Anti-interference method based on residual chi-square and sequential probability ratio test joint decision

Abstract

The invention discloses an anti-interference method based on residual chi-square and sequential probability ratio test joint decision, relates to the technical field of navigation anti-interference, and has the advantages of lower calculation amount and better effect on mutation type interference and slow deformation type interference. For an SINS / GNSS integrated navigation system composed of a strapdown inertial navigation system (SINS) and a global navigation satellite system (GNSS), the method comprises the following steps: firstly, establishing an SINS / GNSS integrated navigation system model; and establishing a Kalman filter of the SINS / GNSS integrated navigation system. And performing residual chi-square test, and outputting an interference detection result 1. And performing sequential probability ratio test, and outputting an interference detection result 2. When the interference detection result I and the interference detection result II do not detect interference, judging that the integrated navigation system is not interfered; and if one of the two detects interference, judging that the integrated navigation system is interfered. After the interference in the integrated navigation system is detected, the measurement information is isolated, and the interference information is prevented from polluting the Kalman filter.

Description

Technical Field

[0001] The invention relates to the technical field of navigation anti-interference, and in particular to an anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making. Background Art

[0002] Inertial Navigation System (INS) is completely autonomous and is not disturbed by the external environment, but the navigation error accumulates over time. Inertial navigation system is divided into two types: platform type and strapdown type. Since the strapdown inertial navigation system (SINS) uses a mathematical simulation platform to replace the physical platform, it has the characteristics of small size, light weight, simple structure and strong reliability, so the present invention uses SINS later. Global Navigation Satellite System (GNSS) can provide users with accurate, continuous, all-weather position, speed and time information, but GNSS signals are easily disturbed. In the case of severe obstruction, the quality of GNSS signals is seriously reduced due to factors such as multipath effect and non-line-of-sight error. Therefore, usually two navigation sensors, SINS and GNSS, are combined to obtain continuous, stable and robust navigation results.

[0003] Unfortunately, GNSS signals may be interfered with during use. Interference to GNSS signals is mainly divided into suppression interference and deception interference. Suppression interference mainly affects the normal reception of real signals by receiving devices by transmitting high-power noise interference signals. At present, it can be effectively suppressed at the receiver level through anti-interference technologies such as array antennas. As for deception interference, it is divided into forwarding deception interference and generative deception interference. Since these malicious deception attack interferences are usually purposeful and may present a variety of error characteristics, including the time of deception attack, the amplitude of deception, and the change of deception distance over time, these factors increase the complexity and difficulty of effectively detecting deception signals. Deception interference can be achieved through fault detection, which can effectively reduce the complexity of the algorithm compared to the method of improving the structure of satellite receivers.

[0004] Common fault detection algorithms include the chi-square test algorithm and the sequential probability ratio test algorithm. The chi-square test is a method for judging the degree of deviation between the actual observed value and the theoretical value of a statistical sample. The greater the deviation between the actual observed value and the theoretical value, the greater the chi-square value. The chi-square test includes the residual chi-square test algorithm and the state chi-square test algorithm. The state chi-square test algorithm does not miss detection during the alarm period, but its calculation amount is large, the delay is high, and the sensitivity decreases as the filtering proceeds; while the residual chi-square test method has a small delay and is more sensitive to measurement faults. It can directly use the calculation results of the Kalman filter, so the calculation amount is low. Therefore, the residual chi-square test is more sensitive to hard faults, but has a higher delay on soft faults; the sequential probability ratio test (SPRT) has a high sensitivity to the slowly growing soft faults in the sensor, but this method has the defects of slow detection speed in the early stage of the fault and inability to detect the end of the fault.

[0005] Therefore, how to achieve good anti-interference effects on both sudden interference and slowly varying interference on the basis of lower computational complexity is a problem that needs to be solved urgently. Summary of the invention

[0006] In view of this, the present invention provides an anti-interference method based on the joint decision-making of residual chi-square and sequential probability ratio test, which can combine the residual chi-square test and the sequential probability ratio test to propose a joint discrimination method to detect interference in GNSS signals, with lower calculation amount and better effects on both sudden interference and slowly varying interference.

[0007] To achieve the above object, the technical solution of the present invention - an anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making, for a SINS / GNSS integrated navigation system composed of a strapdown inertial navigation system SINS and a global navigation satellite system GNSS, adopts the following steps to perform anti-interference navigation:

[0008] Step 1: Establish the SINS / GNSS integrated navigation system model.

