A fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering

Abstract

This invention discloses a fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering, belonging to the field of underwater vehicle target detection. The steps are as follows: the underactuated underwater vehicle travels in a straight line, and the turning radius is determined based on turning model data; the fused navigation information is predicted based on discrete Kalman filtering, and the navigation error is compared with the average turning radius; position information and side-scan sonar feature variables are fused, and preliminary fault-tolerant processing is performed; a sliding window is used to statistically analyze and record the current navigation information of the underwater vehicle in real time. Compared with traditional methods that only compare current position information in real time, this invention creatively combines sonar feature information and uses Kalman filtering to predict the underwater vehicle's position information. It combines a sliding window and the range of the turning radius to perform fault-tolerant processing and judgment of the position information, providing reliable motion control for accurate and complete underwater vehicle target detection.

Description

Technical Field

[0001] This invention belongs to the field of underwater target detection for underwater vehicles, and specifically relates to a fault-tolerant processing method for underwater target detection based on Kalman filtering. Background Technology

[0002] As we all know, the Earth's oceans are vast and contain a variety of abundant resources, such as oil and natural gas. The rational exploitation and use of marine resources, as well as the protection of the marine ecosystem, are crucial issues that need to be addressed. However, the marine environment is extremely complex and unpredictable, making underwater exploration via manned submersibles extremely dangerous. Against this backdrop, underwater vehicles have emerged as a solution.

[0003] As the saying goes, "A skilled craftsman must first sharpen his tools," unmanned underwater vehicles (UUVs), as powerful tools for exploring the ocean, shoulder the important responsibility of being pioneers. Research related to UUV exploration has become an indispensable part of this process.

[0004] Underwater target search generally involves four basic processes: target search, target approach, target reconnaissance and information acquisition, and return to the mother ship. Currently, the widely used method for target reconnaissance and information acquisition is to analyze acquired sonar information. When approaching the target, the course is determined by comparing the current position with the target point in the mission plan. After the mission concludes, the host computer acquires the subsequent trajectory and reads side-scan data to obtain more complete target information. However, in actual target detection, while calculating the subsequent trajectory based on the current moment is simple, it suffers from low accuracy and the risk of target loss. Often, even when the target is found, the vehicle's course deviates, resulting in incomplete information acquisition. Currently, there is no fault-tolerant systematic method for motion control in underwater target detection. Summary of the Invention

[0005] The technical problem to be solved:

[0006] To overcome the shortcomings of existing technologies, this invention provides a fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering. In underwater detection missions, the ideal turning radius is first calculated using experimental datasets. Then, combining sonar feature variables and the vehicle's own position information, position information is predicted based on Kalman filtering. Finally, a sliding window is used to perform fault-tolerant processing on the fused information, thereby improving detection accuracy in real time. This method can be used for target detection of underwater vehicles with multiple targets, serving as a foundation for the field of detection and perception.

[0007] The technical solution of this invention is: a fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering, used for underwater vehicles to sense and detect targets, characterized in that:

[0008] Step 1: The underactuated underwater vehicle travels in a straight line, and the turning radius is determined based on the turning model data;

[0009] Step 2: Based on discrete Kalman filtering, predict the fused navigation information and compare the navigation error with the average turning radius;

[0010] Step 3: Fusion of location information and side-scan sonar characteristic variables, and preliminary fault tolerance processing;

[0011] Step 4: The sliding window will collect and record the current navigation information of the vehicle in real time.

[0012] A further technical solution of the present invention is as follows: In step one, the underwater vehicle's speed v and position x are collected and statistically analyzed; the turning radius is ρ, and n sampling points are taken, namely x1, x2, ..., xn. Based on geometric relationships, the turning radius ρ = (ρ1...ρn) can be estimated from any sampling point. n ).

[0013] A further technical solution of the present invention is: average turning radius Calculation:

[0014]

[0015] in, h is the ideal turning radius, representing the vertical distance between the underwater vehicle at position x0 and position x, mapped to the center of the turning circle from position x0. Δx = x - x0, where Δx is the deviation of the underwater vehicle from position x0 to position x under ideal conditions, and θ is the yaw angle of the vehicle from position x0 to position x; ρ1 + ρ2 + ... + ρ n It is the sum of the turning radii of the aircraft at positions x1, x2, ... xn.

