Power damage state identification and fault-tolerant control method for unmanned ship driven by double propellers

Abstract

The invention provides a power damage state identification and fault-tolerant control method for a double-propeller-driven unmanned ship. The method comprises the following steps: firstly, acquiring navigation characteristics of the unmanned ship based on a motion effect of actually measured data, and forming a training and testing set by normal state navigation data and damaged state navigation data; secondly, recording manipulation behaviors and motion responses in the navigation process, comparing training features, and establishing a dynamic damage state recognition algorithm through a deep neural network; and finally, on the basis of a damage degree identification algorithm and a control end adaptive software and hardware system, a propeller instruction setting method is determined, and the unmanned ship can still keep certain autonomous navigation capability in a damaged state by taking speed reduction and direction keeping navigation as a control target. According to the scheme, the problem that an unmanned ship driven by double propellers is low in power system fault handling capacity is solved, and the unmanned ship can still keep the autonomous navigation capacity to a certain degree in the power abnormal state.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned boat navigation control, and in particular to a method for identifying power damage state and fault-tolerant control of an unmanned boat driven by two propellers. Background Art

[0002] Unmanned boats have the characteristics of high cost-effectiveness, strong seaworthiness, wide combat use and high intelligence. They are important new combat equipment in future intelligent warfare. Compared with manned boats, surface unmanned boats have significant advantages in tracking and attack, anti-submarine warfare, environmental monitoring, maritime transportation and other scenarios. They do not need to consider the limitations of manned design, which can promote the upgrade of ship functions, enhance mission execution capabilities and improve mission execution efficiency. In order to give full play to the advantages, in recent years, new power propulsion systems and adaptive control methods suitable for unmanned concepts have gradually become new hot topics for discussion. For unmanned boats, the design process does not need to consider the temperature and noise impact of the main power equipment layout on the navigation personnel. Therefore, the ship has a freer layout range and design upper limit, and is more likely to break through the performance limit of traditional power propulsion systems.

[0003] However, the reliability and stability of the power system have always been the key technical challenges in the development of unmanned surface ships. The failure of the power system is not a special problem in the development of unmanned ships, but also exists on manned ships. However, for unmanned ships, the importance of power system failure is higher and the challenge is greater. First, autonomy and strong survivability are important indicators of unmanned ships. Once an unmanned ship has a power failure, it will not only seriously affect its autonomous navigation performance and mission execution capability, but may also threaten navigation safety and cause unnecessary losses; secondly, the unmanned boat is away from human intervention for a long time and cannot rely on direct intervention of personnel to solve the problem. The maintenance response is not timely, and the characteristic signal of power abnormality cannot be fed back in time. Therefore, the requirements for the self-sustaining power and reliability of the power system are higher; in addition, one of the advantages of unmanned ships is that they can perform tasks under special working conditions and extreme environments. In this case, the ship is more prone to various failures or abnormal conditions due to environmental factors.

[0004] In summary, power damage identification and fault-tolerant control are key challenges for unmanned ships. The strong autonomous navigation capability of unmanned ships requires the ability to identify and respond to power system anomalies as a basis. Therefore, how to accurately identify the abnormal state of unmanned boat power and take effective fault-tolerant control measures has become a key technology to ensure the reliable operation of unmanned ships. Summary of the invention

[0005] The purpose of the present invention is to provide a method for identifying the power damage state and fault-tolerant control of a twin-propeller driven unmanned boat, construct a control motion mapping for the unmanned boat driven by the stern twin propellers, identify the degree of power damage based on a deep neural network, design a control rate according to the damage degree identification result, and perform fault-tolerant control with the goal of reducing speed and maintaining direction to solve the problem of power system abnormality.

[0006] The specific technical solution of the present invention is to provide a method for identifying the power damage state and fault-tolerant control of a double-propeller driven unmanned boat, and the method for identifying the power damage state and fault-tolerant control of a double-propeller driven unmanned boat comprises the following steps:

[0007] Step 1: construct the unmanned boat maneuvering and motion feature mapping based on the measured data, and use the outlier detection algorithm based on the navigation data to ensure the data validity;

[0008] Step 2: Establish a power damage state recognition algorithm through a deep neural network to qualitatively analyze the power damage degree of the unmanned boat;

[0009] Step 3: Determine the thruster command setting method based on the damage identification algorithm.

[0010] Furthermore, in step 1, the following steps are also included:

[0011] Step 11, constructing the unmanned boat manipulation and motion feature mapping;

[0012] Step 12: An outlier detection algorithm based on navigation data is used to ensure data validity;

[0013] Step 12 also includes:

[0014] Step 121: Screening based on Z score;

[0015] Step 122: Screen again based on the local abnormal factors.

