Knowledge and data driving-based inertial navigation system fault diagnosis method and system
Through the knowledge-driven fault diagnosis method based on knowledge and data, combined with fuzzy logic operations and adaptive correction technology, the problems of low fault diagnosis accuracy and poor anti-interference ability of traditional inertial navigation systems are solved, achieving higher fault diagnosis accuracy and system stability.
Patent Information
- Application Number
- CN202411028857.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-15
AI Technical Summary
The fault diagnosis method of traditional inertial navigation systems relies on prior knowledge and fixed fault models, and there are problems such as large error, low efficiency and poor anti-interference ability.
Using a knowledge-driven fault diagnosis method, the sensor data is acquired for preprocessing and smoothing processing, combined with fuzzy logic operations and adaptive intelligent correction technology, a fault diagnosis model is built to realize the output of fault category probability distribution and real-time system parameters. Correction.
It improves the accuracy and anti-interference ability of fault diagnosis, enhances the adaptability and stability of the system, and provides multi-level and multi-type fault display, which facilitates rapid response.
Smart Images

Figure CN118960780A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of inertial navigation, and in particular relates to a knowledge and data-driven inertial navigation system fault diagnosis method and system. Background Art
[0002] High-precision inertial navigation systems (INS) are increasingly used in the fields of ocean exploration, submarine navigation, etc. In the use of high-precision inertial navigation systems, fault diagnosis is required. Traditional fault diagnosis methods are mainly based on analytical models, which rely on prior knowledge of the system and fixed fault models for fault analysis, and have the defects of large errors, low efficiency, and poor adaptability.
[0003] Therefore, it is necessary to provide a knowledge- and data-driven inertial navigation system fault diagnosis method and system to solve the above problems. Summary of the invention
[0004] The present invention provides a knowledge- and data-driven inertial navigation system fault diagnosis method and system, which improves the accuracy of fault diagnosis and the adaptability of the system by introducing knowledge- and data-driven fault diagnosis and adaptive intelligent correction technology, thereby improving the overall performance of the inertial navigation system and solving the problems of low fault diagnosis accuracy and poor anti-interference ability of traditional inertial navigation systems.
[0005] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:
[0006] A knowledge and data-driven inertial navigation system fault diagnosis method includes the following steps:
[0007] Step S1, obtaining raw data from each sensor in the inertial navigation system and preprocessing the raw data;
[0008] Step S2, using a weighted time window to smooth the preprocessed raw data to form an original data set;
[0009] Step S3, converting the data in the original data set into fuzzy sets, defining fuzzy rules according to expert knowledge, using fuzzy logic operation reasoning to obtain fault diagnosis results, and using the fault diagnosis results as labels for corresponding data in the original data set;
[0010] Step S4, constructing a fault diagnosis model, sending the original data set and the labels corresponding to the data in the original data set to the fault diagnosis model for training, until a maximum number of iterations is reached or the fault diagnosis model is fitted;
[0011] Step S5, using the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, outputting the probability distribution of the fault category of the device to which each sensor belongs, setting a probability threshold, taking the fault category with the highest probability as the preliminary prediction result, and judging whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold, if so, judging that the fault category has occurred; if not, judging that no fault has occurred; after the judgment is completed, outputting the final prediction result;
[0012] Step S6, using a multi-level and multi-type fault representation method to classify and hierarchically display the final prediction results, with each fault type corresponding to a different processing strategy and priority;
[0013] Step S7, based on the final prediction result, use an adaptive algorithm to calibrate the system parameters, adjust the system parameters in real time, and correct the impact of the fault.
[0014] As a preferred improvement, in step S1, the original data includes acceleration data, angular velocity data and position data.
[0015] As a preferred improvement, in step S1, the preprocessing includes data cleaning processing and data normalization processing.
[0016] As a preferred improvement, in step S2, the processing of the weighted time window specifically includes the following steps:
[0017] Step S21, selecting a time window size W according to system characteristics and data frequency;
[0018] Step S22, selecting a weight function to calculate the weights of all data points in the preprocessed raw data;
[0019] Step S23, performing weighted averaging on each data point based on the weight of the data point to obtain a smoothed data value.
