A method for diagnosing faults of a spatial gravitational wave detection inertial sensor

Abstract

The application discloses a kind of space gravitational wave detection inertial sensor fault diagnosis methods, belong to space gravitational wave detection inertial sensor fault diagnosis technical field.It includes the following steps: based on the sensor measurement channel of mutual coupling establishes double-channel Informer model and is summarized to obtain sensor fault condition;Obtain state measurement value and the state quantity prediction value of the target based on double-channel Informer model prediction, according to state measurement value and the state quantity prediction value of the target, generate state quantity residual sequence;Residual sequence is generated based on residual sequence, judge residual sequence based on residual sequence, set dynamic threshold based on judgment residual sequence, compare judgment residual sequence and dynamic threshold, obtain comparison result;Fault detection is carried out in combination with comparison result and sensor fault condition;Fault isolation is carried out by analyzing and judging residual sequence representation, complete space gravitational wave detection inertial sensor fault diagnosis.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of fault diagnosis of space gravitational wave detection inertial sensor, and particularly relates to a fault diagnosis method of space gravitational wave detection inertial sensor. BACKGROUND

[0002] High-precision measurement of gravitational waves will be one of the most advanced and important research topics in the field of future basic science. As one of the key payloads of space gravitational wave detection missions, space inertial sensors have the following two main functions: accurately measuring the acceleration of a spacecraft caused by non-conservative forces to achieve drag-free flight of the spacecraft; and serving as part of a laser interferometer system to measure the relative displacement between a test mass and the spacecraft to provide an inertial reference for the laser interferometer.

[0003] Currently, the residual acceleration noise index required by space gravitational wave detection programs around the world is on the order of 10 -15 m / s 2 / Hz 1 / 2 . In the face of the aforementioned precision requirements, whether the space inertial sensor is working properly will directly affect the implementation of the detection mission. Therefore, timely and accurate fault diagnosis of space gravitational wave detection inertial sensors is very important.

[0004] Traditional satellite sensor fault detection methods mainly use hardware redundancy, but space gravitational wave detection inertial sensors cannot meet the redundancy requirements. Model-based fault diagnosis methods require accurate physical models and are not suitable for systems with strong nonlinearity and high coupling characteristics, such as drag-free systems. In addition, traditional time series prediction methods, including ARIMA models and machine learning methods, can only achieve short-term sequence prediction and are difficult to accurately predict long-term sequences. In recent years, LSTM neural networks have been commonly used for time series prediction. Although the LSTM model can achieve effective time series prediction, the prediction accuracy of the model will decrease significantly as the length of the time series increases, and the prediction time will also increase significantly, which is not suitable for real-time fault detection of on-orbit sensors. SUMMARY

[0005] To overcome the shortcomings of the prior art, the purpose of the present application is to provide a fault diagnosis method for space gravitational wave detection inertial sensors, which does not require an accurate physical model of the drag-free system and only needs measurement data of the space gravitational wave detection inertial sensor to achieve fault diagnosis of the sensor.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] The application provides a space gravitational wave detection inertial sensor fault diagnosis method, comprising the following steps:

[0008] A double-channel Informer model is established based on the mutually coupled sensor measurement channels, and sensor fault conditions are obtained by summarizing;

[0009] State measurement values and state quantity prediction values of a target predicted based on the double-channel Informer model are obtained, a residual sequence of the state quantity is generated according to the state measurement values and the state quantity prediction values of the target, a judgment residual sequence is generated based on the residual sequence, a dynamic threshold is set based on the judgment residual sequence, the judgment residual sequence and the dynamic threshold are compared, and a comparison result is obtained.

[0010] Fault detection is performed in combination with the comparison result and the sensor fault conditions, fault isolation is performed by analyzing the judgment residual sequence, and space gravitational wave detection inertial sensor fault diagnosis is completed.

[0011] In the specific implementation process, the determination process of the mutually coupled sensor measurement channels is as follows:

[0012] A comprehensive pose measurement model of the space gravitational wave detection inertial sensor is established based on a sensitive structure of the space gravitational wave detection inertial sensor, and the mutually coupled sensor measurement channels are determined.

[0013] In the specific implementation process, the sensitive structure of the space gravitational wave detection inertial sensor is composed of a cubic test mass and an electrode cage; the test mass is located at the middle position of the electrode cage, the electrode cage has a plurality of surfaces, electrode plates are arranged on the plurality of surfaces, the electrode plates on opposite surfaces are the same, and the electrode plates on the opposite surfaces form a differential capacitance sensor.