[0009] Step 2: Establish the Kalman filter of the SINS / GNSS integrated navigation system.

[0010] Step 3: Perform a residual chi-square test and output interference detection result 1.

[0011] Step 4: Perform a sequential probability ratio test and output interference detection result 2.

[0012] Step 5: Combine the detection results in step 3 and step 4 to obtain the final interference detection result. That is, when interference detection results 1 and 2 do not detect interference, it is determined that the SINS / GNSS integrated navigation system is not interfered with; if one of the interference detection results 1 and 2 detects interference, it is determined that the SINS / GNSS integrated navigation system is interfered with.

[0013] After detecting interference in the SINS / GNSS integrated navigation system, the measurement information is isolated to prevent the interfered information from contaminating the Kalman filter.

[0014] Furthermore, step 1: establish a SINS / GNSS integrated navigation system model; specifically:

[0015] The model of the integrated navigation system is established based on the inertial navigation error equation, and the states of the integrated navigation system are selected as follows:

[0016] X=[φ E φ N φ U δv E δv N δv U δLδλδhε gE ε gN ε gU ε aE ε aN ε aU ] T

[0017] Among them, φ E ,φ N ,φ U are the SINS attitude error angles in the northeast direction; δv E ,δv N ,δv U are the SINS velocity errors in the east, north and celestial directions respectively; δL, δλ and δh are the latitude, longitude and altitude errors of SINS respectively; ε gE , ε gN , ε gU is the zero bias of the gyro's three axes in the northeast sky, ε aE , ε aN , ε aU is the zero bias of the three axes of the northeast sky of the accelerometer; the state equation is as follows:

[0018]

[0019] X k , X k-1 are the states of the integrated navigation system at time k and k-1 respectively, and F k / k-1 is the state transfer matrix from time k-1 to time k, Γ k-1is the noise driving matrix at time k-1, is the k-1 system noise matrix, and the superscript b represents the carrier coordinate system.

[0020] F k / k-1 The discrete formula of is as follows, where F(k) is the state matrix.

[0021] F k / k-1 ≈I 15 +F(k)·△t

[0022] I 15 is a 15-dimensional unit matrix, and △t is the filtering period.

[0023] The system noise matrix and noise driving matrix are:

[0024]

[0025] in is the angular random walk in the gyro output, is the velocity random walk in the accelerometer output, w gE 、w gN and w gU are the angular random walks in the east, north and sky directions respectively, w aE 、w aN and w aU They are the random walks of the speed in the east, north and sky directions respectively; is Γ k-1 For the part in.

[0026] The measurement equation is:

[0027] Z k =H k X k +V k

[0028] Among them, V k is the GNSS position measurement noise, H k is the measurement matrix.

[0029] Further, step 2: establish the Kalman filter of the SINS / GNSS integrated navigation system, specifically:

[0030] Time update:

[0031]

[0032] in is the estimated value of the state from time k-1 to time k, F k / k-1 is the state transfer matrix from time k-1 to time k, is the estimated value of the state at time k-1; Pk / k-1 is the state covariance matrix from time k-1 to time k, P k-1 is the state covariance matrix at time k-1, G k-1 is the system noise distribution matrix at time k-1, Q k-1 is the system noise covariance matrix at time k-1.

[0033] Measurement Update:

[0034]

[0035] P k =(IK k H k ) k / k-1

[0036] Where K k is the gain matrix at time k; H k is the measurement matrix at time k; R k is the measurement noise covariance matrix at time k; is the estimated value of the state at time k; Z k is the measurement value at time k.

[0037] Further, step three: perform a residual chi-square test and output interference detection result 1, specifically:

[0038] Get the Kalman filter measurement residual r k :

[0039]

[0040] where z k To measure information; is the estimated value of the state from time k-1 to time k.

[0041] According to the Kalman filter related information, calculate the fault detection function λ k Values:

[0042]

[0043] in, Comparison of fault detection function A k and threshold T D Size, output interference detection result 1, fault judgment criteria are as follows:

[0044]

[0045] Further, step 4: perform a sequential probability ratio test and output interference detection result 2, specifically:

[0046] Get the Kalman filter measurement residual vk information:

[0047]

[0048] z k is the measurement information; H k is the measurement matrix; is the estimated value of the state from time k-1 to time k.

[0049] Calculate the mean information of the measurement residuals

[0050]

[0051] Where α is the fading factor, which is set to 0.95; v k-1 is the residual at time k-1.

