Cluster GNSS (Global Navigation Satellite System) positioning anomaly detection method based on direction finding information

Abstract

The embodiment of the invention provides a cluster GNSS (Global Navigation Satellite System) positioning anomaly detection method based on direction finding information, which belongs to the technical field of measurement and specifically comprises the following steps of: 1, decomposing a cluster system into a plurality of cluster unilateral systems; step 2, performing anomaly detection on each cluster unilateral system to obtain a detection result; 3, inputting the deception distance, the machine spacing, the GNSS measurement noise and the direction finding sensor measurement noise of the cluster unilateral system into the leakage alarm rate prediction model to obtain a predicted leakage alarm rate; 4, constructing a cluster system fault detection model by adopting a Bayesian network according to the cluster topological structure, the detection result and the prediction leakage alarm rate; and 5, according to the cluster system fault detection model, deducing the state of each machine in the cluster system by adopting maximum posterior probability estimation. According to the scheme of the invention, the method can detect the abnormal positioning condition of a part of agents of the cluster system encountering significant deception interference, and improves the detection efficiency, precision and adaptability.

Description

Technical Field

[0001] The disclosed embodiments relate to the field of measurement technology, and in particular to a cluster GNSS positioning anomaly detection method based on direction finding information. Background Art

[0002] The Global Navigation Satellite System (GNSS) is an air-based radio navigation and positioning system that can provide three-dimensional coordinates for equipment on the surface of the earth or in near-Earth space. As a low-cost, widely applicable, and well-positioned navigation and positioning method, GNSS is widely used in the navigation and positioning of unmanned equipment. However, due to the certain fragility of GNSS signals from space, they are susceptible to natural and human interference, among which human interference includes suppression interference and deceptive interference, especially signal deceptive interference, which will cause great harm to unmanned equipment. At present, research on GNSS anti-interference technology at home and abroad mainly focuses on the following three aspects: on-board anti-interference technology, signal feature recognition and detection technology, and anti-interference technology based on auxiliary information. On-board anti-interference technology mainly suppresses interference at the satellite level to improve the anti-interference ability of GNSS signals. Signal feature recognition and detection technology mainly identifies and suppresses interference signals by analyzing the characteristics of GNSS signals. Anti-interference technology based on auxiliary information mainly uses information from other sources to enhance the anti-interference ability of GNSS signals.

[0003] Although existing technologies can improve positioning accuracy through ranging information between vehicles or drones, they lack effective fault detection and positioning recovery mechanisms in the event of GPS signal failure or spoofing attacks, and most methods are only applicable to specific types of measurement information.

[0004] It can be seen that there is an urgent need for an efficient, accurate and adaptable cluster GNSS positioning anomaly detection method based on direction finding information. Summary of the invention

[0005] In view of this, an embodiment of the present disclosure provides a cluster GNSS positioning anomaly detection method based on direction finding information, which at least partially solves the problems of poor detection efficiency, accuracy and adaptability in the prior art.

[0006] The present disclosure provides a cluster GNSS positioning anomaly detection method based on direction finding information, including:

[0007] Step 1, decompose the cluster system into multiple cluster unilateral systems;

[0008] Step 2: Perform anomaly detection on each cluster unilateral system to obtain the detection result;

[0009] Step 3, inputting the deception distance, the distance between devices, the GNSS measurement noise and the direction finding sensor measurement noise of the cluster unilateral system into the missed alarm rate prediction model to obtain the predicted missed alarm rate;

[0010] Step 4: Based on the cluster topology, detection results and predicted missed alarm rate, a cluster system fault detection model is constructed using a Bayesian network;

[0011] Step 5: According to the cluster system fault detection model, the maximum a posteriori probability estimation is used to infer the status of each machine in the cluster system.

[0012] According to a specific implementation of the embodiment of the present disclosure, step 2 specifically includes:

[0013] Step 2.1, for each cluster unilateral system, construct the dynamic equation and measurement equation of the cluster unilateral system for obtaining GNSS measurement information and inter-machine measurement information;

[0014] Step 2.2, construct a parity matrix S according to the dynamic equation and the measurement equation, and multiply the parity matrix S with the measurement output of the cluster unilateral system to obtain a parity vector p(k);

[0015] Step 2.3, calculate the covariance matrix of the parity vector p(k) and standardize the parity vector p(k) based on it to obtain the standardized parity vector By detecting variables Whether it exceeds the limit, so as to determine whether there is an abnormality in the unilateral system of the cluster and form a detection result.