[0016] A further technical solution of the present invention is as follows: In step two, it is assumed that all disturbances are Gaussian distributed, such that the state variable X(k)=[x(k),y(k),θ(k)] represents the current information of the underwater vehicle, where x(k) is the current position, y(k) is the current speed, and θ(k) is the yaw angle; the position of the underwater vehicle is used as the observation variable, thereby establishing the Kalman measurement equation as follows:

[0017] y(k)=H(k)X(k)+V(k) (2)

[0018] Where H(k) is the observation matrix and V(k) is the observation noise;

[0019] Perform the next step of estimating the system's position, velocity, and yaw angle according to equation (3):

[0020]

[0021] Where k represents a discrete time node; This represents the predicted state variables for time k at time k-1, including heading angle, position, and velocity information; This is the predicted state equation of the system at time k-1 to time k, and it is also the state transition matrix of the system;

[0022]

[0023] Estimated system state update:

[0024]

[0025] Where K(k) represents the Kalman gain matrix at time k, Y(k) represents the system measurement sequence from time 0 to time k, and H(k) represents the system measurement matrix at time k. Represents the predicted state variables of the system at time k;

[0026] Variance of prediction estimation error:

[0027]

[0028] in, represents the system state estimation error covariance at time k-1, which is set to 0 at time 0; Q is the system noise covariance matrix. It is the system's state transition matrix. yes The transpose of the matrix; based on prior experience, Q is set as:

[0029]

[0030] Therefore, the filter gain equation is:

[0031]

[0032] Where K(K|K-1) is the predicted Kalman gain matrix for time k at time k-1; H represents the covariance of the system state estimation error at time k-1; T Let H represent the transpose of the measurement matrix at the current time; H is the measurement matrix at the current time; and R is the system measurement noise covariance matrix.

[0033]

[0034] During target detection, the position at the next moment is predicted using Kalman filtering; and the error Δx is calculated based on the currently known target latitude and longitude [x1, y1] and the latitude and longitude [x2, y2] obtained by the underwater vehicle's own navigation system, and compared with the average turning radius in (1). For comparison, if the longitude or latitude is smaller than the turning radius... Then proceed to step three; the error calculation is as follows:

[0035]

[0036] Where x_i represents the longitude at time i, y_i represents the latitude at time i, and ||x1-x2|| represents the norm of x1-x2.

[0037] A further technical solution of the present invention is as follows: In step three, the direction is determined based on the position information obtained in step two and the side-scan feature value g; the side-scan feature value g is collected in real time by the side-scan sonar data board and sent to the main control, representing the percentage of effective side-scan data.

[0038] A further technical solution of the present invention is: the current longitude error is less than the average turning radius. And since it is a negative number, according to the changes in geographical longitude and latitude, it can be divided into changes in longitude and latitude;

[0039] If the direction is north-south, the turning course depends on the difference in longitude; in this case, if the error is less than 0 and its absolute value is less than the average turning radius in step one. If the target is located to the right of the vehicle, then a left turn should be made; conversely, a right turn should be made if the target is not to the right. Similarly, if the current latitude error is less than 0 and less than the average turning radius in step one, then the target should be located to the left. If the target being detected is located to the left of the underwater vehicle, then it should turn right; otherwise, it should turn left.

[0040] A further technical solution of the present invention is: in step four, the "window" error step size is set to 10; if 10 consecutive sets of errors occur that are greater than the average turning radius... In such cases, the current direction can be changed to flexibly respond to situations where incorrect judgments are made.

[0041] A further technical solution of the present invention is: in step four, the sliding window refers to the method in industrial control algorithms where a fixed-length array is used, the initial interval of which is [1, 10], and the error step size is 10.

[0042] Specifically, when the underwater vehicle begins to detect, and the navigation information is within the interval in real time, the predicted value in step two is compared with the value in step two, and the window array is assigned a value. When the array variable has been assigned 10 values, that is, the window has been filled, it is then determined whether five consecutive position errors are less than or greater than the turning radius. If five consecutive position errors are less than or greater than the turning radius, the underwater vehicle moves in the opposite direction; otherwise, it maintains its original direction.

[0043] The window variable is reset upon completion, and the position error is reassigned until the mission ends or an error occurs and fault tolerance is handled. Step four is executed repeatedly until the detection mission is completed and the underwater vehicle returns to the mother ship.

