colony algorithm, the path selection probability of the ants can be adjusted to affect the algorithm's efficiency and accuracy. The path selection probability can be adjusted dynamically and adaptively, and its form is expressed by the following formula:

[0117]

[0118] in, and This represents the parameters that control the weights of pheromones and heuristic factors; and These are the parameters from the previous iteration. This is the current iteration step size. This represents the maximum iteration step size. By adjusting the path selection probability using the above formula, the ants can explore the solution space more flexibly during the search process, effectively improving the search efficiency of network parameters.

[0119] Furthermore, the fault diagnosis network is trained based on the training dataset according to the optimized network parameters. The errors and their corresponding weights during the model training process are used to update the threshold of the fault diagnosis network. The model training is completed when the training stopping condition is met, and the trained fault diagnosis network is obtained.

[0120] It should be noted that by using the ant colony algorithm to search for the initial network parameters of the fault diagnosis network, the convergence speed of the initial network parameters is improved, thus enabling the optimization of network parameters to be determined more quickly. Furthermore, optimizing the network parameters can effectively improve the generalization ability and robustness of the fault diagnosis network in the fault diagnosis process, thereby improving the accuracy and reliability of the fault diagnosis results.

[0121] Reference Figure 8 As shown in one embodiment of this application, the fault diagnosis result includes the fault type and fault severity, and the operating status parameters include the underwater vehicle's remaining energy status, navigation mission type, and environmental disturbance level; based on the fault diagnosis result and the underwater vehicle's operating status parameters, a fault emergency decision is made, and a fault-tolerant control strategy for handling the corresponding fault is output, including:

[0122] S310. Fuzzy classification is performed on the fault type, fault degree, remaining energy status, navigation mission type and environmental disturbance level to obtain fuzzy variables.

[0123] S320. Based on pre-defined fuzzy rules, perform fuzzy inference on fuzzy variables to obtain a fault-tolerant control strategy.

[0124] Reference Figure 9 As shown, in the overall framework of this application, the decision layer receives the fault diagnosis results output by the diagnostic layer and uses fuzzy decision units to generate corresponding fault-tolerant control strategies based on the underwater vehicle's operating state parameters to cope with emergencies during operation. The fault diagnosis results output by the diagnostic layer may include fault type and fault severity, where the fault confidence level is represented as... The fault type is represented as The degree of failure is expressed as Operational status parameters may include the underwater vehicle's remaining energy state, mission type, and environmental disturbance level, where the remaining energy state is expressed as... The navigation mission type is represented as The environmental disturbance level is expressed as The fault-tolerant control strategy is expressed as: .

[0125] Specifically, fuzzy classifications are used for fault confidence, fault type, fault severity, remaining energy status, navigation mission type, and environmental disturbance level. The universe of discourse for fault confidence is [0,1]. Fault type, including rudder jamming, multi-faceted damage, off-center damage, and no fault, has a universe of discourse of [0,3]. Fault severity includes high-risk and low-risk faults, with a universe of discourse of [0,1]. Rudder jamming faults with a jamming angle of 30° or -30° and multi-faceted damage faults with a damage ratio of 80% are considered high-risk faults; other fault types are considered low-risk. The universe of discourse for remaining energy status is [0,1]. Navigation mission type includes search and operation phases, with a universe of discourse of [0,1]. The universe of discourse for environmental disturbance level is [0,4].

[0126] The above fuzzy inputs are described in linguistic form, where fault confidence is categorized as {S,L}, representing low and high fault confidence, respectively. Fault type is categorized as {NB,NS,PS,PB}, representing no fault, broken fault, centering fault, and stuck rudder fault, respectively. Fault severity is categorized as {S,L}, representing low and high dangerous faults, respectively. Remaining energy status is categorized as {S,L}, representing low and high remaining energy, respectively. Navigation mission type is categorized as {S,L}, representing the search phase and the operational phase, respectively. Environmental disturbance level is categorized as {S,L}, representing low and high environmental disturbance levels, respectively.

[0127] Similarly, the fault-tolerant control strategies corresponding to the fuzzy variables include continue navigation, optimal allocation, fault reconstruction, control remapping, and emergency surfacing, with a universe of discourse of [0,4]. The output of the fuzzy decision unit is described in linguistic form, and the fault-tolerant control strategies are divided into {NB, NS, Z, PS, PB}, representing continue navigation, optimal allocation, control remapping, fault reconstruction, and emergency surfacing, respectively.