[0020] As a preferred improvement, in step S22, the weight function is a Gaussian weight coefficient, and the calculation process is:
[0021]
[0022] In the formula, w(t) represents the weight of the data; t0 represents the center of the time window; t represents the node in the time window; σ represents the standard deviation of the weight function;
[0023] The calculation process of step S23 is expressed as:
[0024]
[0025] In the formula, x″ t represents the smoothed data; i represents the position index of the data point in the original data set; xi represents the i-th data point in the original data set.
[0026] As a preferred improvement, step S3 specifically includes the following process:
[0027] Step S31, converting the data in the original data set into a fuzzy set through a triangular membership function;
[0028] Step S32, using fuzzy rules defined according to expert knowledge and using fuzzy logic operation to infer the possibility of each fault;
[0029] Step S33, defuzzifying by the centroid method, and converting the fuzzy reasoning result into a fault diagnosis result.
[0030] As a preferred improvement, the fault diagnosis model is a mamba model based on an attention mechanism or a graph neural network model based on an attention mechanism.
[0031] As a preferred improvement, the training process of the fault diagnosis model includes the following steps:
[0032] Step S41, extracting features that can reflect system status and fault information from the original data set;
[0033] Step S42, selecting the features that best reflect the system status and fault information from the extracted features through feature selection, thereby reducing the data dimension;
[0034] Step S43, dividing the input original data set into multiple subsequences, and performing independent iterative training on each subsequence;
[0035] Step S44: After training is completed, cross-validation is used to evaluate model performance.
[0036] As a preferred improvement, step S7 specifically includes the following steps:
[0037] The recursive least squares method is used to correct the system parameters:
[0038]
[0039] In the formula, θ t represents the system parameters before correction; θ t+1 Represents the corrected system parameters; K t represents gain; y t represents the output value of the system at time t; φ t represents the input vector; T represents the transposed matrix; P t represents the covariance matrix of the system parameters before correction; λ represents the forgetting factor; P t+1Represents the covariance matrix of the corrected system parameters.
[0040] A system for executing the above-mentioned knowledge- and data-driven inertial navigation system fault diagnosis method comprises:
[0041] A data acquisition module is used to obtain raw data from various sensors in the inertial navigation system and pre-process the raw data;
[0042] The smoothing processing module uses a weighted time window to smooth the preprocessed raw data to form an original data set;
[0043] An expert knowledge base module is used to convert the data in the original data set into a fuzzy set, define fuzzy rules according to expert knowledge, use fuzzy logic operation reasoning to obtain a fault diagnosis result, and use the fault diagnosis result as a label for the corresponding data in the original data set;
[0044] A fault diagnosis model, used for sending the original data set and the labels corresponding to the data in the original data set into the fault diagnosis model for training until a maximum number of iterations is reached or the fault diagnosis model is fitted;
[0045] The fault identification module uses the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, outputs the probability distribution of the fault category of each sensor, sets the probability threshold, takes the fault category with the highest probability as the preliminary prediction result, and determines whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold. If so, it is determined that the fault category has occurred; if not, it is determined that no fault has occurred; after the determination is completed, the final prediction result is output; the classification module uses a multi-level and multi-type fault representation method to classify and hierarchically display the final prediction results, and each fault type corresponds to a different processing strategy and priority;
[0046] The correction module uses an adaptive algorithm to correct system parameters based on the final prediction results, adjusts system parameters in real time, and corrects the impact of faults.