[0016] Furthermore, in step 11, the manipulation motion mapping relationship is:

[0017] a→P

[0018]

[0019] a is the acceleration at the current moment, P is the propulsion power, F is the three-degree-of-freedom force and torque matrix, m is the mass of the unmanned boat, and v is the speed of the unmanned boat;

[0020] Based on the unmanned ship with dual propeller differential drive, the decomposition mapping of acceleration on the inertial system of the body can be written as:

[0021] P2:(a x ,a y )→(Tl ,T r )

[0022] Among them, a x ,a y are the accelerations of the unmanned boat perpendicular to the bow direction and parallel to the bow direction, T l ,T r These are the left and right propeller thrust throttles respectively.

[0023] Furthermore, in step 121, the Z-score is expressed as:

[0024]

[0025] The expected value of any variable X is E(X) and the standard deviation is σ(X);

[0026] The sample size is selected as n, and the upper limit of the absolute value of the Z score of a single data point is taken as σ, that is, if each data point does not constitute a significant difference with other values ​​within t seconds before and after it, it is recorded as valid data.

[0027] Furthermore, in step 122, the selection is performed again based on the local abnormality factor:

[0028] Given a parameter p, the distance d from data point p to data point o is p k (p,o) is expressed as:

[0029] d k (p,o)=max{d k (o),d(p,o)}

[0030] o is the point closest to p, d k (o) is the distance between o and the kth closest point to o, and d(p,o) is the distance between p and o;

[0031] The local reachable density of data point p is defined as:

[0032]

[0033] N k (p) is all distances p less than d k The points of (p) constitute the k nearest neighbors of p;

[0034] The local outlier factor LOF of a data point p is the ratio of the average local reachable density of points in the neighborhood of point p to the local reachable density of data point p:

[0035]

[0036] LOF k (p) local outlier factor represents the neighborhood N of point p k(p) is the average of the ratios of the local reachability density of other points within the point to the local reachability density of point p;

[0037] If the ratio is closer to 1, it means that the density of point p's neighborhood points is similar, and point p may belong to the same cluster as its neighborhood points; if the ratio is less than 1, it means that the density of point p is higher than the density of its neighborhood points, and point p is a dense point; if the ratio is greater than 1, it means that the density of point p is lower than the density of its neighborhood points, and point p may be an outlier.

[0038] Furthermore, in step 2, the following steps are also included:

[0039] Step 21, defining the damage degree;

[0040] The absolute propeller damage is defined as:

[0041]

[0042] Among them, Δ X is the absolute damage degree of the propeller with higher power loss, Δ Y is the absolute damage degree of the propeller with lower power loss, Δ X >Δ Y ; The throttle at the propeller control end with higher power loss is T X , the throttle at the propeller control end on the other side is T Y The equivalent propulsion throttles of the actual motion of the two propellers are

[0043] Damage is defined as:

[0044] Δ=Δ X -Δ Y

[0045] Δ is the degree of damage;

[0046] Step 22, establishing power damage state recognition through deep neural network.

[0047] Furthermore, in step 22, the following steps are also included:

[0048] Step 221: The speed, acceleration, left and right throttles, and the time corresponding to each set of data and the valid navigation information are filtered to form input feature values;

[0049] Step 222: Perform activation function transformation on the hidden variable as the input of the next fully connected layer; its expression is:

[0050] ReLU(x)=max(x,0)

[0051] ReLU(x) is the ReLU function, x is the input feature value of the activation function, which is the filtered valid navigation information;

[0052] Step 223: The fully connected layer is composed of multiple hidden layers with extended widths, and the hidden layer output is transformed by an activation function;

[0053] Step 224: The output layer uses the Huber function as the loss function for the regression problem. The Huber error is defined as follows:

[0054]

[0055] Among them, y is the true value, f(x) is the predicted value, and δ is a hyperparameter used to control the turning point of the error.

[0056] Furthermore, in step 3, the following steps are also included:

[0057] Step 31: Calculate the initial power reduction;

[0058] The difference between the calculated throttle value and the true value is calculated. When the damage Δ is 0, that is, when the ship's propulsion system is normal, the difference should be a curve close to 0. The damage time interval is [t0,t n ], n+1 groups of data are recorded in the interval, which are:

[0059]

[0060] Indicates t n Acceleration information recorded at all times, {a} represents the acceleration time series of the unmanned boat;

[0061] Based on the deep neural network, a power damage state recognition algorithm is established. The heading and vertical acceleration are input, and the output is the equivalent propulsion throttle:

[0062]

[0063] Indicates the equivalent throttle of the left and right propellers calculated from the nth set of recorded acceleration information. It represents the time series of equivalent propulsion throttle of the propeller of the unmanned boat;