[0014] In the specific implementation process, an x-axis, a y-axis and a z-axis of a three-axis rectangular coordinate system of the test mass are set, and the comprehensive pose measurement model of the space gravitational wave detection inertial sensor is:

[0015]

[0016] Wherein, C0 represents the initial capacitance of the test mass and the electrode plate; d0 represents the initial distance between the test mass and the electrode plate; L is the distance between the center of the electrode plate and the initial center surface of the test mass; ΔC 12 and ΔC 34 are the measurement values of the two differential capacitance sensors related to the displacement in the x-axis direction and the rotation angle in the z-axis direction, respectively; d x represents the displacement measurement value of the test mass along the x-axis, represents the rotation angle measurement value of the test mass along the z-axis.

[0017] In the specific implementation process, the ΔC12 and ΔC 34 As follows:

[0018]

[0019] Wherein, when the test mass simultaneously occurs translation and rotation, the actual displacement of the test mass along the x-axis direction is x, and the actual rotation angle along the z-axis direction is θ z , C0 represents the initial capacitance of the test mass and the electrode plate; d0 represents the initial distance between the test mass and the electrode plate; L is the distance between the center of the electrode plate and the initial center surface of the test mass.

[0020] In the specific implementation process, the mutually coupled sensor measurement channels include an x-axis displacement measurement channel and a z-axis direction rotation angle measurement channel, a y-axis displacement measurement channel and an x-axis direction rotation angle measurement channel, and a z-axis displacement measurement channel and a y-axis direction rotation angle measurement channel.

[0021] The double-channel Informer model includes M xz model, M yx model and M zy model.

[0022] Wherein, the state quantity prediction value of the prediction target obtained by the M xz model is the x-axis displacement and the z-axis direction rotation angle; the state quantity prediction value of the prediction target obtained by the M yx model is the y-axis displacement and the x-axis direction rotation angle; and the state quantity prediction value of the prediction target obtained by the M zy model is the z-axis displacement and the y-axis direction rotation angle.

[0023] In the specific implementation process, the sensor fault conditions are as follows:

[0024] When the judgment residual sequences of two state quantities of the mutually coupled sensor measurement channels simultaneously appear abnormal, it is determined that one of the sensors related to the two state quantities has a fault;

[0025] When the judgment residual sequences of two state quantities of the mutually coupled sensor measurement channels do not simultaneously appear abnormal, it is determined that the sensor has no fault.

[0026] In the specific implementation process, the dynamic threshold is determined based on the 3-σ rule of the past judgment residual sequence and the current time judgment residual sequence, and is specifically as follows:

[0027]

[0028]

[0029] wherein, Γ (t) is the residual threshold value at time t, is the mean value of the sequence is the standard deviation of the sequence is the mean value of the sequence is the standard deviation of the sequence

[0030] In the specific implementation process, the comparison result comprises:

[0031] When the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels exceeds the dynamic threshold value, it indicates that the judgment residual sequence is abnormal; when the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels is abnormal at the same time, it is determined that the sensor is faulty.

[0032] When the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels does not exceed the dynamic threshold value, it indicates that the judgment residual sequence is not abnormal; when the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels is not abnormal at the same time, it is determined that the sensor is not faulty.

[0033] In the specific implementation process, the process of analyzing the judgment residual sequence to isolate faults is as follows:

[0034] When the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceeds the same side dynamic threshold value, it is a same direction abnormality; when the judgment residual sequence of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceeds the different side dynamic threshold value, it is a reverse abnormality; based on the same direction abnormality and the reverse abnormality, the corresponding sensor fault condition is judged, and fault isolation is performed.

[0035] Compared with the prior art, the present application has the following beneficial effects:

[0036] The present application provides a space gravitational wave detection inertial sensor fault diagnosis method, which realizes double-channel prediction of a longer time sequence based on Informer time series prediction, obtains higher prediction accuracy and shorter prediction time, and is more suitable for fault diagnosis of space gravitational wave detection inertial sensors. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is an implementation flowchart of the space gravitational wave detection inertial sensor fault diagnosis method based on Informer time series prediction of the embodiments of the present application;

[0038] Figure 2A plot of differential capacitance measurement change for a fault-free case of an embodiment of the application;

[0039] Figure 3 A plot of displacement d for a fault-free case of an embodiment of the application x and the angle of rotation A plot of residual change.