[0016] According to a specific implementation of the embodiment of the present disclosure, step 2.1 specifically includes:

[0017] Step 2.1.1, perform linearization processing on the Taylor expansion of the azimuth information in the cluster unilateral system;

[0018] Step 2.1.2, construct the dynamic equation of the cluster unilateral system according to the physical meaning of the cluster unilateral system;

[0019] Step 2.1.3, the sensor noise is calculated by the output of each sensor in normal state, and the measurement equation of the cluster unilateral system is constructed, wherein the sensor output includes the GNSS positioning output and the direction finding information output.

[0020] According to a specific implementation of the embodiment of the present disclosure, before step 3, the method further includes:

[0021] Step 3.1, simulating and designing different working conditions, wherein the working conditions include deception distance, agent spacing, GNSS measurement noise, and direction finding sensor measurement noise;

[0022] Step 3.2, for various working conditions, a preset number of Monte Carlo simulations are carried out to obtain the method false alarm rate under each working condition, and form a false alarm rate data set;

[0023] Step 3.3, the normalized data of the deception distance, the distance between devices, the GNSS measurement noise, and the direction finding sensor measurement noise data are used as the features of the training data of the missed alarm rate prediction model, and the missed alarm rate data set is used as the target value of the training data of the missed alarm rate prediction model;

[0024] Step 3.4, build a deep neural network, set the optimizer and initial learning rate, and select the loss function;

[0025] Step 3.5, select training data, set the number of iterations, and train to obtain the missed alarm rate prediction model.

[0026] According to a specific implementation of the embodiment of the present disclosure, step 4 specifically includes:

[0027] Step 4.1, designing a structure D of a cluster system fault detection model according to the cluster topology structure, wherein the structure D includes two layers of vertex layer T(V) and edge layer C(E);

[0028] Step 4.2, construct the parameter Θ of the cluster system fault detection model based on the detection results of each cluster unilateral system and the predicted missed alarm rate;

[0029] In step 4.3, the cluster system fault detection model is formed by combining the structure D and the parameter Θ.

[0030] According to a specific implementation of the embodiment of the present disclosure, step 5 specifically includes:

[0031] Step 5.1, calculate the joint probability distribution P(C(V), T(V)) by the cluster system fault detection model;

[0032] Step 5.2, using the idea of ​​variable elimination method, obtain the marginal distribution P(C(V)) by summing irrelevant variables;

[0033] Step 5.3, calculate the posterior probability P(T(V)|C(V)), and take the true state value of each vertex corresponding to the maximum posterior probability as the test state of each vertex.

[0034] The cluster GNSS positioning anomaly detection scheme based on direction finding information in the disclosed embodiment includes: step 1, decomposing the cluster system into multiple cluster unilateral systems; step 2, performing anomaly detection on each cluster unilateral system to obtain a detection result; step 3, inputting the deception distance, machine distance, GNSS measurement noise and direction finding sensor measurement noise of the cluster unilateral system into a missed alarm rate prediction model to obtain a predicted missed alarm rate; step 4, using a Bayesian network to construct a cluster system fault detection model based on the cluster topology structure, the detection results and the predicted missed alarm rate; step 5, using the maximum a posteriori probability estimation to infer the state of each machine in the cluster system based on the cluster system fault detection model.

[0035] The beneficial effects of the embodiments of the present disclosure are as follows: through the scheme of the present disclosure, based on the existence of a direction-finding sensor between two machines in a cluster, the dynamic equation and the measurement equation of the cluster unilateral system are constructed, and the parity space detection method is used to construct a parity vector for the cluster unilateral system to detect whether there is an abnormality in the cluster unilateral system; Monte Carlo simulation is used to obtain the missed alarm rate of the method under different working conditions, and this is used as a data set to train a missed alarm rate prediction model for predicting the missed alarm rate of the method; based on the cluster topology structure and the parity space detection results of each cluster unilateral system, a Bayesian network is used to construct a cluster system fault detection model, and the maximum a posteriori probability estimation is used to infer the state of each machine in the cluster, thereby improving the detection efficiency, accuracy and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 A schematic diagram of a flow chart of a cluster GNSS positioning anomaly detection method based on direction finding information provided in an embodiment of the present disclosure;