[0064] In the time interval [t0,t n ], calculate the cumulative deviation of the boost throttle:

[0065]

[0066] T δX (n) represents the cumulative throttle deviation of the X propeller with higher power loss in the nth group of recorded information, T δY (n) indicates the cumulative throttle deviation of the Y propeller with the lower power loss in the nth group of recorded information. represents the equivalent throttle of the X propeller calculated from the i-th group of recorded acceleration information, T iX Indicates the throttle of the X propeller control end recorded at the time of the i-th group of information, represents the equivalent throttle of the Y propeller calculated from the i-th group of recorded acceleration information, T iY represents the throttle of the Y propeller control end recorded at the i-th group of information;

[0067] And calculate to t n The absolute damage degree of the propeller at the moment:

[0068]

[0069] Δ X (n) represents the absolute damage degree of propeller X calculated from the nth set of recorded information, Δ Y (n) represents the absolute damage degree of propeller Y calculated from the nth set of recorded information;

[0070] Then t n The damage level of the ship at this moment is:

[0071] Δ(n)=Δ X (n)-Δ Y (n)

[0072] Step 32: multiple iterations with the adjusted relative damage degree of the motion effect being less than a preset value as the termination condition.

[0073] The beneficial effects achieved by the present invention are:

[0074] The unmanned boat power damage state identification and fault-tolerant control method proposed in the present invention can identify the problem and take corresponding measures to maximize the stability and safety of the boat when the unmanned boat has power damage or failure. It does not require additional device systems and can be transferred to most double-propeller driven unmanned boats.

[0075] The damage status recognition of the unmanned boat proposed in the present invention is realized based on the measured data during the navigation process, which not only saves the complicated fault diagnosis process, but also avoids the problem of difficult attribution. It recognizes the manipulation behavior through the motion effect, which is simple, reliable and has lower cost. The recognition is based on the training of the neural network in the early test stage, and can output a more immediate response to the navigation data, and quickly make error correction adjustments;

[0076] After the method proposed in the present invention identifies that a power anomaly occurs at a propeller end, a fault-tolerant control method suitable for power damage can be designed to adjust the control instructions of other propellers, thereby achieving the navigation goal of reducing speed and maintaining direction, so that the unmanned ship can still maintain a certain degree of autonomous navigation and mission execution capabilities in the event of a power anomaly, which highly meets navigation requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a flow chart of a method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to an embodiment of the present invention;

[0078] Figure 2 is a diagram showing a calculation of the recognition accuracy of a known degree of damage verified by motion effect according to an embodiment of the present invention;

[0079] Figure 3 It is a schematic diagram of the tracking control effect after identifying the unknown damage degree according to an embodiment of the present invention. DETAILED DESCRIPTION

[0080] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0081] Embodiment 1

[0082] like Figure 1 As shown, this embodiment provides a method for identifying and fault-tolerantly controlling the power damage state of a double-propeller driven unmanned boat.

[0083] The method for identifying and fault-tolerantly controlling the power damage state of a dual-propeller driven unmanned boat comprises the following steps:

[0084] Step 1: Based on the measured data, construct the unmanned boat maneuvering and motion feature mapping, determine the data related to the mapping model and its acquisition and adjustment methods. Develop an outlier detection algorithm based on navigation data to ensure data validity;

[0085] In step 1, the following steps are also included:

[0086] Step 11, constructing the unmanned boat manipulation and motion feature mapping;

[0087] The manipulation motion mapping relationship of the target construction is set as:

[0088] According to Newton's second law, the dynamic process of the maneuvering motion model of a three-degree-of-freedom ship can be expressed as F=Ma, where F is the three-degree-of-freedom force and torque matrix, M is the three-degree-of-freedom mass and inertia matrix, and a is the acceleration at the current moment.

[0089] The propulsion force in a narrow sense can be expressed by the propulsion power and the ship speed as follows:

[0090]

[0091] The propulsion power P can be regarded as a physical quantity that is linearly related to the throttle, so that the relationship between the control behavior and the motion response can be constructed:

[0092]

[0093] Therefore, the mapping to be constructed is:

[0094] P1:a→P

[0095] For the unmanned ship with dual propeller differential drive, the decomposition of acceleration in the inertial system of the accompanying body is examined, and the above mapping can be written as:

[0096] P2:(a x ,a y )→(T l ,T r )

[0097] Among them, T l ,T r They are the left and right propeller thrust throttles respectively. In practical applications, the motion data obtained by the above mapping during navigation are discrete points.