[0052] Calculate the statistical information △λ based on the mean information of the measurement residual:

[0053]

[0054] λ k =λ k-1 +△λ;

[0055] Where P k / k-1 is the state covariance matrix from time k-1 to time k.

[0056] Compare the statistical information and the threshold value to obtain the second detection result; the fault judgment criteria are as follows:

[0057]

[0058] Where P m is the false alarm rate, P f is the false alarm rate.

[0059] Beneficial effects:

[0060] 1: In order to detect the deceptive interference information in the GNSS signal, the present invention proposes a method for detecting the deceptive interference signal in the GNSS output by means of fault detection. The present invention analyzes the defects of SPRT in fault detection and proposes a fading SPRT method. By introducing a fading factor, the influence of historical information on the statistical change rate and statistics at the time of the fault is reduced, and the sensitivity of SPRT to the fault is enhanced, thereby reducing the fault detection delay and solving the problem that SPRT cannot detect the end of the fault. The present invention combines the residual chi-square test and the sequential probability ratio test to propose a joint discrimination method to detect interference in GNSS signals, which has a lower computational complexity and has better effects on both sudden and slow-changing interference. . BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is the trajectory information, which displays the carrier's attitude, speed, position and other information.

[0062] Figure 2 This is the integrated navigation error diagram when the system is normal, showing the attitude, velocity and position errors of the integrated navigation system when there is no interference signal in the GNSS signal.

[0063] Figure 3 This is the error diagram of the integrated navigation system, which shows the attitude, velocity and position errors output by the integrated navigation system when there is a slowly varying interference signal in the GNSS signal.

[0064] Figure 4 This is the error diagram of the integrated navigation system, which shows the attitude, velocity and position errors output by the integrated navigation system when there is a sudden interference signal in the GNSS signal.

[0065] Figure 5 It is an error diagram of the integrated navigation system, showing the attitude, velocity and position errors output by the integrated navigation system after the anti-interference algorithm mentioned in the present invention is used to detect and isolate the interfered information when there is a slowly varying interference signal in the GNSS signal.

[0066] Figure 6 It is an error diagram of the integrated navigation system, showing the attitude, velocity and position errors output by the integrated navigation system after the anti-interference algorithm mentioned in the present invention is used to detect and isolate the interfered information when there is a sudden interference signal in the GNSS signal;

[0067] Figure 7 A flow chart of the anti-interference method for joint decision-making based on residual chi-square and sequential probability ratio test provided by the present invention. DETAILED DESCRIPTION

[0068] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0069] The present invention provides an anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making, for the SINS / GNSS integrated navigation system composed of a strapdown inertial navigation system SINS and a global navigation satellite system GNSS, the specific process of the present invention is as follows Figure 7 As shown, the following steps are used for anti-interference navigation:

[0070] Step 1: Establish a SINS / GNSS integrated navigation system model; specifically:

[0071] The model of the integrated navigation system is established based on the inertial navigation error equation, and the states of the integrated navigation system are selected as follows:

[0072] X=[φ E φ N φ U δv E δv N δv U δLδλδhε gE ε gN ε gU ε aE ε aN ε aU ] T

[0073] Among them, φ E ,φ N ,φ U are the SINS attitude error angles in the northeast direction; δv E ,δv N ,δv U are the SINS velocity errors in the east, north and celestial directions respectively; δL, δλ and δh are the latitude, longitude and altitude errors of SINS respectively; ε gE , ε gN , ε gU is the zero bias of the gyro's three axes in the northeast sky, ε aE , ε aN , ε aU is the zero bias of the three axes of the northeast sky of the accelerometer; the state equation is as follows:

[0074]

[0075] X k , X k-1 are the states of the integrated navigation system at time k and k-1 respectively, and F k / k-1 is the state transfer matrix from time k-1 to time k, Γ k-1 is the noise driving matrix at time k-1, is the k-1 system noise matrix, and the superscript b represents the carrier coordinate system;

[0076] F k / k-1 The discrete formula of is as follows, where F(k) is the state matrix;

[0077] F k / k-1 ≈I 15 +F(k)·△t

[0078] I 15 is a 15-dimensional unit matrix, and △t is the filtering period.