[0038] Figure 2 A schematic diagram of a specific implementation process of a cluster GNSS positioning anomaly detection method based on direction finding information provided by an embodiment of the present disclosure;

[0039] Figure 3 A schematic diagram of a multi-agent cluster system provided in an embodiment of the present disclosure;

[0040] Figure 4 A schematic diagram of a cluster unilateral system provided by an embodiment of the present disclosure;

[0041] Figure 5 A schematic diagram of a cluster system fault detection model provided by an embodiment of the present disclosure;

[0042] Figure 6 A schematic diagram of a twelve-machine cluster system provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0043] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0044] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0045] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0046] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0047] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0048] The disclosed embodiment provides a cluster GNSS positioning anomaly detection method based on direction finding information, which can be applied to the anomaly detection process of a part of the intelligent bodies in the cluster encountering deception and causing significant deviations in the GNSS positioning of the navigation communication scenario.

[0049] See also Figure 1 , is a flow chart of a cluster GNSS positioning anomaly detection method based on direction finding information provided by an embodiment of the present disclosure. Figure 1 and Figure 2 As shown, the method mainly comprises the following steps:

[0050] Step 1, decompose the cluster system into multiple cluster unilateral systems;

[0051] In specific implementation, we can build a cluster unilateral system based on the two agents in the cluster system that have a direction-finding relationship, and decompose the cluster system into multiple cluster unilateral systems. Figure 3 As shown, the cluster unilateral system is Figure 4 shown.

[0052] Step 2: Perform anomaly detection on each cluster unilateral system to obtain the detection result;

[0053] Based on the above embodiment, step 2 specifically includes:

[0054] Step 2.1, for each cluster unilateral system, construct the dynamic equation and measurement equation of the cluster unilateral system for obtaining GNSS measurement information and inter-machine measurement information;

[0055] Step 2.2, construct a parity matrix S according to the dynamic equation and the measurement equation, and multiply the parity matrix S with the measurement output of the cluster unilateral system to obtain a parity vector p(k);

[0056] Step 2.3, calculate the covariance matrix of the parity vector p(k) and standardize the parity vector p(k) based on it to obtain the standardized parity vector By detecting variables Whether it exceeds the limit, so as to determine whether there is an abnormality in the unilateral system of the cluster and form a detection result.

[0057] Furthermore, the step 2.1 specifically includes:

[0058] Step 2.1.1, perform linearization processing on the Taylor expansion of the azimuth information in the cluster unilateral system;

[0059] Step 2.1.2, construct the dynamic equation of the cluster unilateral system according to the physical meaning of the cluster unilateral system;

[0060] Step 2.1.3, the sensor noise is calculated by the output of each sensor in normal state, and the measurement equation of the cluster unilateral system is constructed, wherein the sensor output includes the GNSS positioning output and the direction finding information output.

[0061] In specific implementation, anomaly detection can be performed on each cluster unilateral system to obtain the detection results. The specific process is as follows:

[0062] 1) For each cluster unilateral system decomposed in step 1, construct the dynamic equation and measurement equation of the cluster unilateral system that can obtain GNSS measurement information and inter-machine measurement information;

[0063] A. Taylor expansion of the azimuth information in the cluster unilateral system, ignoring the high-order terms, and linearization;

[0064]

[0065] Where:

[0066]

[0067] B. Construct the system dynamic equation based on the physical meaning of the cluster unilateral system;

[0068]

[0069] In the formula, is the system state vector at the kth moment, x(k)=[x agtA (k),x agtB (k),y agtA (k),y agtB (k)] T ; is the system input at the kth moment; is the system output at the kth moment, A is the state matrix or system matrix, A = I 4×4 ; B is the input matrix or control matrix, B = I 4×4 ; C 5×4 is the output matrix, which represents the relationship between the system output and the system state variables.

[0070]

[0071] C. The sensor noise is calculated by the sensor data under normal conditions, and the measurement equation of the cluster unilateral system is constructed. The sensor output includes GNSS positioning output and direction finding information output.

[0072] z(k)=Cx(k)+v(k) (5)

[0073] In the formula, is the sensor output of the cluster unilateral system, The distribution obeyed by each sensor noise v(k) is obtained by calculating the normal data, and the covariance matrix is ​​H.