[0098] Step 12: An outlier detection algorithm based on navigation data is used to ensure data validity;

[0099] The significance and setting of outlier detection algorithm are:

[0100] In actual applications, GPS and inertial navigation systems may be subject to various interferences, resulting in unavailable perception data. The causes include obstructions, multipath effects, weather conditions, electromagnetic interference, reading drift, etc. For the navigation perception system of unmanned ships, these interferences will cause some data to be erroneous or unavailable, and short-term unavailability will not affect the navigation and mission execution of the ship. However, for the data set of power damage state recognition, the fallacy of perception data will cause the training set to be contaminated, resulting in the deterioration of training effect. In addition, due to the large amount of data, manual screening is not realistic. Therefore, it is necessary to develop an effective navigation information screening algorithm. The goal of screening is to filter the perception data, remove outliers, correct errors, etc. by adopting appropriate algorithms and technologies to improve the quality and availability of data. Combined with the experience of actual ship testing, the error phenomena caused by various interferences are different. For example, weather reasons may cause the GPS update rate to decrease, and multiple data values ​​(latitude and longitude, speed, heading angle, etc.) obtained continuously may be the same; electromagnetic interference may cause data jumps, and several discrete data points are obviously inconsistent with the data values ​​in their neighborhood.

[0101] To solve the above problems, an outlier detection algorithm based on navigation data can be used. The algorithm can be specifically divided into:

[0102] Step 121: Z score screening. For many types of failed navigation information, rough precision failure value screening can be performed by judging the Z score of a certain value in the sample composed of nearby data. Assuming that the expected value of any variable X is E(X) and the standard deviation is σ(X), the Z score is expressed as

[0103]

[0104] The sample size is selected as n, and the upper limit of the absolute value of the Z score of a single data point is taken as σ, that is, if each data point does not constitute a significant difference with other values ​​within t seconds before and after it, it is recorded as valid data.

[0105] Step 122: Screening of local abnormal factors. The Z-score method can only realize the screening of sample points under coarse precision, and has certain limitations in the establishment of a data set for power damage state identification. Since the propulsion signal of the unmanned boat is a step signal, the propeller speed can change rapidly, resulting in the motion response being a step under the discreteness of adjacent samples. When a smaller upper bound σ of the absolute value of the Z-score is used, such steps may be regarded as abnormal points and eliminated, while when a larger upper bound σ of the absolute value of the Z-score is selected, a jump value close to the neighborhood value may be regarded as a normal step and retained, thereby polluting the data set. Therefore, this step performs a refined secondary screening based on local abnormal factors on the data set after the coarse precision screening of the Z-score.

[0106] Local Outlier Factor (LOF) is a method for describing outliers, which is often used to identify outliers that do not conform to the trend in a large number of data sets. The LOF algorithm measures the degree of abnormality by examining the relative density of the point and the adjacent data points, allowing for uneven data distribution and different densities. Therefore, when facing the abnormal problem of navigation data in this part, it can avoid screening misjudgments caused by data steps. The process of this method is as follows:

[0107] In the same sample set, consider point p, the point closest to p is o, and the distance between p and the point closest to p is d k (p), the distance in this problem is the absolute value of the difference between the values.

[0108] Define all distances p less than d k The points of (p) constitute the k nearest neighbors of p, denoted by N k (p).

[0109] When parameter p is given, the p-reachable distance from data point p to data point o is expressed as

[0110] d k (p,o)=max{d k (o),d(p,o)}

[0111] The limit on the reachable distance significantly reduces the statistical fluctuations of all data points p close to o. The strength of this smoothing effect can be controlled by the parameter k. The higher the k value, the more similar the reachable distances of objects in the same neighborhood, and the smoother the result.

[0112] Next, define the local reachable density of data point p as

[0113]

[0114] The local reachable density is the inverse of the average reachable distance of the k nearest neighbors of p. The larger the distance, the smaller the density. It is the average distance from the k neighbors of the observation point to the observation point. The sample capacity of the dynamic sample is set to n. According to the definition of local reachable density, if the observation data point is relatively discrete relative to other data points, its local reachable density is small. The local relative density (local outlier factor LOF) of data point p is the ratio of the average local reachable density of points in the neighborhood of point p to the local reachable density of data point p, that is:

[0115]

[0116] The local outlier factor represents the neighborhood N of point p k (p) is the average of the ratios of the local reachable density of other points in the cluster to the local reachable density of point p. If the ratio is closer to 1, it means that the density of the neighborhood points of point p is similar, and point p may belong to the same cluster as the neighborhood; if the ratio is less than 1, it means that the density of point p is higher than the density of its neighborhood points, and point p is a dense point; if the ratio is greater than 1, it means that the density of point p is lower than the density of its neighborhood points, and point p may be an abnormal point. Considering that the speed and acceleration information may fluctuate when the working conditions change, combined with the test results, γ is selected as the upper limit of the local anomaly factor LOF. When the upper limit is exceeded, it is regarded as a reception anomaly, and the interpolation of the neighborhood data in the same group is used as the information at that moment.