[0079] The system noise matrix and noise driving matrix are:

[0080]

[0081] in is the angular random walk in the gyro output, is the velocity random walk in the accelerometer output, w gE 、w gN and w gU are the angular random walks in the east, north and sky directions respectively, w aE 、w aN and w aU They are the random walks of the speed in the east, north and sky directions respectively; is Γ k-1 For the part in;

[0082] The measurement equation is:

[0083] Z k =H k X k +V k

[0084] Among them, V k is the GNSS position measurement noise, H k is the measurement matrix.

[0085] Step 2: Establish the Kalman filter of the SINS / GNSS integrated navigation system; specifically:

[0086] Time update:

[0087]

[0088] in is the estimated value of the state from time k-1 to time k, F k / k-1 is the state transfer matrix from time k-1 to time k, is the estimated value of the state at time k-1; P k / k-1 is the state covariance matrix from time k-1 to time k, P k-1 is the state covariance matrix at time k-1, G k-1 is the system noise distribution matrix at time k-1, Q k-1 is the system noise covariance matrix at time k-1;

[0089] Measurement Update:

[0090]

[0091] P k =(IK k H k ) k / k-1

[0092] Where K k is the gain matrix at time k; H kis the measurement matrix at time k; R k is the measurement noise covariance matrix at time k; is the estimated value of the state at time k; Z k is the measurement value at time k.

[0093] Step 3: Perform a residual chi-square test and output interference detection result 1; specifically:

[0094] Get the Kalman filter measurement residual r k :

[0095]

[0096] where z k To measure information; is the estimated value of the state from time k-1 to time k;

[0097] According to the Kalman filter related information, calculate the fault detection function λ k Values:

[0098]

[0099] in, Comparison of fault detection function A k and threshold T D Size, output interference detection result 1, fault judgment criteria are as follows:

[0100]

[0101] Step 4: Perform a sequential probability ratio test and output interference detection result 2; specifically:

[0102] Get the Kalman filter measurement residual v k information:

[0103]

[0104] z k is the measurement information; H k is the measurement matrix; is the estimated value of the state from time k-1 to time k;

[0105] Calculate the mean information of the measurement residuals

[0106]

[0107] α is the forgetting factor, v k-1 is the residual at time k-1;

[0108] Calculate the statistical information △λ based on the mean information of the measurement residual:

[0109]

[0110] P k / k-1 is the state covariance matrix from time k-1 to time k;

[0111] Compare the statistical information and the threshold value to obtain the second detection result; the fault judgment criteria are as follows:

[0112]

[0113] P m is the false alarm rate, P f is the false alarm rate.

[0114] Step 5: Combining the detection results in step 3 and step 4, the final interference detection result is obtained; that is, when the interference detection results 1 and 2 do not detect a fault, it is determined that the SINS / GNSS integrated navigation system is not interfered with; if one of the interference detection results 1 and 2 detects a fault, it is determined that the SINS / GNSS integrated navigation system is interfered with;

[0115] After detecting interference in the SINS / GNSS integrated navigation system, the measurement information is isolated to prevent the interfered information from contaminating the Kalman filter.

[0116] Example 1

[0117] In order to verify the effectiveness of the present invention, the method of the present invention was simulated by Matlab programming. Some functions in the Matlab-based open source toolbox PINS developed by Professor Yan Gongmin of Northwestern Polytechnical University were used in the simulation process.

[0118] Step 1: The SINS attitude error equation is used to describe the misalignment error angle φ between the ideal navigation coordinate system and the calculated navigation coordinate system. The differential equations corresponding to the ideal attitude matrix and the calculated attitude matrix are as follows:

[0119]

[0120] in, represents the projection of the angular velocity of system b relative to system i on system b, represents the projection of the angular velocity of system n relative to system i on system n, and Indicates the error term and

[0121] According to the above formula, the SINS attitude error equation is obtained by ignoring high-order small quantities and simplifying by moving terms as follows:

[0122]

[0123] Similarly, the ideal specific force equation and the actual specific force equation considering parameter errors are as follows:

[0124]

[0125] In the formula, It represents the projection of the angular velocity of the e system relative to the i system on the n system. represents the projection of the angular velocity of the n system relative to the e system on the n system, represents the projection of the velocity of the n system relative to the e system in the n system, is the speed obtained by the navigation computer, Represents the specific force information output by the accelerometer, g n Represents the acceleration of gravity. According to the above equation, ignoring the second-order small quantity related to the velocity error, the SINS velocity error equation is:

[0126]