[0040] Figure 4 A plot of differential capacitance measurement change for a fault-free case of an embodiment of the application;

[0041] Figure 5 A plot of displacement d for a fault-free case of an embodiment of the application x and the angle of rotation A plot of residual change. DETAILED DESCRIPTION

[0042] In order to make the personnel in the art better understand the application scheme, the technical scheme in the embodiments of the application will be described clearly and completely below in conjunction with the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative labor should belong to the protection scope of the application.

[0043] The exemplary embodiments of the present application are described herein with reference to the accompanying drawings, which are included to provide a thorough understanding of the embodiments of the application by a person of ordinary skill in the art, and should be considered in conjunction with the preceding description, and are included to provide a thorough understanding of the embodiments of the application by a person of ordinary skill in the art, and should be considered in conjunction with the preceding description. Therefore, those of ordinary skill in the art should realize that various changes and modifications can be made to the embodiments described herein, without departing from the scope and spirit of the application. Similarly, for the sake of clarity and conciseness, the description below omits the description of well-known functions and structures.

[0044] It should be noted that the terminal involved in the embodiments of the present application can include, but is not limited to, a mobile phone, a personal digital assistant (PDA), a wireless handheld device, a tablet computer, a personal computer (PC), an MP3 player, an MP4 player, a wearable device (for example, smart glasses, a smart watch, a smart bracelet, etc.), a smart home device, and the like.

[0045] In addition, the term "and / or" in this paper is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are a "or" relationship.

[0046] The application provides a space gravitational wave detection inertial sensor fault diagnosis method, comprising the following steps:

[0047] A comprehensive pose measurement model of the space gravitational wave detection inertial sensor is established based on a sensitive structure of the space gravitational wave detection inertial sensor, and a sensor measurement channel coupled with each other is determined;

[0048] A double-channel Informer model is established based on the sensor measurement channel coupled with each other, and a sensor fault condition is obtained by summarizing;

[0049] A state measurement value and a state quantity prediction value of a target predicted based on the double-channel Informer model are obtained, a residual sequence of the state quantity is generated according to the state measurement value and the state quantity prediction value of the target, a judgment residual sequence is generated based on the residual sequence, a dynamic threshold is set based on the judgment residual sequence, the judgment residual sequence and the dynamic threshold are compared, and a comparison result is obtained;

[0050] Fault detection is performed in combination with the comparison result and the sensor fault condition, fault isolation is performed by analyzing and judging the residual sequence, and space gravitational wave detection inertial sensor fault diagnosis is completed.

[0051] The Informer model can realize double-channel prediction of a long time sequence at the same time, has high prediction accuracy and short prediction time, and the intelligent fault diagnosis method based on data driving does not need an accurate model, but only needs sensor data to realize fault diagnosis, so that the model is suitable for space gravitational wave detection inertial sensor fault diagnosis.

[0052] The space gravitational wave detection inertial sensor fault diagnosis method does not need an accurate physical model of a non-dragging system, but only needs measurement data of the space gravitational wave detection inertial sensor, so that fault diagnosis of the sensor can be realized, and the method is simple and efficient, has great economic benefits and wide application prospect.

[0053] The application will be further described in detail in combination with the drawings:

[0054] Referring to Figure 1 The space gravitational wave detection inertial sensor fault diagnosis method is as follows:

[0055] Step 1: Determine the composition of the sensitive structure of the space gravitational wave detection inertial sensor, determine the shape of the proof mass, the electrode distribution of the space gravitational wave detection inertial sensor and the number of differential capacitance sensors;

[0056] The sensitive structure of the spatial gravitational wave detection inertial sensor is composed of a cubic-shaped proof mass and an electrode cage, the proof mass is located in the middle of the electrode cage, the electrode cage has a plurality of faces, two electrode plates are distributed on each face of the electrode cage, the electrode plates on opposite faces are completely the same and constitute a variable-spacing type capacitive sensor in a differential structure, i.e., a differential capacitive sensor, each pair of opposite faces includes two differential capacitive sensors, and the entire spatial gravitational wave detection inertial sensor has six differential capacitive sensors.