[0117] In addition, the LOF algorithm can also solve the problem of multiple identical or close data caused by reception delay, signal interruption, etc. If there are more than or equal to k repeated or very close points in the sample, the average reachable distance of these points is close to zero, and the local reachable density tends to infinity. When the local reachable density ρ in the sample set is k When it is greater than the set threshold β, it can be determined that a data error caused by signal interference has occurred, and an error signal should be transmitted to the control end.

[0118] Step 2: Propose the concept of damage degree, and establish a power damage state recognition algorithm through a deep neural network to qualitatively analyze the power damage degree of the unmanned boat;

[0119] In step 2, the following steps are also included:

[0120] Step 21, defining the damage degree;

[0121] The definition and calculation method of damage degree are set as follows:

[0122] Without loss of generality, assume that in the unmanned ship with dual propeller differential propulsion, the throttle at the propeller control end with higher power loss is TX , the throttle at the propeller control end on the other side is T Y The equivalent propulsion throttles of the actual motion of the two propellers are The absolute damage degree of the propeller is defined as

[0123]

[0124] Among them, Δ X >Δ Y .

[0125] The damage degree Δ is defined as

[0126] Δ=Δ X -Δ Y

[0127] The damage degree here is a relative value, that is, it is possible that both propellers on both sides are damaged. This is because the present invention takes the control effect as the design goal of the identification algorithm, and does not quantitatively analyze the output power of the propellers on both sides, but analyzes their relative propulsion throttle relationship in order to achieve the control effect of speed reduction and direction preservation. Therefore, it is only necessary to construct the equivalent propulsion throttle of the actual movement of the two propellers. The propeller damage degree can be obtained by the calculation method. This calculation method is essentially equivalent to mapping P2. However, since the floating of micro data points cannot be eliminated, it is necessary to construct a quantitative calculation method for the overall trend over a period of time, that is, to establish a mapping:

[0128]

[0129] Step 22, establishing power damage state recognition through a deep neural network;

[0130] Furthermore, in step 2, the neural network for studying the mapping relationship between manipulation behavior and motion effect is composed of feature input, activation function, batch normalization, fully connected layer and regression function, which is specifically set as follows:

[0131] Step 221: Feature input. The feature input layer is the data collection part of the neural network. The input content here includes speed, acceleration, left and right throttles, and the time corresponding to each set of data. After the data setting and effective navigation information screening mentioned above, the input feature value is formed. The acceleration is in units of gravity acceleration g, and the thrust throttle is a positive integer of [20,100], which represents the ratio of the current throttle to the limit throttle.

[0132] Step 222: Activation function. The hidden variables are transformed by activation function and used as the input of the next fully connected layer. In deep neural network training, the present invention uses ReLU function as the activation function, which is expressed as

[0133] ReLU(x)=max(x,0)

[0134] Step 223: Fully connected layer. The fully connected layer consists of multiple hidden layers with extended widths, and the hidden layer output is transformed by an activation function. The hidden layer extracts more meaningful and abstract features from the original input data by learning weights and bias parameters. By increasing the number of hidden layers and the number of neurons, the expressive power of the neural network can be increased, and redundancy and noise can be removed while retaining important information. The Adam algorithm is used to update the weight parameters of the neural network. The Adam algorithm combines the ideas of the momentum method and the adaptive learning rate, comprehensively considers the first-order moment estimation and the second-order moment estimation of the gradient, and then calculates the update step size.

[0135] Step 224: Output layer. The output layer uses the Huber function as the loss function for the regression problem, which compromises the square loss and absolute value loss of the error within a certain range. The definition of Huber error is as follows:

[0136]

[0137] Among them, y is the true value, f(x) is the predicted value, and δ is a hyperparameter used to control the turning point of the error.

[0138] The input data of the neural network is the motion response, and the output is the control behavior. The training set contains input and output information, and the validation set gives the input data to detect whether the output effect is consistent with the actual throttle control. For any i, the motion effect data within a period of time is input, such as acceleration information {(a ix ,a iy )}, the manipulation behavior data is obtained, and the damage degree identification algorithm goal can be achieved. The actual construction mapping of the neural network proposed in the present invention is

[0139]

[0140] Through the above steps, the solution proposed by the present invention can calculate the relative damage degree of the propeller of the double-propeller driven unmanned boat through the acceleration record value during the navigation movement, and further predict the actual propulsion effect of the propeller.

[0141] Step 3: Based on the damage degree identification algorithm and the adaptive software and hardware system of the control end, determine the thruster command setting method, with reduced speed and directional navigation as the control target, so that the unmanned ship can still maintain a certain degree of autonomous navigation capability in a damaged state.