[0127] Finally, consider the SINS position update equations are and are the differential equations for longitude, latitude, and altitude, respectively. M and R N are the meridian radius and meridian radius of a point on the earth’s surface, L, λ and h are the longitude, latitude and altitude, respectively. v = [v N v E v U ] T The vectors correspond to the velocity vectors in the east, north, and sky directions respectively. By calculating the deviation of the position update equation, the position error equation can be obtained as follows:

[0128]

[0129] in, and

[0130] The above equations together constitute the first 9-dimensional state quantities of the system. The remaining 6-dimensional state quantities are the 3-dimensional gyro random constant drift and 3D accelerometer random constant drift Finally, the state equation of the system is obtained after discretization according to the above equation:

[0131] X=[φ E φ N φ U δv E δv N δv U δLδλδhε gE ε gN εgU ε aE ε aN ε aU ] T

[0132]

[0133] Among them, the system noise matrix and noise driving matrix are as follows:

[0134]

[0135] In the formula, represents the gyroscope angular velocity measurement white noise, represents the accelerometer measuring white noise.

[0136] The one-step transfer matrix of the system error and its discretization formula are as follows:

[0137]

[0138] F k / k-1 ≈I 15 +F(t)·△t

[0139] Among them, △t represents the filtering period, M ij (i=a,v,p;j=a,v,p) are the coefficient matrices obtained after simplifying the error model. For their specific meanings, please refer to relevant textbooks and will not be repeated here.

[0140] The system measurement equation consists of a 3D position error, where represents the position information of the SINS and the position information of the GNSS. The measurement equation is in the following form:

[0141]

[0142] in, and Represent the position information of SINS and GNSS respectively. The measurement equation is:

[0143] Z k =H k X k +V k

[0144] Where V k Represents the white noise of satellite receiver position measurement. So far, the SINS / GNSS integrated navigation system model has been established.

[0145] Step 2: Establish the Kalman filter of the SINS / GNSS integrated navigation system, determine the Kalman filter period according to the system characteristics, and perform time update before the measurement information arrives. The time update formula is as follows:

[0146]

[0147] Measurement Update:

[0148]

[0149] P k =(IK k H k ) k / k-1

[0150] At this point, the construction of the Kalman filter of the integrated navigation system has been completed.

[0151] Step 3: Get the residual of the Kalman filter:

[0152]

[0153] In the formula, the predicted value is:

[0154]

[0155] When the system is working normally, the residual is zero-mean Gaussian white noise with a variance of:

[0156] A(k)=H(k)P(k / k-1)H T (k)+R(k)

[0157] When a system fails, the residual mean is no longer equal to 0. Therefore, the residual detection result can be used to determine whether the system fails. The following binary assumptions are made for the residual r(k):

[0158] H0: No fault E{r(k)}=0, E{r(k)r T (k)}=A

[0159] H1: Faulty E{r(k)}=μ,E{(r(k)-μ)(r(k)-μ) T =A

[0160] Its fault detection function is:

[0161] λ(k)=r T (k)A -1 (k)r(k)

[0162] λ(k) obeys the chi-square distribution with m degrees of freedom, where m is the dimension of the measurement information. The fault judgment criterion is:

[0163]

[0164] Where, T DIndicates the fault threshold value, according to the false alarm rate P f Finally, the interference detection result is determined. Step 4: The measurement equation of the filter can be expressed as:

[0165] Z=HX+V

[0166] When the system is fault-free, the new information sequences in the filter are independent of each other and obey the normal distribution N(0,Σ). At this time, the new information expectation and variance can be expressed as:

[0167]

[0168] When the system fails, the expected and variance of the new information are:

[0169]

[0170] According to the above formula, the following hypothesis test can be set:

[0171]

[0172] And we know that the probability density of the new information sample under the two assumptions is

[0173]

[0174] in,

[0175]

[0176] Among them, according to the probability density functions of the two types of hypotheses, the likelihood ratio under the two types of hypotheses can be calculated as:

[0177]

[0178] Assume that the rate of change of the statistic at each moment can be expressed as:

[0179]

[0180] The SPRT fault diagnosis test function is:

[0181]

[0182] The above SPRT test method has a high sensitivity for detecting slowly increasing faults, but it still has the defects of delay and failure to detect the end of the fault. To solve this problem, the above formula should be analyzed. When the measurement system works normally for a long time, the average value of the measurement new information The value of the fault detection function approaches 0, which causes the change in the fault detection function to approach 0. This means that the statistic needs a long time to detect the start of the fault, resulting in a long period of missed alarms. In the event of a system failure, when the fault ends, the measurement system has also returned to normal. In this case, it is difficult to reduce the fault detection function below the threshold. In order to solve the above defects, a fading factor is introduced when calculating the mean and statistics of the new information. The historical statistics and historical new information in the SPRT are gradually eliminated to reduce the impact of historical information at the time of the fault and increase the sensitivity of the statistics to the fault.