[0057] Step two: according to the composition of the sensitive structure of the spatial gravitational wave detection inertial sensor in step one, the comprehensive measurement model of this type of spatial gravitational wave detection inertial sensor is based on differential capacitive sensors. First, a three-axis rectangular coordinate system of the proof mass is established, which is x-axis, y-axis and z-axis, and the actual displacement of the proof mass along the x-axis direction is set as x, and the actual rotation angle along the z-axis direction is set as θ z The measurement values of the two differential capacitive sensors related to the displacement along the x-axis direction and the rotation angle along the z-axis direction are:

[0058]

[0059] Wherein, C0 represents the initial capacitance of the proof mass and the electrode plate, d0 represents the initial spacing of the proof mass and the electrode plate, the center of the electrode plate and the initial center surface distance of the proof mass are L, ΔC 12 and ΔC 34 are the measurement values of the two differential capacitive sensors. The two differential capacitive sensors are EL 12 and EL 34 , and through the difference mode and common mode of the measurement values of the two sensors, the comprehensive measurement model of the spatial gravitational wave detection inertial sensor pose is:

[0060]

[0061] Wherein, d x represents the measurement value of the displacement of the proof mass along the x-axis, represents the measurement value of the rotation angle of the proof mass along the z-axis direction.

[0062] Step three: according to the comprehensive measurement model of the spatial gravitational wave detection inertial sensor pose established in step two, the mutually coupled sensor measurement channels are determined.

[0063] When the differential capacitive measurement values ΔC 12 or ΔC 34 are abnormal, d x and The x-axis displacement measurement channel and the z-axis direction angle measurement channel are coupled with each other, so that the two state quantities are simultaneously affected and abnormal; similarly, according to the measurement model of other state quantities, it can be obtained that the y-axis displacement measurement channel and the x-axis direction angle measurement channel are coupled with each other, and the z-axis displacement measurement channel and the y-axis direction angle measurement channel are coupled with each other.

[0064] According to the mutually coupled sensor measurement channels, the three parallel double-channel Informer models to be trained are M xz , M yx and M zy , which correspond to the predicted state quantities of the models M xy --x-axis displacement and z-axis angle, the model M yx --y-axis displacement and x-axis angle, and the model M zx --z-axis displacement and y-axis angle. That is, the predicted state quantity prediction value of the prediction target obtained by the M xz model is the x-axis displacement and the z-axis direction angle; the predicted state quantity prediction value of the prediction target obtained by the M yx model is the y-axis displacement and the x-axis direction angle; and the predicted state quantity prediction value of the prediction target obtained by the M zy model is the z-axis displacement and the y-axis direction angle.

[0065] Step four: according to the mutually coupled sensor measurement channels determined in step three, classify the sensor fault conditions (space gravitational wave detection inertial sensor fault diagnosis conditions), wherein the sensor fault conditions include:

[0066] When the two state quantity judgment residual sequences of the mutually coupled sensor measurement channels are abnormal at the same time, it is determined that one of the differential capacitors (sensors) related to the two state quantities has failed, and if the two judgment residual sequences are not abnormal at the same time, the sensor is not faulty;

[0067] When the sensor is detected to have a fault, the differential capacitor sensor with the fault needs to be further isolated according to the specific performance of the abnormal judgment residual sequence and the coupling relationship of the measurement channel.

[0068] Taking the differential capacitor sensors EL 12 and EL 34 as examples, the fault diagnosis rule can be represented by Table 1.

[0069] Table 1 Fault diagnosis rule example

[0070]

[0071] Step 5: Using the three parallel dual-channel Informer models trained in Step 3, predict the state variables of the target and obtain the state variable measurements. The residual sequence of the above state variables is obtained by subtracting the state variable measurements from the state variable predictions. The residual sequence is then smoothed using an exponentially weighted moving average method to obtain the judgment residual sequence, specifically:

[0072] ε (t) =βε (t-1) +(1-β)e (t) (3)

[0073] Where, ε (t) and ε (t-1) These represent the residuals at time t and time (t-1) in the residual sequence, respectively. (t) Let t be the residual at time t, and β be the exponentially weighted average weight.

[0074] Step Six: Set a dynamic threshold based on the smoothed judgment residual sequence from Step Five. The dynamic threshold is determined by the past judgment residual sequences and the current judgment residual sequence based on the 3-σ rule, specifically:

[0075]

[0076]

[0077] Among them, Γ (t) Let t be the residual threshold. For sequence The mean, For sequence The standard deviation of h is the length of the dynamic threshold lookback window.