[0128] Furthermore, the fuzzy rules for the fuzzy decision unit are shown in Table 1. After determining the fault-tolerant control strategy, the centroid method is used for defuzzification to output control parameters for controlling the underwater vehicle. It should be noted that the Mamdani inference algorithm is used as the fuzzy inference algorithm in the fuzzy weight allocator.

[0129] Table 1. Fuzzy rule table for fuzzy decision-making units

[0130]

[0131] Accordingly, please refer to Figure 10This application provides an underwater vehicle fault diagnosis device based on a physical data hybrid modeling strategy. The physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network. The device includes:

[0132] The multi-strategy rudder effect prediction module 1010 is used to input the dynamic parameters of the underwater vehicle into the rudder effect prediction physical model to predict the rudder effect of the underwater vehicle and obtain the theoretically driven predicted rudder effect; and to input the dynamic parameters into the rudder effect prediction neural network to predict the rudder effect of the underwater vehicle and obtain the data-driven predicted rudder effect; wherein, the theoretically driven predicted rudder effect corresponds to the theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to the data prediction confidence level.

[0133] The vehicle fault diagnosis module 1020 is used to diagnose underwater vehicles based on theoretically driven predicted rudder effectiveness, theoretical prediction confidence, data-driven predicted rudder effectiveness, data prediction confidence, and dynamic parameters, and obtain the fault diagnosis results of the underwater vehicles.

[0134] The underwater vehicle emergency decision module 1030 is used to make emergency decisions based on fault diagnosis results and the operating status parameters of the underwater vehicle, and output fault-tolerant control strategies for handling the corresponding faults.

[0135] In some optional implementations, the multi-strategy rudder effect prediction module 1010 includes:

[0136] The static rudder effect calculation unit is used to calculate the torque exerted by each rudder on the underwater vehicle based on the rudder angle and its corresponding hydrodynamic coefficient, thereby obtaining the static rudder effect of the underwater vehicle.

[0137] The bow speed correction unit is used to perform nonlinear correction on the static rudder effect based on the bow speed to obtain the theoretically driven predicted rudder effect.

[0138] In some optional implementations, the multi-strategy rudder effect prediction module 1010 includes:

[0139] The acceleration unit is used to calculate acceleration at set time intervals based on the sailing speed to obtain the sailing acceleration; and to calculate acceleration at set time intervals based on the sailing angular velocity to obtain the sailing angular acceleration.

[0140] The parameter mapping unit is used to input the navigation acceleration and angular acceleration into the rudder effect prediction neural network, and to linearly combine the navigation acceleration and angular acceleration using basis functions with different weights to obtain the parameter feature map.

[0141] The rudder effect prediction unit is used to predict rudder effect based on parameter feature mapping, and obtain data-driven predicted rudder effect.

[0142] In some optional implementations, the aircraft fault diagnosis module 1020 includes:

[0143] The confidence correction unit is used to correct the theoretically driven predicted rudder effect based on the theoretical prediction confidence to obtain the adaptive theoretically predicted rudder effect; and to correct the data-driven predicted rudder effect based on the data prediction confidence to obtain the adaptive data-driven predicted rudder effect.

[0144] The fault diagnosis unit is used to predict rudder effect and dynamic parameters based on adaptive theory and adaptive data, and to perform fault diagnosis on the underwater vehicle using a fault diagnosis network to obtain the fault diagnosis results of the underwater vehicle.

[0145] In some optional implementations, the multi-strategy rudder effect prediction module 1010 includes:

[0146] The initial confidence level acquisition unit is used to input the theoretically driven predicted rudder effect and the data-driven predicted rudder effect into the fault diagnosis network for confidence level prediction, and to obtain the initial theoretical confidence level corresponding to the theoretically driven predicted rudder effect and the initial data confidence level corresponding to the data-driven predicted rudder effect.