[0183] In order to enhance the rate of change of the mean at the time of fault occurrence and reduce the detection delay at the time of fault occurrence, the fading factor α can be introduced when calculating the mean to reduce the weight of the new information at the historical moment in the calculation, thereby enhancing the influence of the new information sample at the current moment. The improved formula is as follows:

[0184]

[0185] From the above formula, we know that when the system has no fault, there is still Approaching 0. When a fault occurs in the system, the introduction of the fading factor increases the proportion of new information at the new moment, making the information mean at the beginning of the fault more sensitive to the occurrence of the fault, thereby reducing the delay of fault detection at the beginning of the fault and reducing the missed fault alarm. At the same time, in order to detect the end of the fault in time, the statistical accumulation method can be improved as follows:

[0186] λ(k)=α·λ(k-1)+△λ(k)

[0187] From the above formula, we know that after adopting the fading factor, as time goes by, the impact of historical moments with longer time on the statistics will become smaller and smaller. Therefore, after the fault ends, as time goes by, due to the introduction of the fading factor, the impact of the historical statistical change rate at the fault time on the statistics after the fault ends will gradually disappear, making the statistical information drop to 0 quickly, thereby avoiding the situation where the statistics are continuously greater than the detection threshold, resulting in the failure to be detected.

[0188] Step 5: Considering that when detecting faults, the residual chi-square test is more sensitive to hard faults, and it is easy to miss the fault in the early stage when detecting soft faults; the sequential probability ratio test is more sensitive to soft faults and is easy to miss the fault when a hard fault occurs in the system. Therefore, the results of the two methods are combined to determine the final detection result. Here, the two fault detection results are taken or the final result is determined to reduce the occurrence of missed detection. After the interference is detected, the measurement information is isolated to prevent the interfered measurement information from polluting the Kalman filter.

[0189] The simulation conditions are set as follows:

[0190] Generate a simulation trajectory with a total duration of 900 seconds; Figure 1 The trajectory information shows the attitude, speed and position of the carrier. The strapdown inertial navigation sampling period is 0.01 seconds, and the GNSS sampling period is 1 second; the pitch-roll-azimuth misalignment angle is set to [0.5; -0.5; 0.2], in degrees; the east-north-sky velocity error is set to [0.1; 0.1; 0.1], in m / s; the latitude-longitude-altitude position error is set to [0.1 / Re; 0.1 / Re; 0.1], where Re is the semi-major axis of the earth, the gyro bias is 0.5ο / h, and the angle random walk is The acceleration zero bias is 50μg, the acceleration measurement noise is 10μg, and the rest of the information can be found in the trajectory data. The amplitude of the slowly varying disturbance is 3m / s, the injection time is 870s, and the duration is 30s. The amplitude of the suddenly varying disturbance is 5m, the injection time is 870s, and the duration is 30s.

[0191] Figure 2 This is the integrated navigation error diagram when the system is normal, showing the attitude, velocity and position errors of the integrated navigation system when there is no interference signal in the GNSS signal.

[0192] pass Figure 3 , Figure 4 It can be seen that the interference in the GNSS signal will have a great impact on the navigation system. After using the interfered information, the navigation error increases seriously, seriously affecting the performance of the navigation system.

[0193] Figure 5 , Figure 6 It shows that after using the navigation anti-interference algorithm proposed in the present invention, the impact of interfered data on the navigation system can be effectively weakened. After detecting the interference and taking relevant isolation measures, the navigation error of the navigation system is equivalent to that when the GNSS signal is not interfered with, which shows that this algorithm has a high application value.

