Knowledge data hybrid-driven rate gyroscope accurate health state assessment method
By constructing a confidence rule base that integrates expert knowledge and historical data, and combining it with a gradient descent optimization algorithm, the problems of data imbalance and accuracy in rate gyroscope health status assessment were solved, achieving highly accurate health status assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHENGDU AIRCRAFT DESIGN INST OF AVIATION IND CORP OF CHINA
- Filing Date
- 2025-12-27
- Publication Date
- 2026-05-12
AI Technical Summary
Rate gyroscopes suffer from uneven data volume across different modes and insufficient accuracy in health status assessment, especially in the early stages of a fault or when the fault modes are complex, making it difficult for traditional methods to capture subtle changes.
We adopt a knowledge-data hybrid approach, which integrates expert knowledge and historical data by constructing a confidence rule base (BRB). We design zero-order term drift and first-order term drift as input indicators, and combine the gradient descent optimization algorithm of Huber loss function to adjust model parameters to improve evaluation accuracy.
It effectively solved the problem of unbalanced data volume, significantly improved the accuracy and reliability of health status assessment, and increased the prediction accuracy from 66.6% to 97.2%.
Smart Images
Figure CN122015904A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of rate gyroscope health status assessment technology, and in particular relates to a knowledge data hybrid-driven method for accurate health status assessment of rate gyroscopes. Background Technology
[0002] Rate gyroscopes, as a type of high-precision angular velocity sensor, are widely used in high-precision systems such as navigation, guidance, and attitude control due to their excellent measurement accuracy and stability. Their performance directly affects the overall performance and reliability of the system. However, during long-term operation, rate gyroscopes are susceptible to various factors such as environmental changes, mechanical wear, and electronic component aging, leading to performance degradation and even malfunctions. Common malfunctions manifest as abnormal changes in zero-term drift and first-term drift, which severely impact the system's accuracy and stability. Therefore, designing accurate and reliable health status assessment methods to monitor rate gyroscope performance changes in real time and implementing corresponding maintenance measures is of significant practical importance for improving the robustness and reliability of the system.
[0003] Therefore, in order to improve the robustness and accuracy of traditional rate gyroscope health status assessment methods, a new health status assessment scheme based on knowledge data hybrid driving is proposed, which mainly addresses the following two issues: 1) The problem of uneven data volume among different modalities.
[0004] In the health status assessment of rate gyroscopes, the acquisition of multimodal data often suffers from an imbalance in the amount of data. Some sensor data is easy to obtain, while other data is relatively scarce, which affects the training and generalization ability of the assessment model.
[0005] 2) The accuracy of health status assessment results.
[0006] Traditional assessment methods rely on simple threshold judgments or statistical analysis, which are difficult to capture subtle changes and complex characteristics of the health status of rate gyroscopes, especially in the early stages of failure or in cases of complex failure modes. The accuracy of the assessment results needs to be improved. Summary of the Invention
[0007] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a knowledge data hybrid-driven method for accurate health status assessment of rate gyroscopes, so as to solve the problems of uneven data volume of different modes and insufficient accuracy of health status assessment results, and achieve accurate and reliable assessment of the health status of rate gyroscopes.
[0008] This application provides a knowledge-data hybrid-driven method for accurate health status assessment of a rate gyroscope, the method comprising: Step 1: Acquire rate gyroscope data; Step 2: Design key indicators based on the rate gyroscope output information as inputs to the confidence rule base (BRB); Step 3: Build a confidence rule base; Step 4: Online inference of the BRB model; Step 5: Training data label transformation; Step Six: BRB Model Parameter Optimization. Through the optimization process, the model can automatically adjust parameters based on historical data, thereby improving the accuracy and reliability of the rate gyroscope health status assessment.
[0009] Preferably, step one includes: Rate gyroscopes operate in navigation, guidance, and attitude control systems, and their performance is affected by various factors, including environmental changes, mechanical wear, and electronic component aging. To establish a health status assessment model for rate gyroscopes, it is first necessary to collect key performance indicator data under different operating conditions and health states, including: Zero-term drift: Reflects the deviation of the gyroscope's output from zero value under static conditions, and is an important indicator for measuring the static performance of the gyroscope; First-order drift: reflects the change in the gyroscope's scale factor, affecting the dynamic measurement accuracy of the gyroscope.