[0147] The confidence iteration unit is used to adjust the weight allocation of the theoretically driven predicted rudder effect based on the initial theoretical confidence level to obtain the iterative theoretical predicted rudder effect; and to adjust the weight allocation of the data-driven predicted rudder effect based on the initial data confidence level to obtain the iterative data predicted rudder effect.

[0148] The iterative unit is used to input the iterative theoretical prediction of rudder effect and the iterative data prediction of rudder effect into the fault diagnosis network for confidence prediction, so as to update the initial theoretical confidence and the initial data confidence; repeat the above confidence prediction and weight allocation adjustment steps until the preset stopping condition is reached.

[0149] The prediction confidence acquisition unit is used to take the initial theoretical confidence as the theoretical prediction confidence and the initial data confidence as the data prediction confidence.

[0150] In some optional implementations, the aircraft fault diagnosis module 1020 includes:

[0151] The parameter optimization unit is used to search for the initial network parameters of the fault diagnosis network using the ant colony algorithm, and obtain the optimized network parameters of the fault diagnosis network.

[0152] The model training unit is used to train the fault diagnosis network based on the training dataset and the optimized network parameters, so as to update the optimized network parameters and obtain the target network parameters; and to set the parameters of the fault diagnosis network according to the target network parameters to complete the training.

[0153] In some alternative implementations, the aircraft emergency decision module 1030 includes:

[0154] Fuzzy partitioning units are used to fuzzily partition fault type, fault degree, remaining energy status, navigation mission type and environmental disturbance level to obtain fuzzy variables.

[0155] The fuzzy decision unit is used to perform fuzzy inference on fuzzy variables based on pre-defined fuzzy rules to obtain a fault-tolerant control strategy.

[0156] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0157] In this embodiment, the underwater vehicle fault diagnosis device based on the physical data hybrid modeling strategy is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0158] Please see Figure 11 , Figure 11 This is a schematic diagram of a computer device according to an embodiment of this application. As shown in the figure, the computer device includes one or more processors 10, a memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components communicate with each other using different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some alternative embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 11 Take a processor 10 as an example.

[0159] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0160] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0161] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0162] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0163] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0164] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0165] This application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method of any embodiment of this application.

[0166] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and all such modifications and variations fall within the scope defined by the appended claims.

[0167] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0168] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0169] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0170] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0171] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0172] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0173] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0174] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0175] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0176] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A fault diagnosis method for underwater vehicles based on a physical data hybrid modeling strategy, characterized in that, The physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network; the method includes: The rudder angle and heading speed of the underwater vehicle are input into the rudder effect prediction physical model. Based on the rudder angle and corresponding hydrodynamic coefficient of each rudder on the underwater vehicle, the torque exerted by each rudder on the underwater vehicle is calculated to obtain the static rudder effect of the underwater vehicle. The static rudder effect is nonlinearly corrected based on the heading speed to obtain the theoretically driven predicted rudder effect. The acceleration of the underwater vehicle is calculated at set time intervals based on its speed to obtain the speed acceleration. The acceleration of the underwater vehicle is calculated at set time intervals based on its angular velocity to obtain the angular acceleration. The speed acceleration and the angular acceleration are input into the rudder effect prediction neural network. The speed acceleration and the angular acceleration are linearly combined using basis functions with different weights to obtain a parameter feature map. Rudder effect prediction is performed based on the parameter feature map to obtain the data-driven predicted rudder effect. The theoretically driven predicted rudder effect corresponds to a theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to a data prediction confidence level. The underwater vehicle is diagnosed based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and the dynamic parameters, and the fault diagnosis results of the underwater vehicle are obtained. Based on the fault diagnosis results and the operating status parameters of the underwater vehicle, emergency fault decisions are made, and a fault-tolerant control strategy for handling the corresponding fault is output.

2. The method according to claim 1, characterized in that, The fault diagnosis of the underwater vehicle based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and dynamic parameters, to obtain the fault diagnosis result of the underwater vehicle, includes: The theoretically predicted rudder effect is corrected based on the theoretical prediction confidence level to obtain the adaptive theoretically predicted rudder effect; the data-driven predicted rudder effect is corrected based on the data prediction confidence level to obtain the adaptive data-driven predicted rudder effect. Based on the adaptive theory-predicted rudder effect, the adaptive data-predicted rudder effect, and the dynamic parameters, a fault diagnosis network is used to diagnose the underwater vehicle and obtain the fault diagnosis results of the underwater vehicle.