[0078] Step 7: Determine the status of the residuals based on the dynamic thresholds set in Step 6, and perform fault detection;

[0079] If the residual exceeds the dynamic threshold, it indicates that the residual sequence is abnormal; otherwise, the sequence is normal. According to the classification of sensor fault conditions (fault diagnosis of inertial sensors for space gravitational wave detection), when the residual sequences of mutually coupled measurement channels simultaneously exceed the threshold, the sensor is determined to be faulty; otherwise, the sensor is normal.

[0080] The specific comparison results are as follows:

[0081] When the judgment residual sequence of two state quantities of the coupled sensor measurement channels exceeds the dynamic threshold, it indicates that the judgment residual sequence is abnormal; when the judgment residual sequences of two state quantities of the coupled sensor measurement channels are abnormal at the same time, it is determined that the sensor is faulty.

[0082] When the determination residual sequences of the two state quantities of the mutually coupled sensor measurement channels do not exceed the dynamic threshold, it indicates that the determination residual sequences do not appear abnormal; when the determination residual sequences of the two state quantities of the mutually coupled sensor measurement channels do not appear abnormal at the same time, it is determined that the sensor does not appear failure.

[0083] Step eight: analyze the determination residual sequence performance to isolate the fault, complete the fault diagnosis of the space gravitational wave detection inertial sensor.

[0084] When the determination residual sequences of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceed the same side dynamic threshold, it is a same direction abnormality; when the determination residual sequences of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceed different side dynamic thresholds, it is a reverse abnormality; based on the same direction abnormality and the reverse abnormality, the corresponding sensor failure condition is determined, and fault isolation is performed.

[0085] The fault isolation method is based on different performances of the determination residual sequences of the mutually coupled sensor measurement channels. The two abnormal determination residual sequences can have two performances of simultaneously exceeding the same side dynamic threshold and simultaneously exceeding the different side dynamic threshold. Simultaneously exceeding the same side dynamic threshold is a same direction abnormality, and simultaneously exceeding the different side dynamic threshold is a reverse abnormality. Different abnormalities indicate that different differential capacitance sensors appear failure.

[0086] Taking the example in Table 1 as an example, when d x 、 The residual same direction abnormality indicates that ΔC 34 appears abnormal in formula (2), that is, the differential capacitance sensor E 34 fails, and when d x 、 The residual reverse abnormality indicates that ΔC 12 appears abnormal in formula (2), that is, the differential capacitance sensor E 12 fails.

[0087] Embodiment

[0088] The following simulation by Matlab / Simulink and Python explains the specific diagnosis process of the space gravitational wave detection inertial sensor fault diagnosis method based on the Informer time series prediction.

[0089] First, the measurement data of the space gravitational wave detection inertial sensor needs to be generated for training the Informer model. A six-degree-of-freedom dynamics model and a comprehensive pose measurement model of the test mass are built by using Matlab / Simulink, and further considering the noise model of the differential capacitance sensor:

[0090]

[0091] where subscript i represents the measurement index, subscript j represents the sensing channel, A ij is the high-frequency noise part, B and C j are the low-frequency noise part coefficients, Δ ij is the test mass displacement, and f is the noise frequency. The simulation parameters of the model are set as follows:

[0092] The electrode plate length b = 30 mm, the electrode plate width a = 16 mm, the test mass side length S = 46 mm, the test mass mass m = 1.96 kg, the initial plate spacing d0 = 4 mm, the initial center distance L = 10.75 mm, the sampling frequency fs = 1 Hz, A ij = 0.78aF·Hz -1 / 2 , B = 31 ppm Hz -1 / 2 , C x12 = 0.57aF·Hz -1 / 2 , C x34 = 0.54aF·Hz -1 / 2 .

[0093] After obtaining the measurement data set of the spatial gravitational wave detection inertial sensor, the data set is used for training of the Informer model. Some hyperparameters of the trained model are set as follows:

[0094] The encoder input sequence length L e = 120, the decoder input sequence length L d = 90, the decoder output sequence length O d = 30, the lookback window length L b = 60, the number of encoder layers S e = 3, the number of decoder layers S d = 3, the Embed vector dimension L embed = 512, the batch size S batch = 32, the initial learning rate lr = 0.0001, and the loss function is MSE.