[0044] An underwater vehicle employing a fault-tolerant target detection method based on Kalman filtering is characterized in that: the vehicle is equipped with a Doppler velocimeter, inertial navigation, GPS, and side-scan sonar; the current position is obtained by combining the Doppler velocimeter with inertial navigation and GPS; and the side-scan sonar is used to detect the specific size and shape information of the target.

[0045] Beneficial effects

[0046] The beneficial effects of this invention are as follows: Compared with the traditional method of simply comparing current position information in real time, this invention creatively combines the characteristic information of sonar and uses Kalman filtering to predict the position information of underwater vehicles. It also combines the range of sliding window and turning radius to perform fault-tolerant processing and judgment of position information, providing reliable motion control for accurate and complete underwater vehicle target detection.

[0047] Before the mission begins, this invention calculates the ideal turning radius using a test dataset, which serves as a standardized threshold for subsequent fault-tolerant processing. Then, it combines percentage features of the sonar with its own trajectory to estimate position information, predicts the position information based on Kalman filtering, and finally uses a sliding window to perform fault-tolerant processing on the fused information, thereby improving detection accuracy in real time. Figure 5 and 6 As shown. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the process framework of the present invention;

[0049] Figure 2 This is a schematic diagram of the turning radius of the AUV of the present invention;

[0050] Figure 3 This is a flowchart of the sliding window of the present invention;

[0051] Figure 4 This is a schematic diagram of the fault-tolerant processing of the present invention;

[0052] Figure 5A schematic diagram of left turn measurement and position prediction for the fault-tolerant turning position (left turn when specified in the positive direction) during target detection for underwater vehicles;

[0053] Compared to the position information before prediction, the predicted information after Kalman filtering has a smaller error under noise interference. After fault-tolerant processing, the predicted information continuously reduces its error with the turning radius, eventually circling the target.

[0054] Figure 6 A schematic diagram of right turn position measurement and position prediction during target detection fault tolerance processing of an underwater vehicle. After fault tolerance processing, the vehicle turns right to reduce the error between itself and the target, and the turning radius error is smaller than the heading error before fault tolerance. Detailed Implementation

[0055] 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.

[0056] Reference Figures 1 to 4 This embodiment provides a fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering.

[0057] In terms of hardware, this example, for underwater target detection missions, equips the underwater vehicle with a primary detection device: side-scan sonar. The side-scan sonar is connected to a separate data acquisition board for easy reading and analysis of the detection data. Furthermore, in accordance with the methodology described, the data acquisition board transmits the valid sonar information bits to the main control board in real time via a serial port. Additionally, the navigation system integrates Doppler and inertial navigation to achieve underwater position information output.

[0058] In terms of software, for the position information of underwater dead reckoning of underwater vehicles, Kalman filtering is used to remove system noise and measurement noise, and the true position information is predicted. Combined with side scan feature variables, the position information is judged and optimized by using a sliding window.

[0059] A fault-tolerant processing method for underwater vehicles based on Kalman filtering for underwater target detection, as follows: Figure 1 As shown. Includes the following steps:

[0060] Step 1: The underactuated underwater vehicle performs straight-line navigation, and the data acquisition and turning radius determination of the turning model are carried out.

[0061] Due to factors such as ocean currents, the turning radius of an underactuated underwater vehicle (UUV) is variable. Different ocean currents result in varying resistance for the vehicle at the same speed. During a turn, the turning radius changes due to this resistance. Therefore, data collection is necessary before each mission to determine the turning radius.

[0062] In step one, the collected and statistically analyzed data includes the vehicle's speed v, current position x, and heading angle per unit time. To calculate the turning radius ρ, n sampling points are taken, based on geometric relationships such as... Figure 2 As shown, h is the length of the perpendicular segment from the radius of the circle where the underwater vehicle turns to the radius of the turning circle under the ideal turning model. It can be seen that, at any sampling point, the turning radius ρ = (ρ1...ρ2) / (ρ3) can be estimated. n The calculation is as follows: Assuming the spacecraft's current positions are x and x1 at times t and t_1, then at... Figure 2 The calculation of Δx is as follows:

[0063] Δx=x-x0 (1)

[0064] Since the heading angle changes during a turn, in order to describe the turning radius more accurately, we should try to find a relationship between the radius ρ and the heading angle θ as follows:

[0065]

[0066]

[0067] Further, using trigonometric relationships, we can obtain the formula for ρ with respect to Δx:

[0068]

[0069] It is evident that, under ideal conditions, the turning radius of a vehicle is related to its heading angle and the positional error it travels in the radial direction per unit time. The effect of water resistance reduces the positional error in the radial direction per unit time, thus decreasing the turning radius.