3. The method according to claim 2, characterized in that, The theoretical prediction confidence level and the data prediction confidence level are obtained in the following ways: The theoretically driven predicted rudder effect and the data-driven predicted rudder effect are input into the fault diagnosis network for confidence prediction, so as to obtain the initial theoretical confidence level corresponding to the theoretically driven predicted rudder effect and the initial data confidence level corresponding to the data-driven predicted rudder effect. The theoretically driven predicted rudder effect is weighted and adjusted according to the initial theoretical confidence level to obtain the iterative theoretical predicted rudder effect. The weight allocation of the data-driven predicted rudder effect is adjusted according to the initial data confidence level to obtain the iterative data predicted rudder effect. The iterative theoretical prediction of rudder effect and the iterative data prediction of rudder effect are input into the fault diagnosis network for confidence prediction, so as to update the initial theoretical confidence and the initial data confidence; the above confidence prediction and weight allocation adjustment steps are repeated until the preset stopping condition is reached; The initial theoretical confidence level is used as the theoretical prediction confidence level, and the initial data confidence level is used as the data prediction confidence level.

4. The method according to claim 2, characterized in that, The fault diagnosis network is trained in the following manner: The initial network parameters of the fault diagnosis network are searched using the ant colony algorithm to obtain the optimized network parameters of the fault diagnosis network. The fault diagnosis network, configured according to the optimized network parameters, is trained based on the training dataset to update the optimized network parameters and obtain the target network parameters; the fault diagnosis network is then configured with parameters based on the target network parameters to complete the training.

5. The method according to any one of claims 1 to 4, characterized in that, The fault diagnosis results include the fault type and fault severity, and the operating status parameters include the remaining energy status of the underwater vehicle, the navigation mission type, and the environmental disturbance level. The step of making emergency fault decisions based on the fault diagnosis results and the operating status parameters of the underwater vehicle, and outputting a fault-tolerant control strategy for handling the corresponding fault, includes: Fuzzy variables are obtained by fuzzy partitioning the fault type, the fault degree, the remaining energy status, the navigation mission type, and the environmental disturbance level; The fault-tolerant control strategy is obtained by performing fuzzy inference on the fuzzy variables based on pre-defined fuzzy rules.

6. A fault diagnosis device for underwater vehicles based on a physical data hybrid modeling strategy, characterized in that, The physical data hybrid modeling strategy employs at least a rudder effect prediction physical model and a rudder effect prediction neural network; the device includes: A multi-strategy rudder effect prediction module is used to input the rudder angle and heading speed of an underwater vehicle into the rudder effect prediction physical model. Based on the rudder angle of each rudder and its corresponding hydrodynamic coefficient, the module calculates the torque exerted by each rudder on the underwater vehicle to obtain the static rudder effect. The static rudder effect is then nonlinearly corrected based on the heading speed to obtain the theoretically driven predicted rudder effect. Acceleration is calculated at set time intervals based on the underwater vehicle's speed to obtain the navigation acceleration. Acceleration is also calculated at set time intervals based on the underwater vehicle's angular velocity to obtain the navigation angular acceleration. The navigation acceleration and navigation angular acceleration are input into the rudder effect prediction neural network, and a linear combination of the navigation acceleration and navigation angular acceleration is performed using basis functions with different weights to obtain a parameter feature map. Rudder effect prediction is then performed based on the parameter feature map to obtain the data-driven predicted rudder effect. The theoretically driven predicted rudder effect corresponds to a theoretical prediction confidence level, and the data-driven predicted rudder effect corresponds to a data prediction confidence level. The underwater vehicle fault diagnosis module is used to diagnose faults in the underwater vehicle based on the theoretically driven predicted rudder effect, the theoretical prediction confidence level, the data-driven predicted rudder effect, the data prediction confidence level, and dynamic parameters, and to obtain the fault diagnosis results of the underwater vehicle. The vehicle emergency decision module is used to make emergency decisions based on the fault diagnosis results and the operating status parameters of the underwater vehicle, and output a fault-tolerant control strategy for handling the corresponding fault.

7. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the method of any one of claims 1 to 5.

Images