[0074] 2) Construct the parity matrix S based on the established mathematical model and calculate the parity vector p(k). The specific process is as follows:

[0075] A. Calculate the left zero matrix S of the system output matrix C, that is, S is a parity matrix, and the parity matrix satisfies SC = 0;

[0076] B. Multiply the parity matrix by the system measurement output to obtain the parity vector p(k).

[0077] p(k)=Sz(k) (6)

[0078] 3) Obtain the normalized parity vector By detecting variables Whether it exceeds the limit, so as to determine whether there is an abnormality in the unilateral system of the cluster. The specific process is as follows:

[0079] A. Calculate the covariance matrix P = SHS of p(k) T , normalize the parity vector p(k)

[0080]

[0081] Obtain a standard normal distribution

[0082] B. Variables Satisfying the chi-square distribution with 1 degree of freedom, given the confidence level 1-α, the detection threshold is obtained At this time, the false alarm rate P fa =α;

[0083] C. Judgment variables Whether the detection threshold is exceeded If the limit is exceeded, it means that there is an abnormality in the cluster unilateral system. Monte Carlo simulation is used to evaluate the performance of the cluster unilateral system detection method based on the odd-even space detection method. The average fault missed alarm rate results of a single intelligent agent are shown in Table 1, and the average fault missed alarm rate results of two intelligent agents are shown in Table 2.

[0084] Table 1

[0085]

[0086] Table 2

[0087]

[0088]

[0089] Step 3, inputting the deception distance, the distance between devices, the GNSS measurement noise and the direction finding sensor measurement noise of the cluster unilateral system into the missed alarm rate prediction model to obtain the predicted missed alarm rate;

[0090] Furthermore, before step 3, the method further includes:

[0091] Step 3.1, simulating and designing different working conditions, wherein the working conditions include deception distance, distance between devices, GNSS measurement noise, and direction finding sensor measurement noise;

[0092] Step 3.2, for various working conditions, a preset number of Monte Carlo simulations are carried out to obtain the method false alarm rate under each working condition, and form a false alarm rate data set;

[0093] Step 3.3, the normalized data of the deception distance, the distance between devices, the GNSS measurement noise, and the direction finding sensor measurement noise data are used as the features of the training data of the missed alarm rate prediction model, and the missed alarm rate data set is used as the target value of the training data of the missed alarm rate prediction model;

[0094] Step 3.4, build a deep neural network, set the optimizer and initial learning rate, and select the loss function;

[0095] Step 3.5, select training data, set the number of iterations, and train to obtain the missed alarm rate prediction model.

[0096] In specific implementation, before using the missed alarm rate prediction model to obtain the predicted missed alarm rate, the missed alarm rate prediction model needs to be trained. The training process is as follows:

[0097] 1) Design different simulation conditions and use Monte Carlo simulation to obtain the false alarm rate data set of this method;

[0098] A. Design different simulation conditions with different levels of deception distance, agent spacing, GNSS measurement noise, and direction-finding sensor measurement noise

[0099] B. For each of the designed working conditions, 100,000 Monte Carlo simulations are carried out to obtain the method false alarm rate under each working condition and form a false alarm rate data set.

[0100] 2) The deception distance, the distance between the two vehicles, the GNSS measurement noise and the direction-finding sensor measurement noise are normalized and used as the features of the training data of the missed alarm rate prediction model, and the missed alarm rate data is used as the target value of the training data;

[0101] The deception distance, inter-machine distance, GNSS measurement noise and direction-finding sensor measurement noise data are preprocessed and normalized using decimal calibration.

[0102]

[0103] The k values ​​for the deception distance, inter-machine distance, GNSS measurement noise, and direction-finding sensor measurement noise data are taken as 3, 3, 1, and -1, respectively, to eliminate the dimensional differences between different features. The normalized data are used as the features of the training data for the missed alarm rate prediction model.

[0104] B. The missed alarm rate data is used as the target value of the missed alarm rate prediction model training data.

[0105] 3) Construct a deep neural network and train it to obtain a missed alarm rate prediction model;

[0106] A. Build a deep neural network, use the Adam optimizer, set the initial learning rate to 0.001, and use the mean squared error (MSE) as the loss function. The MSE loss function is simple and easy to calculate, and is suitable for regression problems.