[0142] In step 3, the following steps are also included:

[0143] The thruster command setting method is set as:

[0144] The recognition algorithm in step 2 can determine the power damage of propeller X on one side relative to propeller Y on the other side, and then actively reduce the propulsion command of propeller Y. The difference of the initial reduction is the throttle corresponding to the damage, so that the propellers on both sides rotate at the same power state under the damage of the initial verification. Afterwards, considering the complex nonlinear mapping relationship between the throttle control and the power response, the control effect of the initial throttle reduction may still not achieve the goal of keeping direction. At this time, the navigation data is obtained again through the perception system, and added to the recognition algorithm and control algorithm iteration as feedback, and finally the goal of keeping direction control is achieved.

[0145] Step 31: Calculation of initial power reduction: Within a period of time, the calculated throttle value is subtracted from the actual value. When the damage Δ is 0, that is, when the ship's propulsion system is normal, the difference should be a curve close to 0. The damage time interval is [t0, t n ], n+1 groups of data are recorded in the interval, which are

[0146]

[0147] According to the deep neural network algorithm designed in step 2, the heading and vertical acceleration are input, and the output is the equivalent propulsion throttle

[0148]

[0149] In the time interval [t0,t n ], calculate the cumulative deviation of the boost throttle

[0150]

[0151] From this we can calculate t n The absolute damage degree of the propeller at any time

[0152]

[0153] Then t n The ship's damage at this moment is

[0154] Δ(n)=Δ X (n)-Δ Y (n)

[0155] The volatility of damage decreases with the increase of ship sailing time. When the volatility of damage Δ(n) is less than the set threshold η within a continuous period of τ, the damage identification is considered to be completed, and the compensation reduction coefficient calculated initially can be obtained.

[0156] Step 32: Iterative control law setting. Although in some cases, relatively accurate results can be obtained in the initial calculation, there may still be throttle intervals with fewer training data sets, lower recognition accuracy, and sparser eigenvalues ​​in the sample space composed of all maneuvering behaviors. At this time, the instructions calculated initially may have lower accuracy. Therefore, multiple iterative calculations are required to achieve sufficient accuracy. After power compensation reduction, repeat the contents of steps 2 and 3, re-identify the degree of damage, and update the power reduction coefficient. The present invention uses the adjusted relative degree of damage of the motion effect power to be less than α as the termination condition, and α is initially set to 5% according to the test ship type.

[0157] Embodiment 2

[0158] In this embodiment, two experiments were carried out to verify the feasibility of the scheme.

[0159] The hull parameters selected for the test are shown in Table 1.

[0160] Table 1 Test ship parameters

[0161]

[0162]

[0163] In the test, the parameter values ​​of the Z-score method selected according to the test ship type are shown in Table 2.

[0164] Table 2 Z score method parameters

[0165]

[0166] In the test, the parameter values ​​of the local anomaly factor method selected according to the test ship type are shown in Table 3.

[0167] Table 3 Local anomaly factor method parameters

[0168]

[0169] In the test, the neural network training parameters selected according to the data conditions and training results are shown in Table 4.

[0170] Table 4 Neural network training parameters based on measured data

[0171]

[0172]

[0173] Test 1: During the actual navigation process, the power attenuation is actively set on the output end of one propeller, and the accuracy of power damage recognition is verified by the navigation test. In this test, the propeller propulsion throttle loss on one side is artificially set to 20%, and the goal is to identify the 20% damage through the navigation effect.

[0174] Mapping based on actual construction of neural network Input heading and vertical acceleration, and the output is equivalent propulsion throttle

[0175]

[0176] In the time interval [t0,t n ], the cumulative deviation of the boost throttle is obtained as follows: Figure 1 -(a), Figure 1 -(b) as shown.

[0177]

[0178] From this we can calculate t n The absolute damage degree of the propeller at any time

[0179]

[0180] Then t n At time Δ(n) = Δ X (n)-Δ Y (n) Figure 1 -(c). The volatility of the damage decreases as the ship's sailing time increases. When the volatility of the damage Δ(n) is less than the set threshold η within a continuous period of τ, the damage identification is considered to be completed, and the initial calculated compensation reduction coefficient can be obtained. In this test, the relative damage calculated was about -20.8256%, that is, the right propeller power reduction was about 20%, which is consistent with the setting. This test accurately reflects the degree of power damage.

[0181] In Test 2, when the propeller was entangled with foreign objects such as water plants and fishing nets, resulting in abnormal power output, the reliability of fault-tolerant control was verified by identifying the power damage degree and the control effect under tracking navigation state.