[0010] Preferably, step two includes: Using the output data of the rate gyroscope, two key performance indicators, zero-term drift and first-term drift, are designed as inputs to the BRB health status assessment model. These two indicators can effectively reflect the performance changes of the rate gyroscope under different health states and are easily affected by external interference and internal aging. By collecting these indicator data and combining them with expert knowledge and historical data, an initial confidence rule base (BRB) is constructed to provide a rule basis for subsequent health status assessment.
[0011] Preferably, step three includes: Based on expert knowledge and historical operational data from the rate gyroscope, an initial confidence rule base (BRB) is constructed. This BRB integrates expert experience and data information, addresses data imbalance, and provides a rule foundation for subsequent health status assessments. Based on the model input, the k-th confidence rule of the rate gyroscope health status assessment model can be defined as:
[0012] in, This represents the observation data of the m-th index at time t. This represents the reference value of the m-th antecedent in the k-th rule; Indicates the level of health status; This represents the reference value of the nth consequent in the kth rule; Indicates the first Confidence rules The weights; express In the rules The attribute weights are as follows; initially, Determined by empirical knowledge.
[0013] Preferably, step four includes:
[0014] Where t = 1, 2, …, T, and T is the total number of sensor data; The (k+1)th and (k)th reference points are output from the m-th sensor, and are predetermined by experts; the current sensor data input of the rate gyroscope is represented as... A rule is activated when the matching degree of sensor m is non-zero; here, L represents the total number of rules in the BRB model. express For reference values The matching degree is the attribute matching degree. It can be calculated using the following method:
[0015] Next, the rule matching degree matrix is obtained as follows:
[0016] in, Based on the rule matching degree matrix The activation weight of the k-th rule It can be calculated using the following formula:
[0017] Next, all activated confidence rules are fused using an evidence reasoning parsing algorithm. The fusion process can be described as follows:
[0018] in, The fusion confidence level represents the nth health status level; finally, the output of the rate gyroscope health status assessment model can be obtained through the following utility transformation: .
[0019] Preferably, step five includes: Convert each labeled data sample into a target confidence level; .
[0020] Preferably, The calculation is as follows:
[0021] Where N is the number of output confidence scores, and T is the total number of all data.
[0022] Preferably, step six includes: By using historical data to fine-tune the unknown parameters of the BRB model, the optimization problem of the BRB model can be expressed as:
[0023] Where n represents the sample size, y i Let θ represent the observed values related to unknown external disturbances, and θ be the vector of unknown parameters to be optimized. The objective function is the Huber loss function, which is a robust loss function that exhibits squared error when the error is small and absolute error when the error is large, thus providing better robustness in handling outliers. The Huber loss function is defined as follows:
[0024] in, This is used to control the switching of the loss function between squared error and absolute error; During the optimization process, the model's parameter update strategy includes updating the confidence level, rule weights, fuzzy set parameters, and attribute weights. The specific update formulas are as follows: Confidence Update:
[0025] Rule weight Update:
[0026] Attribute weight Update:
[0027] in, It is the learning rate; This is the gradient to be calculated; Gradient calculation is crucial for parameter updates, and the specific calculation formula is as follows: Subsequent confidence gradient It can be calculated as follows:
[0028] in:
[0029] Regular weight gradient It can be calculated as follows:
[0030] in:
[0031] Attribute weight gradient It can be calculated as follows:
[0032] in:
[0033] The calculation formula is the same as above;
[0034] To maintain the interpretability of the BRB model, the parameters need to be constrained after each update: .
[0035] The beneficial technical effects of this application are as follows: This invention proposes a confidence rule base model that combines expert knowledge and historical data. By integrating multimodal data and expert experience, it effectively solves the problem of data imbalance and uses a gradient descent optimization algorithm to train the model, thereby improving the accuracy of health status assessment. Attached Figure Description
[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0037] Figure 1 A flowchart for rate gyroscope health status assessment provided in this application embodiment; Figure 2 This application provides a model structure diagram of a knowledge-data hybrid-driven rate gyroscope accurate health status assessment method according to an embodiment of the present application. Figure 3 A schematic diagram illustrating the health assessment results of the rate gyroscope using the BRB model after one tuning iteration, as provided in this embodiment of the application. Figure 4 This is a schematic diagram showing the health assessment results of the rate gyroscope by the BRB model after two optimizations, as provided in the embodiments of this application. Detailed Implementation
[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0039] The features and illustrative embodiments of various aspects of the present invention will now be described in detail. Numerous specific details are set forth in the following detailed description to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention may be practiced without requiring some of these specific details. The following description of embodiments is merely intended to provide a better understanding of the invention by illustrating examples of the invention. The invention is by no means limited to any specific setups and methods set forth below, but covers any improvements, substitutions, and modifications to structures, methods, and devices without departing from the spirit of the invention. Well-known structures and techniques are not shown in the drawings and the following description to avoid unnecessarily obscuring the invention.