[0047] The beneficial effects of the present invention are:
[0048] (1) Improve the accuracy of fault diagnosis: By integrating expert knowledge and historical data, a more accurate fault diagnosis model is constructed, which significantly improves the accuracy of fault diagnosis;
[0049] (2) Enhanced anti-interference capability: The weighted time window smoothing and fuzzy processing technology is used to effectively reduce the impact of environmental noise and data fluctuations on the diagnosis results, and improve the system's anti-interference capability;
[0050] (3) Adaptive correction: Through the real-time adaptive correction mechanism, the system can be dynamically adjusted according to the actual situation, thus improving the stability and reliability of the system;
[0051] (4) Multi-level and multi-type fault display: Multi-level and multi-type fault display makes the fault diagnosis results more intuitive and easy to understand, facilitating users to respond quickly. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 A flow chart showing the steps of the knowledge- and data-driven inertial navigation system fault diagnosis method provided by the present invention;
[0053] Figure 2 A framework diagram showing the knowledge- and data-driven inertial navigation system fault diagnosis method provided by the present invention. DETAILED DESCRIPTION
[0054] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0055] Please refer to Figure 1 and Figure 2 The present invention provides a knowledge and data-driven inertial navigation system fault diagnosis method, comprising the following steps:
[0056] Step S1, acquiring raw data from each sensor in an inertial navigation system, and preprocessing the raw data.
[0057] The raw data is collected by various sensors installed on the inertial navigation system, and the raw data includes acceleration data, angular velocity data and position data.
[0058] The preprocessing process includes data cleaning and data normalization.
[0059] Data cleaning includes:
[0060] Handling missing values: Delete samples with missing values or use possible values to interpolate missing values. The interpolation method can be mean substitution, interpolation, etc.
[0061] Eliminate outliers: Detect and eliminate abnormal data points using the triple standard deviation method or box plot method;
[0062] Denoising: Use wavelet transform, low-pass filtering and other methods to remove noise from the data.
[0063] Data normalization normalizes the data to a uniform range (such as [0, 1] or [-1, 1]) to eliminate dimensional differences and facilitate model processing. The normalization formula is as follows:
[0064] or
[0065] In the formula, X represents the set of original data; x represents any original data; and x′ represents the normalized value of the original data.
[0066] Step S2, using a weighted time window to smooth the preprocessed raw data to form an original data set.
[0067] In underwater inertial navigation systems, data collected by sensors (such as accelerometers, gyroscopes, etc.) are often affected by various noises and interferences. The weighted time window can effectively reduce data fluctuations and improve the accuracy of fault diagnosis. For example, the angular velocity data output by the gyroscope can more accurately reflect the actual motion state of the submarine after being smoothed by the weighted time window, thereby improving the accuracy and reliability of the navigation system.
[0068] The basic principle of weighted time window is to assign different weights to the data points in the time window, and perform weighted averaging on the data points according to these weights to obtain smoothed data values. The weights are usually determined based on the positions of the data points in the time window, and the closer the data points are to the center, the higher the weights.
[0069] In step S2, the processing of the weighted time window specifically includes the following steps:
[0070] Step S21, selecting a time window size W according to system characteristics and data frequency;
[0071] Step S22, select a weight function, calculate the weights of all data points in the preprocessed raw data, the weight coefficient selects one of a Gaussian weight function and a triangular weight function. In this embodiment, a Gaussian weight function is selected, and the calculation process is as follows:
[0072]
[0073] In the formula, w(t) represents the weight of the data; t0 represents the center of the time window; t represents the node in the time window; σ represents the standard deviation of the weight function;
[0074] Step S23, weighted average is performed on each data point based on the weight of the data point to obtain a smoothed data value. The calculation process is expressed as:
[0075]
[0076] In the formula, x″ trepresents the smoothed data; i represents the position index of the data point in the original data set; x i represents the i-th data point in the original data set.
[0077] Step S3, converting the data in the original data set into fuzzy sets, defining fuzzy rules according to expert knowledge, using fuzzy logic operation reasoning to obtain fault diagnosis results, and using the fault diagnosis results as labels for corresponding data in the original data set.
[0078] In underwater inertial navigation systems, sensor data is often affected by noise and interference, resulting in a certain degree of uncertainty and ambiguity in the data. Through fuzzy processing, these data can be used more effectively for fault diagnosis and correction.
[0079] The fuzzy processing process includes the following steps:
[0080] Step S31, converting the data smoothed in step S2 into a fuzzy set.