[0182] Mapping based on actual construction of neural network Input heading and vertical acceleration, and the output is equivalent propulsion throttle

[0183]

[0184] In the time interval [t0,t n ], calculate the cumulative deviation of the boost throttle

[0185]

[0186] From this we can calculate t n The absolute damage degree of the propeller at any time

[0187]

[0188] Then t n At time Δ(n) = Δ X (n)-Δ Y (n) Figure 3 -(a). The volatility of damage decreases with the increase of the ship's sailing time. When the volatility of damage Δ(n) is less than the set threshold η within the continuous τ time, the damage identification is considered to be completed, and the initial calculated compensation reduction coefficient can be obtained. In this experiment, the relative damage calculated was about 45.5232%, that is, the power reduction of the left propeller was about 45%. The accuracy of this value was verified by the fault-tolerant tracking control effect.

[0189] The right throttle is actively reduced to 54.4768% of the original value as the initial precision instruction of the direction-keeping control. The PID control method is used for verification, and the navigation target point is set. The target point connection line is used as the path for tracking navigation. The effect of fault-tolerant control can be verified by the tracking effect on the actual ship to determine whether the goal of direction-keeping navigation is achieved. The navigation trajectory and throttle timing curve are shown in Figure 3 -(b), Figure 3 -(c). In the experiment, the damage degree can achieve the ideal effect after one calculation, without iterative correction.

[0190] Depend on Figure 3 -(b) It can be seen that the target track (orange broken line) and the actual track (blue curve) have a high degree of fit. When the actual track is approximately a straight line, the left and right throttle values ​​at the control end are not the same. The throttle value of the left propeller is approximately 22-28, and the throttle value of the right propeller is approximately 10-20. The actual throttle value of the left propeller at the output end needs to be reduced to 54.4768% of the original. At this time, the actual throttle values ​​of the left and right propellers are almost the same, and the actual ship should be approximately straight, which is consistent with the actual motion effect. Therefore, the calculation of the power damage degree and the fault-tolerant control effect of Test 2 can be preliminarily verified.

[0191] The technical solution of the present invention is described in detail above in conjunction with the accompanying drawings. The present invention proposes a method for identifying the power damage state and fault-tolerant control of a twin-propeller driven unmanned boat, constructs a control motion mapping for an unmanned boat driven by a stern twin propeller differential, and identifies the degree of power damage based on a deep neural network. The control rate is designed according to the damage degree identification result, and fault-tolerant control is performed with the goal of reducing speed and maintaining direction to solve the problem of power system abnormality.

[0192] The steps in the present invention can be adjusted in order, combined or deleted according to actual needs.

[0193] Although the present invention is disclosed in detail with reference to the accompanying drawings, it should be understood that the above embodiments are only for illustrating the technical ideas of the present invention and cannot be used to limit the scope of protection of the present invention. The scope of protection of the present invention is defined by the appended claims and may include various modifications, alterations and equivalents made to the invention without departing from the scope and spirit of the present invention.

[0194] The present invention is not limited to the above-mentioned specific implementation modes. A person skilled in the art may implement the present invention in various other specific implementation modes according to the embodiments and the disclosure of the drawings. Therefore, any design that adopts the design structure and concept of the present invention and makes some simple transformations or changes shall fall within the scope of protection of the present invention.

Claims

1. A method for identifying and fault-tolerantly controlling the power damage state of a double-propeller driven unmanned boat, characterized in that: The method for identifying and fault-tolerantly controlling the power damage state of a dual-propeller driven unmanned boat comprises the following steps: Step 1: construct the unmanned boat maneuvering and motion feature mapping based on the measured data, and use the outlier detection algorithm based on the navigation data to ensure the data validity; Step 2: Establish a power damage state recognition algorithm through a deep neural network to qualitatively analyze the power damage degree of the unmanned boat; Step 3: Determine the thruster command setting method based on the damage identification algorithm.

2. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 1 is characterized in that: In step 1, the following steps are also included: Step 11, constructing the unmanned boat manipulation and motion feature mapping; Step 12: An outlier detection algorithm based on navigation data is used to ensure data validity; Step 12 also includes: Step 121: Screening based on Z score; Step 122: Screen again based on the local abnormal factors.

3. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 2 is characterized in that: In step 11, the manipulation motion mapping relationship is: a→P a is the acceleration at the current moment, P is the propulsion power, F is the three-degree-of-freedom force and torque matrix, m is the mass of the unmanned boat, and v is the speed of the unmanned boat; Based on the unmanned ship with dual propeller differential drive, the decomposition mapping of acceleration on the inertial system of the body can be written as: P2:(a x ,a y )→(T l ,T r ) Among them, a x , a y are the accelerations of the unmanned boat perpendicular to the bow direction and parallel to the bow direction, T l , T r These are the left and right propeller thrust throttles respectively.

4. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 3 is characterized in that: In step 121, the Z-score is expressed as: The expected value of any variable X is E(X) and the standard deviation is σ(X); The sample size is selected as n, and the upper limit of the absolute value of the Z score of a single data point is taken as σ, that is, if each data point does not constitute a significant difference with other values ​​within t seconds before and after it, it is recorded as valid data.

5. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 3 is characterized in that: In step 122, the selection is performed again based on the local abnormal factors: Given a parameter p, the distance d from data point p to data point o is p k (p, i) is expressed as: d k (p,i)=max{d k (i),d(p,i)} o is the point closest to p, d k (o) is the distance between i and the kth closest point to i, and d(p, i) is the distance between p and o; The local reachable density of data point p is defined as: N k (p) is all distances p less than d k The points of (p) constitute the k nearest neighbors of p; The local outlier factor LOF of a data point p is the ratio of the average local reachable density of points in the neighborhood of point p to the local reachable density of data point p: LOF k (p) local outlier factor represents the neighborhood N of point p k (p) is the average of the ratios of the local reachability density of other points within the point to the local reachability density of point p; If the ratio is closer to 1, it means that the density of point p's neighborhood points is similar, and point p may belong to the same cluster as its neighborhood points; if the ratio is less than 1, it means that the density of point p is higher than the density of its neighborhood points, and point p is a dense point; if the ratio is greater than 1, it means that the density of point p is lower than the density of its neighborhood points, and point p may be an outlier.

6. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 1 is characterized in that: In step 2, the following steps are also included: Step 21, defining the damage degree; The absolute propeller damage is defined as: Among them, Δ X is the absolute damage degree of the propeller with higher power loss, Δ Y is the absolute damage degree of the propeller with lower power loss, Δ X >Δ Y ; The throttle at the propeller control end with higher power loss is T X , the throttle at the propeller control end on the other side is T Y The equivalent propulsion throttles of the actual motion of the two propellers are Damage is defined as: D=D X -D Y Δ is the degree of damage; Step 22, establishing power damage state recognition through deep neural network.

7. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 6 is characterized in that: In step 22, the following steps are also included: Step 221: The speed, acceleration, left and right throttles, and the time corresponding to each set of data and the valid navigation information are filtered to form input feature values; Step 222: Perform activation function transformation on the hidden variable as the input of the next fully connected layer; its expression is: ReLU(x)=max(x,0) ReLU(x) is the ReLU function, x is the input feature value of the activation function, which is the filtered valid navigation information; Step 223: The fully connected layer is composed of multiple hidden layers with extended widths, and the hidden layer output is transformed by an activation function; Step 224: The output layer uses the Huber function as the loss function for the regression problem. The Huber error is defined as follows: Among them, y is the true value, f(x) is the predicted value, and δ is a hyperparameter used to control the turning point of the error.

8. The method for identifying power damage state and fault-tolerant control of a double-propeller driven unmanned boat according to claim 1 is characterized in that: In step 3, the following steps are also included: Step 31: Calculate the initial power reduction; The difference between the calculated throttle value and the true value is calculated. When the damage Δ is 0, that is, when the ship's propulsion system is normal, the difference should be a curve close to 0. The damage time interval is [t0,t n ], n+1 groups of data are recorded in the interval, which are: Indicates t n Acceleration information recorded at all times, {a} represents the acceleration time series of the unmanned boat; Based on the deep neural network, a power damage state recognition algorithm is established. The heading and vertical acceleration are input, and the output is the equivalent propulsion throttle: Indicates the equivalent throttle of the left and right propellers calculated from the nth set of recorded acceleration information. It represents the time series of equivalent propulsion throttle of the propeller of the unmanned boat; In the time interval [t0,t n ], calculate the cumulative deviation of the boost throttle: T δX (n) represents the cumulative throttle deviation of the X propeller with higher power loss in the nth group of recorded information, T δY (n) indicates the cumulative throttle deviation of the Y propeller with the lower power loss in the nth group of recorded information. represents the equivalent throttle of the X propeller calculated from the i-th group of recorded acceleration information, T iX Indicates the throttle of the X propeller control end recorded at the time of the i-th group of information, represents the equivalent throttle of the Y propeller calculated from the i-th group of recorded acceleration information, T iY represents the Y propeller control end throttle recorded at the i-th group of information; And calculate to t n The absolute damage degree of the propeller at the moment: Δ X (n) represents the absolute damage degree of X propeller calculated from the nth set of recorded information, Δ Y (n) represents the absolute damage degree of propeller Y calculated from the nth set of recorded information; Then t n The damage level of the ship at this moment is: Δ(n)=Δ X (n)-D Y (n) Step 32: multiple iterations with the adjusted relative damage degree of the motion effect being less than a preset value as the termination condition.

Images