[0095] The trained two-channel Informer model is used for time series prediction and fault diagnosis. Here, only the test of the sensor fault diagnosis based on the measurement channel residual of the test mass x-axis direction displacement and z-axis direction rotation is considered. The time series prediction and fault diagnosis parameters are set as follows: the total prediction length T = 600 s, the prediction step dT = 30 s, the exponential weighted average weight β = 0.9, and the dynamic threshold lookback window length h = 299.

[0096] The faults are generated by injecting intermittent bias faults three times in the normal differential capacitance measurement values, the fault occurrence times are 100s, 300s and 500s respectively, the second fault occurs at different positions, the fault durations are 5s, 7s and 10s respectively, and the fault sizes are 0.1fF, 0.14fF and 0.2fF respectively.

[0097] The simulation results in the fault-free case are shown in Figure 2 and Figure 3 . As can be seen from Figure 2 , the generated fault-free differential capacitance measurement values are in the order of 10 -16 F, and contain obvious noise. Since no fault is injected, the differential capacitance measurement value curve is relatively smooth. Correspondingly, the x-axis displacement and z-axis rotation angle measurement value residuals both conform to the normal distribution, and there is no obvious simultaneous exceedance of the set threshold, which indicates that the residual sequence is normal, and according to the conditions given in Table 1, the sensor is fault-free at this time.

[0098] The simulation results in the fault-free case are shown in Figure 4 and Figure 5 . As can be seen from Figure 4 , three different intermittent faults occur at 100s, 300s and 500s, and the second fault occurs at a different differential capacitance sensor from the other two faults. Correspondingly, as can be seen from Figure 5 , the residuals at 100s, 300s and 500s all exceed the threshold, indicating that abnormalities are successfully detected at these three times. According to the diagnostic rules given in Table 1, since the x-axis displacement and z-axis rotation angle measurement value residuals all appear abnormal at the same time, the sensor has faults at these three times, thereby achieving fault detection. The residuals corresponding to the first and third faults exceed the thresholds of different sides, i.e. reverse abnormalities occur, so it can be determined that the two faults occur on the differential capacitance sensor EL 12 ; and the residual corresponding to the second fault exceeds the threshold of the same side, i.e. a same-direction abnormality occurs, so it can be determined that the fault occurs on the differential capacitance sensor EL 34 , thereby achieving fault isolation.

[0099] The embodiment provides a space gravitational wave detection inertial sensor fault diagnosis method, based on Informer time series prediction, by explicitly defining the sensitive structure composition of the space gravitational wave detection inertial sensor, establishing a space gravitational wave detection inertial sensor comprehensive position and posture measurement model, determining the mutually coupled sensor measurement channels, training three parallel double-channel Informer models according to the coupling relationship to predict the state quantity prediction value of the target and specify the fault diagnosis rule; obtaining the state quantity measurement value, generating the corresponding residual sequence according to the state quantity measurement value and the state quantity prediction value of the predicted target, generating a new judgment residual sequence by smoothing the residual sequence and setting a dynamic threshold, comparing whether the judgment residual sequence exceeds the set threshold, and simultaneously according to the specified fault diagnosis rule, judging the fault condition of the sensor to realize fault detection, and after detecting the fault, further realizing fault isolation according to the residual performance.

[0100] The above is only used for describing the technical idea of the present application, and cannot be used to limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical scheme falls within the protection scope of the claims of the present application.