[0040] In the description of this invention, it should be noted that the directions or positional relationships indicated by terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing and simplifying the invention, and should not be construed as limiting the invention. Furthermore, the use of ordinal numbers (e.g., "first and second," etc.) is for distinguishing objects and is not limited to this order, and should not be construed as indicating or implying relative importance.
[0041] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly, encompassing both direct connection and indirect connection via an intermediate medium. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances.
[0042] It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other, and the various embodiments can be referenced and cited in each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0043] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.
[0044] This invention proposes a confidence rule base model that combines expert knowledge and historical data. By integrating multimodal data and expert experience, it effectively solves the problem of data imbalance and employs a gradient descent optimization algorithm to train the model, thereby improving the accuracy of health status assessment. Although much progress has been made in health status assessment to date, its application to rate gyroscopes is still relatively limited, especially in assessment methods for multimodal data imbalance and complex fault modes, which require further research.
[0045] Please see the appendix Figure 1 The solution provided in this application mainly includes the following steps: Step 1: Rate Gyroscope Data Acquisition Rate gyroscopes operate in navigation, guidance, and attitude control systems, and their performance is affected by various factors, including environmental changes, mechanical wear, and electronic component aging. To establish a health status assessment model for rate gyroscopes, it is first necessary to collect key performance indicator data under different operating conditions and health states. These include, but are not limited to: Zero-term drift: Reflects the deviation of the gyroscope's output from zero value under static conditions, and is an important indicator for measuring the static performance of the gyroscope.
[0046] First-order drift: reflects the change in the gyroscope's scale factor, affecting the dynamic measurement accuracy of the gyroscope.
[0047] Data acquisition can be conducted through experimental testing, simulation, or on-site operational data acquisition. To ensure the effectiveness of the evaluation model, it is necessary to collect data samples covering different health states (e.g., healthy, sub-healthy, minor fault, severe fault) and different operating conditions, and to accurately label the health status of the data samples to provide a data foundation for subsequent model training and evaluation.
[0048] Step 2: Design key indicators based on the rate gyroscope output information as inputs to the confidence rule base (BRB). Using the output data of a rate gyroscope, two key performance indicators, zero-term drift and first-term drift, were designed as inputs to the BRB health status assessment model. These two indicators can effectively reflect the performance changes of the rate gyroscope under different health states and are susceptible to external interference and internal aging. By collecting these indicator data and combining them with expert knowledge and historical data, an initial confidence rule base (BRB) was constructed to provide a rule basis for subsequent health status assessments.
[0049] Step 3: Build a confidence rule base Based on expert knowledge and historical operational data from the rate gyroscope, an initial confidence rule base (BRB) is constructed. The confidence rule base is a core component of the method of this invention, used to integrate expert experience and data information, address data imbalance issues, and provide a rule-based foundation for subsequent health status assessments.
[0050] Step 4: Online Inference of the BRB Model The training and update process of the BRB model is available. Figure 2 This indicates that, through the optimization process described above, the model can automatically adjust its parameters based on historical data, thereby improving the accuracy and reliability of the rate gyroscope's health status assessment.
[0051] The feasibility of the knowledge-data hybrid-driven rate gyroscope accurate health status assessment model established in this invention can be verified through simulation experiments. It mainly includes the following steps: Step 1: Rate Gyroscope Sample Data Acquisition Experimental data were collected from the operating records of a certain model of rate gyroscope. The data includes zero-term drift, first-term drift, and corresponding health status labels. The health status labels specifically include: healthy, sub-healthy, and minor malfunction. A healthy state indicates that the zero-term drift and first-term drift fluctuate within the normal range; a sub-healthy state indicates that the zero-term drift or first-term drift slightly exceeds the normal range, but is still within acceptable limits; a minor malfunction indicates that the zero-term drift or first-term drift significantly exceeds the normal range, but the gyroscope can still function basically normally.