[0081] For example, the acceleration data is fuzzified using the triangular membership function, and three fuzzy sets of "low acceleration", "medium acceleration" and "high acceleration" are defined:
[0082]
[0083] Step S32, defining fuzzy rules based on expert knowledge, and using fuzzy logic operations to infer the possibility of each fault.
[0084] For example: if the fuzzy set of the acceleration data is "high acceleration" and the fuzzy set of the angular velocity data is "low angular velocity", the system is defined as having "fault 1"; if the fuzzy set of the velocity data is "medium acceleration" and the fuzzy set of the angular velocity data is "medium angular velocity", the system is defined as "normal".
[0085] Step S33, converting the fuzzy reasoning result into a fault diagnosis result by defuzzification.
[0086] For example, using the centroid method to defuzzify, the calculation process is expressed as:
[0087]
[0088] Among them, μ i is the membership degree of the fuzzy set, y i is the corresponding output value.
[0089] Step S4, constructing a fault diagnosis model, sending the original data set and the labels corresponding to the data in the original data set to the fault diagnosis model for training, until a maximum number of iterations is reached or the fault diagnosis model is fitted.
[0090] The fault diagnosis model is a mamba model based on an attention mechanism or a graph neural network model based on an attention mechanism. In this embodiment, it is specifically a mamba model based on an attention mechanism.
[0091] The training process of the fault diagnosis model includes the following steps:
[0092] Step S41, extracting features that can reflect system status and fault information from the original data set.
[0093] The extracted features include time domain features and frequency domain features, wherein the time domain features include mean, variance, peak, skewness and kurtosis; the frequency domain features include frequency components and power spectral density.
[0094] The calculation process of each feature in the time domain feature is as follows:
[0095] (1) Mean μ, which represents the average value of the data in the original data set, is calculated as follows:
[0096]
[0097] In the formula, x i represents the i-th data in the original data set; N represents the total number of data in the original data set;
[0098] (2) Peak value, which indicates the maximum value of the data in the original data set, is calculated as follows: Peak = max(x i );
[0099] (3) Skewness, which indicates the symmetry of the data distribution in the original data set, is calculated as:
[0100]
[0101] In the formula, σ represents the standard deviation of the data in the original data set;
[0102] (4) Kurtosis, which indicates the peak degree of data distribution in the original data set, is calculated as:
[0103]
[0104] The frequency domain features are obtained by converting time domain data into frequency domain, and the conversion method used is selected from one of Fourier transform and wavelet transform.
[0105] Step S42, selecting the features that best reflect the system status and fault information from the extracted features through feature selection, thereby reducing the data dimension.
[0106] The feature selection method is selected from one of the principal component analysis (PCA), linear discriminant analysis (LDA) and recursive feature elimination (RFE).
[0107] Step S43, dividing the input original data set into multiple subsequences, and performing independent iterative training on each subsequence;
[0108] The original data set is divided into multiple subsequences through the network, and each subsequence is processed independently, so as to make full use of local information while maintaining global information. The processing unit of each subsequence can be trained independently, thereby reducing the number of model parameters, greatly reducing the computational complexity of the model, and improving the performance of the fault diagnosis model.
[0109] Step S44: After training is completed, cross-validation is used to evaluate model performance.
[0110] The cross-validation process includes the following steps:
[0111] Divide the training set into k subsets;
[0112] Use k-1 subsets to train the model, and use the remaining 1 subset to verify the model, and repeat k times;
[0113] The average of the performance indicators for each validation is calculated as the final performance indicator of the model.
[0114] The performance indicators include accuracy, precision, recall and F1 value. The calculation process of each performance indicator is as follows:
[0115] (1) Accuracy: It indicates the ratio of correctly classified samples to the total number of samples. The calculation formula is:
[0116]
[0117] Among them, TP represents a true positive example; TN represents a true negative example; FP represents a false positive example; FN represents a false negative example;
[0118] (2) Precision: It indicates the proportion of samples predicted to be positive that are actually positive. The calculation formula is:
[0119]
[0120] (3) Recall rate Recall refers to the proportion of samples that are actually positive that are correctly predicted as positive. The calculation formula is:
[0121]
[0122] (4) F1 value F1Score represents the harmonic mean of precision and recall, and the calculation formula is:
[0123]
[0124] Step S5, use the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, output the probability distribution of the fault category of the device to which each sensor belongs, set the probability threshold, take the fault category with the highest probability as the preliminary prediction result, and judge whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold. If so, it is determined that the fault category has occurred; if not, it is determined that no fault has occurred; after the judgment is completed, the final prediction result is output.