[0107] B. Select training data and set the number of iterations to 100 to ensure that the model fully learns the characteristics of the data without overfitting, and train to obtain a missed alarm rate prediction model.

[0108] After the prediction missed alarm rate model is trained, the deception distance, inter-machine distance, GNSS measurement noise and direction-finding sensor measurement noise of the cluster unilateral system can be input into the missed alarm rate prediction model to obtain the predicted missed alarm rate.

[0109] Step 4: Based on the cluster topology, detection results and predicted missed alarm rate, a cluster system fault detection model is constructed using a Bayesian network;

[0110] Furthermore, the step 4 specifically includes:

[0111] Step 4.1, designing a structure D of a cluster system fault detection model according to the cluster topology structure, wherein the structure D includes two layers of vertex layer T(V) and edge layer C(E);

[0112] Step 4.2, construct the parameter Θ of the cluster system fault detection model based on the detection results of each cluster unilateral system and the predicted missed alarm rate;

[0113] In step 4.3, the cluster system fault detection model is formed by combining the structure D and the parameter Θ.

[0114] In specific implementation, the cluster system fault detection model B can be constructed based on the cluster topology, detection results and predicted missed alarm rate using Bayesian network.<D,Θ> , the specific process can be as follows:

[0115] A. The structure of the cluster system fault detection model designed based on the cluster topology structure D, such as Figure 5 As shown in the figure, the structure consists of two layers: vertex layer T(V) and edge layer C(E). The vertex layer T(V) contains the true state attributes T(v) of each vertex. i ), which represents the real state of each mobile agent in the cluster system. The edge layer C(E) contains each edge check state C(v i ,v j ), indicating the verification status of each cluster unilateral system in the cluster system. The nodes in the vertex layer T(V) point to the nodes in the edge layer C(E) that have a dependency relationship with it. ;

[0116] B. The parameter Θ of the cluster system fault detection model is constructed based on the parity space detection results of each cluster unilateral system and the missed alarm rate predicted by the missed alarm rate prediction model. Since the GNSS information of each mobile agent in the cluster may be deceived, it can be considered that the probability of the GNSS information of each mobile agent being deceived and not being deceived is equal. Therefore, the node T(v) in the vertex layer T(V) is given. i )'s prior probability table As shown in Table 3, according to the missed alarm rate predicted by the missed alarm rate prediction model, the edge layer node C(e k ) As shown in Table 4.

[0117] Table 3

[0118]

[0119] Table 4

[0120]

[0121] Step 5: According to the cluster system fault detection model, the maximum a posteriori probability estimation is used to infer the status of each machine in the cluster system.

[0122] Based on the above embodiment, step 5 specifically includes:

[0123] Step 5.1, calculate the joint probability distribution P(C(V), T(V)) by the cluster system fault detection model;

[0124] Step 5.2, using the idea of ​​variable elimination method, obtain the marginal distribution P(C(V)) by summing irrelevant variables;

[0125] Step 5.3, calculate the posterior probability P(T(V)|C(V)), and take the true state value of each vertex corresponding to the maximum posterior probability as the test state of each vertex.

[0126] In specific implementation, the specific process of using maximum a posteriori probability estimation to infer the status of each machine in the cluster is as follows:

[0127] A. Calculate the joint probability distribution P(C(V), T(V)) by the cluster system fault detection model;

[0128]

[0129] B. Using the idea of ​​variable elimination, the marginal distribution P(C(V)) is obtained by summing irrelevant variables;

[0130]

[0131] C. Calculate the posterior probability P(T(V)|C(V)). The true state value of each vertex corresponding to the maximum posterior probability is in Get the test status of each vertex

[0132] D. For example Figure 6 The twelve multi-agent cluster systems shown in the figure are used to simulate the cluster system network vertex fault detection method, and the detection results are shown in Table 5.