[0070] To universally represent the turning radius of an underwater vehicle, the mean-value theorem states that the approximately optimal turning radius is:

[0071]

[0072] Step 2: Predict underwater vehicle navigation information based on discrete Kalman filtering;

[0073] In step two, it is assumed that all disturbances are Gaussian distributed, such that the state variable X(k) = [x(k), y(k), θ(k)] represents the current information of the underwater vehicle, where x(k) is the current position, y(k) is the current velocity, and θ(k) is the yaw angle. To facilitate effective detection, the position of the underwater vehicle is used as the observation variable, thus establishing the Kalman measurement equation as follows:

[0074] y(k)=H(k)X(k)+V(k) (6)

[0075] Where H(k) is the observation matrix and V(k) is the observation noise.

[0076] Perform the next step of system state estimation according to equation (7):

[0077]

[0078] in The system state equations are given, assuming the underwater vehicle moves at a constant velocity, meaning the system state transition matrix is:

[0079]

[0080] Based on Kalman filtering theory, a recursive prediction scheme is applied to the augmented observation equation, continuously estimating the system state update through time updates and measurement updates:

[0081]

[0082] Variance of prediction estimation error:

[0083]

[0084] Where Q is the system noise covariance matrix, Q is set as:

[0085]

[0086] Therefore, the filter gain equation is:

[0087]

[0088] Where R is the system measurement noise covariance matrix:

[0089]

[0090] During target detection, the position at the next moment is predicted using Kalman filtering. The error Δx is calculated based on the approximate latitude and longitude [x1, y1] of the known target and the latitude and longitude [x2, y2] obtained from the underwater vehicle's own navigation system (inertial navigation + Doppler velocimeter), and compared with the turning radius ρ in (5). If the latitude or longitude is less than the turning radius, proceed to step three. If it is greater than the turning radius, the vehicle autonomously searches the path planned by the mission until it is less than the turning radius.

[0091] The error is calculated as follows:

[0092]

[0093] Where x_i represents longitude, y_i represents latitude, and ||x1-x2|| represents the norm of x1-x2.

[0094] Step 3: Perform preliminary fault tolerance processing by fusing the location information and side-scan sonar feature variables.

[0095] Location information is provided by the underwater vehicle's navigation system. Since GPS information cannot be obtained underwater, a combined navigation system is needed to perform dead reckoning to obtain location information. For underwater vehicles equipped with side-scan sonar, the XTF data acquired by the side scan needs to be uploaded to shore via WiFi after each mission. Furthermore, the XTF format results in a large data storage volume each time, requiring a separate data board for processing. The fault tolerance processing in this method is mainly calculated on the main control board. Traditional methods of judging location information alone cannot effectively utilize the effective side-scan data and are prone to errors in dead reckoning during target detection. Therefore, this method combines traditional single location information with the percentage of effective side-scan data for preliminary judgment.

[0096] The position information obtained in step two, combined with the side-scan feature value g (which is collected in real time by the side-scan sonar data board and sent to the main control, representing the percentage of effective side-scan data), is used to determine the direction.

[0097] If the current side-scan feature value g > 50%, then the current longitude error is less than ρ and is negative. Based on the changes in geographical latitude and longitude, this can be specifically divided into changes in longitude and latitude. According to geographical latitude and longitude, in the Northern Hemisphere, latitude gradually decreases from north to south, while longitude gradually increases from west to east. If the direction is north-south, the turning course depends on the difference in longitude. If the error is less than 0 (error definition: vehicle latitude and longitude information minus target latitude and longitude information) and its absolute value is less than the turning radius in step one, then the target is located to the right of the vehicle. Therefore, a left turn should be made. Conversely, a right turn. Similarly, if the current latitude error is less than 0 and less than the turning radius in step one, the detected target is located to the left of the underwater vehicle, and a right turn should be made. Conversely, a left turn should be made. Figure 4 As shown.

[0098] Step 4: The sliding window continuously analyzes and records the current navigation information of the vehicle, especially its position. The error step size is set to 10. If, after determining the direction of motion in Step 3, 10 consecutive cases occur where the turning radius ρ is greater than the target radius, the current direction is changed to flexibly address incorrect judgments.