Claims

1. A method for diagnosing a fault of a spatial gravitational wave probe inertial sensor, characterized by, The method comprises the following steps: A double-channel Informer model is established based on the mutually coupled sensor measurement channels, and sensor fault conditions are summarized; State measurement values and state quantity prediction values of a target predicted based on the double-channel Informer model are obtained, and a residual sequence of the state quantity is generated according to the state measurement values and the state quantity prediction values of the target; A judgment residual sequence is generated based on the residual sequence, a dynamic threshold is set based on the judgment residual sequence, and a comparison result is obtained by comparing the judgment residual sequence and the dynamic threshold; Fault detection is performed in combination with the comparison result and the sensor fault conditions, fault isolation is performed by analyzing and judging the residual sequence, and fault diagnosis of the inertial sensor for space gravitational wave detection is completed; The determination process of the mutually coupled sensor measurement channels is as follows: A comprehensive pose measurement model of the inertial sensor for space gravitational wave detection is established based on a sensitive structure of the inertial sensor for space gravitational wave detection, and the mutually coupled sensor measurement channels are determined; An x-axis, a y-axis and a z-axis of a three-axis rectangular coordinate system of a test mass are set, and the comprehensive pose measurement model of the inertial sensor for space gravitational wave detection is as follows: ； wherein, represents the initial capacitance of the test mass and the electrode plate; represents the initial distance between the test mass and the electrode plate; L is the distance between the center of the electrode plate and the initial center plane of the test mass; and are the measurement values of the two differential capacitance sensors related to the displacement in the x-axis direction and the rotation angle in the z-axis direction, respectively; represents the measurement value of the displacement of the test mass along the x-axis, represents the measurement value of the rotation angle of the test mass along the z-axis. The and As follows: ； wherein, when the test mass simultaneously has a translation and a rotation, the actual displacement of the test mass along the x-axis is set as , and the actual rotation angle of the test mass along the z-axis is set as , represents an initial capacitance between the test mass and the electrode plate; represents an initial distance between the test mass and the electrode plate; and L represents a distance between the center of the electrode plate and the initial center surface of the test mass. The mutually coupled sensor measurement channels include an x-axis displacement measurement channel and a z-axis direction angle measurement channel mutually coupled sensor measurement channel, a y-axis displacement measurement channel and an x-axis direction angle measurement channel mutually coupled sensor measurement channel, and a z-axis displacement measurement channel and a y-axis direction angle measurement channel mutually coupled sensor measurement channel; The dual-channel Informer model comprises a model, a model, and a model; wherein, the state quantity prediction value of the prediction target obtained by the model is an x-axis displacement amount and a z-axis direction rotation angle; the state quantity prediction value of the prediction target obtained by the model is a y-axis displacement amount and an x-axis direction rotation angle; the state quantity prediction value of the prediction target obtained by the model is a z-axis displacement amount and a y-axis direction rotation angle.

2. The method according to claim 1, wherein The sensitive structure of the inertial sensor for space gravitational wave detection comprises a cubic test mass and an electrode cage; the test mass is located at a middle position of the electrode cage; the electrode cage has a plurality of surfaces, and electrode plates are arranged on the plurality of surfaces; the electrode plates on opposite surfaces of the plurality of surfaces are the same, and the electrode plates on the opposite surfaces form a differential capacitance sensor.

3. The method according to claim 1, wherein The sensor fault conditions are as follows: When the judgment residual sequences of two state quantities of the mutually coupled sensor measurement channels are abnormal at the same time, it is determined that one of the sensors related to the two state quantities has a fault; When the judgment residual sequences of the two state quantities of the mutually coupled sensor measurement channels are not abnormal at the same time, it is determined that the sensor has no fault.

4. The method according to claim 3, wherein The dynamic threshold is determined based on 3-σ rules of past judgment residual sequences and current time judgment residual sequences, and specifically as follows: ； ； in, Let t be the residual threshold. For sequence The mean, For sequence The standard deviation of h is the length of the dynamic threshold lookback window.

5. The method according to claim 4, wherein The comparison result includes: When the judgment residual sequences of the two state quantities of the mutually coupled sensor measurement channels exceed the dynamic threshold, it is indicated that the judgment residual sequences are abnormal; when the judgment residual sequences of the two state quantities of the mutually coupled sensor measurement channels are abnormal at the same time, it is determined that the sensor has a fault; When the judgment residual sequences of the two state quantities of the mutually coupled sensor measurement channels do not exceed the dynamic threshold, it is indicated that the judgment residual sequences are not abnormal; when the judgment residual sequences of the two state quantities of the mutually coupled sensor measurement channels are not abnormal at the same time, it is determined that the sensor has no fault.

6. The space-based gravitational wave explorer inertial sensor fault diagnosis method of claim 5, wherein, The process of analyzing and judging the residual sequence for fault isolation is as follows: The same-direction abnormality is when the determination residual sequence of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceeds the dynamic threshold of the same side; the reverse abnormality is when the determination residual sequence of the two state quantities of the mutually coupled sensor measurement channels simultaneously exceeds the dynamic threshold of different sides; The corresponding sensor fault condition is determined based on the same-direction abnormality and the reverse abnormality, and fault isolation is performed.

Images