[0133] Table 5

[0134]

[0135] The cluster GNSS positioning anomaly detection method based on direction finding information provided in this embodiment makes full use of the existing direction finding sensors in the cluster system, adopts the odd-even space detection method, and proposes a cluster unilateral system anomaly detection method. The method has high applicability and low cost, and the detection is fast and accurate. The reason is that in order to realize cluster tasks, most cluster systems need to carry inter-machine measurement sensors to obtain inter-machine measurement information. Therefore, the proposed method does not need to add additional sensors for the cluster system, so the use of this method has high applicability and low cost; at the same time, a missed alarm rate prediction model is trained, and while ensuring the accuracy of the detection results, the missed alarm rate is also relatively fast; taking into full consideration the cluster topology structure and the abnormal detection results of each cluster unilateral system, a Bayesian network is used to establish a cluster system fault detection model, which can effectively infer the true state of each intelligent body in the cluster system. Compared with the commonly used voting election method, this method has higher interpretability.

[0136] It should be understood that various parts of the present disclosure may be implemented in hardware, software, firmware or a combination thereof.

[0137] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A cluster GNSS positioning anomaly detection method based on direction finding information, characterized in that: include: Step 1, decompose the cluster system into multiple cluster unilateral systems; Step 2: Perform anomaly detection on each cluster unilateral system to obtain the detection result; Step 3, inputting the deception distance, the distance between devices, the GNSS measurement noise and the direction finding sensor measurement noise of the cluster unilateral system into the missed alarm rate prediction model to obtain the predicted missed alarm rate; Step 4: Based on the cluster topology, detection results and predicted missed alarm rate, a cluster system fault detection model is constructed using a Bayesian network; Step 5: According to the cluster system fault detection model, the maximum a posteriori probability estimation is used to infer the status of each machine in the cluster system.

2. The method according to claim 1, characterized in that The step 2 specifically includes: Step 2.1, for each cluster unilateral system, construct the dynamic equation and measurement equation of the cluster unilateral system for obtaining GNSS measurement information and inter-machine measurement information; Step 2.2, construct a parity matrix S according to the dynamic equation and the measurement equation, and multiply the parity matrix S with the measurement output of the cluster unilateral system to obtain a parity vector p(k); Step 2.3, calculate the covariance matrix of the parity vector p(k) and standardize the parity vector p(k) based on it to obtain the standardized parity vector By detecting variables Whether it exceeds the limit, so as to determine whether there is an abnormality in the unilateral system of the cluster and form a detection result.

3. The method according to claim 2, characterized in that The step 2.1 specifically includes: Step 2.1.1, perform linearization processing on the Taylor expansion of the azimuth information in the cluster unilateral system; Step 2.1.2, construct the dynamic equation of the cluster unilateral system according to the physical meaning of the cluster unilateral system; Step 2.1.3, the sensor noise is calculated by the output of each sensor in normal state, and the measurement equation of the cluster unilateral system is constructed, wherein the sensor output includes the GNSS positioning output and the direction finding information output.

4. The method according to claim 3, characterized in that Before step 3, the method further includes: Step 3.1, simulating and designing different working conditions, wherein the working conditions include deception distance, agent spacing, GNSS measurement noise, and direction finding sensor measurement noise; Step 3.2, for various working conditions, a preset number of Monte Carlo simulations are carried out to obtain the method false alarm rate under each working condition, and form a false alarm rate data set; Step 3.3, the normalized data of the deception distance, the distance between devices, the GNSS measurement noise, and the direction finding sensor measurement noise data are used as the features of the training data of the missed alarm rate prediction model, and the missed alarm rate data set is used as the target value of the training data of the missed alarm rate prediction model; Step 3.4, build a deep neural network, set the optimizer and initial learning rate, and select the loss function; Step 3.5, select training data, set the number of iterations, and train to obtain the missed alarm rate prediction model.

5. The method according to claim 4, characterized in that The step 4 specifically includes: Step 4.1, designing a structure D of a cluster system fault detection model according to the cluster topology structure, wherein the structure D includes two layers of vertex layer T(V) and edge layer C(E); Step 4.2, construct the parameter Θ of the cluster system fault detection model based on the detection results of each cluster unilateral system and the predicted missed alarm rate; In step 4.3, the cluster system fault detection model is formed by combining the structure D and the parameter Θ.

6. The method according to claim 5, characterized in that The step 5 specifically includes: Step 5.1, calculate the joint probability distribution P(C(V), T(V)) by the cluster system fault detection model; Step 5.2, using the idea of ​​variable elimination method, obtain the marginal distribution P(C(V)) by summing irrelevant variables; Step 5.3, calculate the posterior probability P(T(V)|C(V)), and take the true state value of each vertex corresponding to the maximum posterior probability as the test state of each vertex.

Images