[0099] In step four, the sliding window, as shown... Figure 3 As shown. This refers to defining an array whose initial range is a range: [a i ,b i ].

[0100] The boundary width z_i is determined according to the actual situation.

[0101] z i =b i -a i (14)

[0102] A dynamic data analysis method that continuously adds or deletes boundaries based on the required data characteristics, as shown in Formula 15.

[0103] arr[i]>p, 1<=i<=10 (15)

[0104] When detection begins, the position information error from step two is sequentially added to the defined interval and counted into the window, and values ​​are assigned to the variable array. When the number of array variables reaches 10, i.e., the window is full, the data in the window is compared one by one with the turning radius, as shown in Formula 15.

[0105] If the absolute value of the position error between all 10 position data points and the detected object is less than the turning radius, the boundary is narrowed, and the underwater vehicle maintains its original direction of motion. Otherwise, if five consecutive position data points within the window are greater than the turning radius, the underwater vehicle reverses its direction of motion. Other cases are handled the same as the first case, maintaining the same direction of motion. After the next data point enters the window, a fault-tolerant judgment is performed for the next position data point until the mission ends and the underwater vehicle returns to the mother ship.

[0106] The invention will now be described in further detail with reference to simulation examples and accompanying drawings:

[0107] To verify the effectiveness of the invented fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering, simulation verification was conducted on the position information and errors during the target detection process. Following step one of the invention's description, the current target detection area was set as Liquan Lake Reservoir in Shaanxi Province, with a longitude of approximately 108.4316 and a latitude of approximately 34.52. The error was within 0.1%. According to the underwater vehicle detection mission flow, it was assumed that the target was known to be located within this area.

[0108] Based on step one, assuming the underwater vehicle's latitude and longitude at times t and t0 are (108.432583, 34.5244) and (108.43098, 34.5245) respectively, the direct distance can be calculated using a specific formula. Then, using the geometric diagram in the figure and formulas 4) and 5), considering that for equal latitude, the distance difference is approximately 100m for every 0.001 degree; and for equal longitude, the distance difference is 111m for every 0.001 degree of latitude, the current average turning radius ρ is 9.2m.

[0109] By using the underwater vehicle's position information prediction during target detection in step two, and combining it with the sliding window mechanism in step three, the predicted information is evaluated. Using data collected in previous experiments and the turning radius calculated in step one, position information prediction via Kalman filtering and fault-tolerant processing of the predicted information via the sliding window are achieved. The results are as follows... Figures 5-6 As shown, Figure 5 Under noise interference, the aircraft uses the proposed target detection fault-tolerant processing method to process sensor measurement information through a sliding window and then reverses the fault tolerance. Figure 5 and Figure 6 Based on the original measurements, after processing with a fault-tolerant method, the original method remains unchanged because the data within the sliding window mechanism is smaller than the average radius. The proposed method can improve detection accuracy and reduce the target detection error rate by predicting position information during the underwater vehicle detection phase. Compared with target detection methods that do not use Kalman filtering prediction followed by fault-tolerant processing, the proposed method can effectively reduce detection errors and noise interference.