[0194] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making, characterized in that: For the SINS / GNSS integrated navigation system composed of the strapdown inertial navigation system SINS and the global navigation satellite system GNSS, the following steps are used for anti-interference navigation: Step 1: Establish SINS / GNSS integrated navigation system model; Step 2: Establish the Kalman filter of the SINS / GNSS integrated navigation system; Step 3: Perform residual chi-square test and output interference detection result 1; Step 4: Perform a sequential probability ratio test and output interference detection result 2; Step 5: Combining the detection results in step 3 and step 4, obtaining a final interference detection result; that is, when interference detection results 1 and 2 do not detect interference, it is determined that the SINS / GNSS integrated navigation system is not interfered with; if one of the interference detection results 1 and 2 detects interference, it is determined that the SINS / GNSS integrated navigation system is interfered with; After detecting interference in the SINS / GNSS integrated navigation system, the measurement information is isolated to prevent the interfered information from contaminating the Kalman filter.

2. The anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making as claimed in claim 1, characterized in that: The step 1: establishing a SINS / GNSS integrated navigation system model; specifically: The model of the integrated navigation system is established based on the inertial navigation error equation, and the states of the integrated navigation system are selected as follows: X=[φ E f N f U dv E dv N dv U δL δλ δh e gE e gN e gU e aE e aN e aU ] T Among them, φ E ,φ N ,φ U are the SINS attitude error angles in the northeast direction; δv E ,δv N ,δv U are the SINS velocity errors in the east, north and celestial directions respectively; δL, δλ and δh are the latitude, longitude and altitude errors of SINS respectively; ε gE , ε gN , ε gU is the zero bias of the gyro's three axes in the northeast sky, ε aE , ε aN , ε aU is the zero bias of the three axes of the northeast sky of the accelerometer; the state equation is as follows: X k , X k-1 are the states of the integrated navigation system at time k and k-1 respectively, and F k / k-1 is the state transfer matrix from time k-1 to time k, Γ k-1 is the noise driving matrix at time k-1, is the k-1 system noise matrix, and the superscript b represents the carrier coordinate system; F k / k-1 The discrete formula of is as follows, where F(k) is the state matrix; F k / k-1 ≈I 15 +F(k)·△t I 15 is a 15-dimensional unit matrix, △t is the filtering period; The system noise matrix and noise driving matrix are: in is the angular random walk in the gyro output, is the velocity random walk in the accelerometer output, w gE 、w gN and w gU are the angular random walks in the east, north and sky directions respectively, w aE 、w aN and w aU They are the random walks of the speed in the east, north and sky directions respectively; is Γ k-1 For the part in; The measurement equation is: Z k =H k X k +V k Among them, V k is the GNSS position measurement noise, H k is the measurement matrix.

3. The anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making as claimed in claim 1, characterized in that: Step 2: Establish the Kalman filter of the SINS / GNSS integrated navigation system, specifically: Time update: in is the estimated value of the state from time k-1 to time k, F k / k-1 is the state transfer matrix from time k-1 to time k, is the estimated value of the state at time k-1; P k / k-1 is the state covariance matrix from time k-1 to time k, P k-1 is the state covariance matrix at time k-1, G k-1 is the system noise distribution matrix at time k-1, Q k-1 is the system noise covariance matrix at time k-1; Measurement Update: P k =(I-K k H k )P k / k-1 Where K k is the gain matrix at time k; H k is the measurement matrix at time k; R k is the measurement noise covariance matrix at time k; is the estimated value of the state at time k; Z k is the measurement value at time k.

4. The anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making as claimed in claim 1, characterized in that: Step 3: Perform a residual chi-square test and output interference detection result 1, specifically: Get the Kalman filter measurement residual r k : where z k To measure information; is the estimated value of the state from time k-1 to time k; According to the Kalman filter related information, calculate the fault detection function λ k Values: in, Comparison of fault detection function A k and threshold T D Size, output interference detection result 1, fault judgment criteria are as follows:

5. The anti-interference method based on residual chi-square and sequential probability ratio test joint decision-making as claimed in claim 1, characterized in that: The step 4: performing a sequential probability ratio test and outputting interference detection result 2, specifically: Get the Kalman filter measurement residual v k information: z k is the measurement information; H k is the measurement matrix; is the estimated value of the state from time k-1 to time k; Calculate the mean information of the measurement residuals Where α is the fading factor, which is set to 0.95; v k-1 is the residual at time k-1; Calculate the statistical information △λ based on the mean information of the measurement residual: l k =λ k-1 +△λ; Where P k / k-1 is the state covariance matrix from time k-1 to time k; Compare the statistical information and the threshold value to obtain the second detection result; the fault judgment criteria are as follows: Where P m is the false alarm rate, P f is the false alarm rate.

Images