[0125] Step S6, using a multi-level and multi-type fault representation method, classifies and hierarchically displays the final prediction results, and each fault type corresponds to a different processing strategy and priority.
[0126] Step S7, based on the final prediction result, use an adaptive algorithm to calibrate the system parameters, adjust the system parameters in real time, and correct the impact of the fault.
[0127] Step S7 specifically includes the following steps:
[0128] The recursive least squares (RLS) method is used to calibrate the system parameters:
[0129]
[0130] In the formula, θ t represents the system parameters before correction; θ t+1 Represents the corrected system parameters; K t represents gain; y t represents the output value of the system at time t; φ t represents the input vector; T represents the transposed matrix; P t represents the covariance matrix of the system parameters before correction; λ represents the forgetting factor; P t+1 Represents the covariance matrix of the corrected system parameters.
[0131] The present invention also provides a system for executing the above-mentioned knowledge- and data-driven inertial navigation system fault diagnosis method, comprising:
[0132] A data acquisition module is used to obtain raw data from various sensors in the inertial navigation system and pre-process the raw data;
[0133] The smoothing processing module uses a weighted time window to smooth the preprocessed raw data to form an original data set;
[0134] An expert knowledge base module is used to convert the data in the original data set into a fuzzy set, define fuzzy rules according to expert knowledge, use fuzzy logic operation reasoning to obtain a fault diagnosis result, and use the fault diagnosis result as a label for the corresponding data in the original data set;
[0135] A fault diagnosis model, used for sending the original data set and the labels corresponding to the data in the original data set into the fault diagnosis model for training until a maximum number of iterations is reached or the fault diagnosis model is fitted;
[0136] The fault identification module uses the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, outputs the probability distribution of the fault category of each sensor, sets the probability threshold, takes the fault category with the highest probability as the preliminary prediction result, and determines whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold. If so, it is determined that the fault category has occurred; if not, it is determined that no fault has occurred; after the determination is completed, the final prediction result is output; the classification module uses a multi-level and multi-type fault representation method to classify and hierarchically display the final prediction results, and each fault type corresponds to a different processing strategy and priority;
[0137] The correction module uses an adaptive algorithm to correct system parameters based on the final prediction results, adjusts system parameters in real time, and corrects the impact of faults.
[0138] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.
Claims
1. A knowledge and data-driven inertial navigation system fault diagnosis method, characterized in that: The steps include: Step S1, obtaining raw data from each sensor in the inertial navigation system and preprocessing the raw data; Step S2, using a weighted time window to smooth the preprocessed raw data to form an original data set; Step S3, converting the data in the original data set into fuzzy sets, defining fuzzy rules according to expert knowledge, using fuzzy logic operation reasoning to obtain fault diagnosis results, and using the fault diagnosis results as labels for corresponding data in the original data set; Step S4, constructing a fault diagnosis model, sending the original data set and the labels corresponding to the data in the original data set to the fault diagnosis model for training, until a maximum number of iterations is reached or the fault diagnosis model is fitted; Step S5, using the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, outputting the probability distribution of the fault category of the device to which each sensor belongs, setting a probability threshold, taking the fault category with the highest probability as the preliminary prediction result, and judging whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold, if so, judging that the fault category has occurred; if not, judging that no fault has occurred; after the judgment is completed, outputting the final prediction result; Step S6, using a multi-level and multi-type fault representation method to classify and hierarchically display the final prediction results, with each fault type corresponding to a different processing strategy and priority; Step S7, based on the final prediction result, use an adaptive algorithm to calibrate the system parameters, adjust the system parameters in real time, and correct the impact of the fault.
2. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: In step S1, the raw data includes acceleration data, angular velocity data and position data.
3. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: In step S1, preprocessing includes data cleaning and data normalization.
4. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: In step S2, the processing of the weighted time window specifically includes the following steps: Step S21, selecting a time window size W according to system characteristics and data frequency; Step S22, selecting a weight function to calculate the weights of all data points in the preprocessed raw data; Step S23, performing weighted averaging on each data point based on the weight of the data point to obtain a smoothed data value.
5. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 4 is characterized in that: In step S22, the weight function is a Gaussian weight coefficient, and the calculation process is: In the formula, w(t) represents the weight of the data; t0 represents the center of the time window; t represents the node in the time window; σ represents the standard deviation of the weight function; The calculation process of step S23 is expressed as: In the formula, x″ t represents the smoothed data; i represents the position index of the data point in the original data set; x i represents the i-th data point in the original data set.
6. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: Step S3 specifically includes the following process: Step S31, converting the data in the original data set into a fuzzy set through a triangular membership function; Step S32, using fuzzy rules defined according to expert knowledge and fuzzy logic operation reasoning to obtain the possibility of each fault; Step S33, defuzzifying by the centroid method, and converting the fuzzy reasoning result into a fault diagnosis result.
7. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: The fault diagnosis model is a mamba model based on an attention mechanism or a graph neural network model based on an attention mechanism.
8. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: The training process of the fault diagnosis model includes the following steps: Step S41, extracting features that can reflect system status and fault information from the original data set; Step S42, selecting the features that best reflect the system status and fault information from the extracted features through feature selection, thereby reducing the data dimension; Step S43, dividing the input original data set into multiple subsequences, and performing independent iterative training on each subsequence; Step S44: After training is completed, cross-validation is used to evaluate model performance.
9. The knowledge and data-driven inertial navigation system fault diagnosis method according to claim 1, characterized in that: Step S7 specifically includes the following steps: The recursive least squares method is used to correct the system parameters: In the formula, θ t represents the system parameters before correction; θ t+1 Represents the corrected system parameters; K t represents gain; y t Represents the output value of the system at time t; φ t represents the input vector; T represents the transposed matrix; P t represents the covariance matrix of the system parameters before correction; λ represents the forgetting factor; P t+1 Represents the covariance matrix of the corrected system parameters.
10. A system for executing the knowledge and data-driven inertial navigation system fault diagnosis method according to any one of claims 1 to 9, characterized in that: include: A data acquisition module is used to obtain raw data from various sensors in the inertial navigation system and pre-process the raw data; The smoothing processing module uses a weighted time window to smooth the preprocessed raw data to form an original data set; An expert knowledge base module is used to convert the data in the original data set into a fuzzy set, define fuzzy rules according to expert knowledge, use fuzzy logic operation reasoning to obtain a fault diagnosis result, and use the fault diagnosis result as a label for the corresponding data in the original data set; A fault diagnosis model, used for sending the original data set and the labels corresponding to the data in the original data set into the fault diagnosis model for training until a maximum number of iterations is reached or the fault diagnosis model is fitted; The fault identification module uses the trained fault diagnosis model to perform fault diagnosis on the inertial navigation system, outputs the probability distribution of the fault category of each sensor, sets the probability threshold, takes the fault category with the highest probability as the preliminary prediction result, and determines whether the fault probability of the fault category in the preliminary prediction result is not less than the probability threshold. If so, it is determined that the fault category has occurred; if not, it is determined that no fault has occurred; after the determination is completed, the final prediction result is output; The classification module uses a multi-level and multi-type fault representation method to classify and hierarchically display the final prediction results. Each fault type corresponds to a different processing strategy and priority. The correction module uses an adaptive algorithm to correct system parameters based on the final prediction results, adjusts system parameters in real time, and corrects the impact of faults.
Citation Information
Cited By
Fault diagnosis system of organic waste gas catalytic decomposition device
CN120333541A