[0110] 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 fault-tolerant processing method for underwater vehicle target detection based on Kalman filtering, used for underwater vehicle target sensing and detection, characterized in that: Step 1: The underactuated underwater vehicle travels in a straight line, and the turning radius is determined based on the turning model data; Step 2: Predict the fused navigation information based on discrete Kalman filtering, and compare the navigation error with the average turning radius; the specific process is as follows: Assuming all disturbances are Gaussian distributed, the state variable X(k) = [x(k), y(k), θ(k)] represents the current information of the underwater vehicle, where x(k) is the current position, y(k) is the current velocity, and θ(k) is the yaw angle. Using the underwater vehicle's position as the observation variable, the Kalman measurement equation is established as follows: （2） Where H(k) is the observation matrix and V(k) is the observation noise; Perform the next step of estimating the system's position, velocity, and yaw angle according to equation (3): （3） Where k represents a discrete time node; This represents the predicted state variables for time k at time k-1, including heading angle, position, and velocity information; This is the predicted state equation of the system at time k-1 to time k, and it is also the state transition matrix of the system; （4） Estimated system state update: （5） in This represents the Kalman gain matrix at time k. This represents the system measurement sequence from time 0 to time k. This represents the measurement matrix of the system at time k. Represents the predicted state variables of the system at time k; Variance of prediction estimation error: （6） in, represents the system state estimation error covariance at time k-1, which is set to 0 at time 0; Q is the system noise covariance matrix. It is the system's state transition matrix. yes The transpose of the matrix; based on prior experience, Q is set as: （7） Therefore, the filter gain equation is: （8） in It is the predicted Kalman gain matrix for time k at time k-1; This represents the covariance of the system state estimation error at time k-1; This represents the transpose of the measurement matrix at the current time. This is the measurement matrix at the current moment; where R is the system measurement noise covariance matrix: （9） During target detection, the position at the next moment is predicted using Kalman filtering; and based on the currently known target latitude and longitude [x]... i ',y i '] and the latitude and longitude obtained by the current underwater vehicle's own navigation system [x i ,y i ]Calculate the error ∆x and compare it with the average turning radius in (1). For comparison, if the longitude or latitude is smaller than the turning radius... Then proceed to step three; the error calculation is as follows: （10） Where, x i Let y represent the longitude at time i. i Let represent the latitude at time i. express The norm; Step 3: Fusion of position information and side-scan sonar feature variables, and preliminary fault tolerance processing; the preliminary fault tolerance processing package is: based on the position information obtained in Step 2, combined with the side-scan feature value g, to determine the direction; the side-scan feature value g is collected in real time by the side-scan sonar data board and sent to the main control, representing the percentage of effective side-scan data; Step 4: The sliding window will collect and record the current navigation information of the vehicle in real time.

2. The underwater vehicle target detection fault-tolerant processing method based on Kalman filtering according to claim 1, characterized in that: In step one, the turning model data collected and statistically analyzed include the underwater vehicle's velocity v and position x; the turning radius is ρ, and n sampling points are taken, namely x1, x2, ..., xn. Based on geometric relationships, the turning radius can be estimated from any sampling point. .

3. The underwater vehicle target detection fault-tolerant processing method based on Kalman filtering according to claim 2, characterized in that: Average turning radius Calculation: （1） in, h is the ideal turning radius of the underwater vehicle. Location and Location mapping to The vertical distance between the position and the center of the turning circle ; △x' is the underwater vehicle under ideal conditions. Location and The positional deviation, θ, is the distance the spacecraft travels from... Location to The yaw angle of the position; It is the sum of the turning radii of the aircraft at positions x1, x2, ... xn.

4. The underwater vehicle target detection fault-tolerant processing method based on Kalman filtering according to claim 3, characterized in that: The current longitude error is less than the average turning radius. If the value is negative, then according to the changes in geographical longitude and latitude, it can be divided into changes in longitude and latitude. If the direction is north-south, the turning course depends on the difference in longitude; in this case, if the error is less than 0 and its absolute value is less than the average turning radius in step one. If the target is to the right of the vehicle, then turn right; otherwise, turn left. Similarly, the current latitude error is less than 0 and less than the average turning radius in step one. It can be seen that the target being detected is located to the left of the underwater vehicle, so it should turn left; otherwise, it should turn right.

5. The underwater vehicle target detection fault-tolerant processing method based on Kalman filtering according to claim 4, characterized in that: In step four, the sliding window refers to the method in industrial control algorithms where a fixed-length array is used, with an initial range of [1, 10], to represent an error step size of 10. When the vehicle begins to probe, and the navigation information is within the interval in real time, the predicted value in step two is compared and the window array is assigned a value. When the array variable has been assigned 10 values, that is, the window has been filled, it is then determined whether five consecutive position errors are greater than the turning radius. If five consecutive position errors are greater than the turning radius, the underwater vehicle moves in the opposite direction; otherwise, it maintains the original direction. The window variable is reset upon completion, and the position error is reassigned until the task ends or an error occurs and error handling is performed. Repeat step four until the exploration mission is completed and the underwater vehicle returns to the mother ship.

6. A vehicle employing the underwater vehicle target detection fault-tolerant processing method based on Kalman filtering as described in any one of claims 1-5, characterized in that: The vehicle is equipped with a Doppler velocimeter, inertial navigation, GPS, and side-scan sonar. It obtains its current position by combining the Doppler velocimeter with inertial navigation and GPS. The side-scan sonar is used to detect the specific size and shape information of the target.

Images