[0052] Step 2: BRB-based performance evaluation By analyzing the first two fields of data samples obtained by the rate gyroscope under unknown external disturbances of different intensities, the reference values of the two antecedents are set as follows:
[0053] Next, the unknown parameters of the BRB model were fined twice using a gradient descent optimization algorithm based on the Huber loss function. The optimized parameters were then tested using test data, with prediction accuracy used as the model evaluation metric. In multiple tests, the algorithm significantly improved the prediction accuracy of the BRB model, such as... Figures 3-4 As shown, the prediction accuracy of the method after the first optimization was 66.6%, and the prediction accuracy of the method after the second optimization was 97.2%.
[0054] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.
Claims
1. A knowledge-data hybrid-driven method for accurate health status assessment of a rate gyroscope, characterized in that, The method includes: Step 1: Acquire rate gyroscope data; Step 2: Design key indicators based on the rate gyroscope output information as inputs to the confidence rule base (BRB); Step 3: Build a confidence rule base; Step 4: Online inference of the BRB model; Step 5: Training data label transformation; Step Six: BRB Model Parameter Optimization. Through the optimization process, the model can automatically adjust parameters based on historical data, thereby improving the accuracy and reliability of the rate gyroscope health status assessment.
2. The method according to claim 1, characterized in that, Step one includes: Rate gyroscopes operate in navigation, guidance, and attitude control systems, and their performance is affected by various factors, including environmental changes, mechanical wear, and electronic component aging. To establish a health status assessment model for rate gyroscopes, it is first necessary to collect key performance indicator data under different operating conditions and health states, including: Zero-term drift: Reflects the deviation of the gyroscope's output from zero value under static conditions, and is an important indicator for measuring the static performance of the gyroscope; First-order drift: reflects the change in the gyroscope's scale factor, affecting the dynamic measurement accuracy of the gyroscope.
3. The method according to claim 1, characterized in that, Step two includes: Using the output data of the rate gyroscope, two key performance indicators, zero-term drift and first-term drift, are designed as inputs to the BRB health status assessment model. These two indicators can effectively reflect the performance changes of the rate gyroscope under different health states and are easily affected by external interference and internal aging. By collecting these indicator data and combining them with expert knowledge and historical data, an initial confidence rule base (BRB) is constructed to provide a rule basis for subsequent health status assessment.
4. The method according to claim 1, characterized in that, Step three includes: Based on expert knowledge and historical operational data from the rate gyroscope, an initial confidence rule base (BRB) is constructed. This BRB integrates expert experience and data information, addresses data imbalance, and provides a rule foundation for subsequent health status assessments. Based on the model input, the k-th confidence rule of the rate gyroscope health status assessment model can be defined as: in, This represents the observation data of the m-th index at time t. This represents the reference value of the m-th antecedent in the k-th rule; Indicates the level of health status; This represents the reference value of the nth consequent in the kth rule; Indicates the first Confidence rules The weights; express In the rules The attribute weights are as follows; initially, Determined by empirical knowledge.
5. The method according to claim 1, characterized in that, Step four includes: Where t = 1, 2, …, T, and T is the total number of sensor data; The (k+1)th and (k)th reference points are output from the m-th sensor, and are predetermined by the expert; the current sensor data input of the rate gyroscope is represented as... A rule is activated when the matching degree of sensor m is non-zero; here, L represents the total number of rules in the BRB model. express For reference values The matching degree is the attribute matching degree. It can be calculated using the following method: Next, the rule matching degree matrix is obtained as follows: in, Based on the rule matching degree matrix The activation weight of the k-th rule It can be calculated using the following formula: Next, all activated confidence rules are fused using an evidence reasoning parsing algorithm. The fusion process can be described as follows: in, The fusion confidence level represents the nth health status level; finally, the output of the rate gyroscope health status assessment model can be obtained through the following utility transformation: 。 6. The method according to claim 1, characterized in that, Step five includes: Convert each labeled data sample into a target confidence level; 。 7. The method according to claim 6, characterized in that, The calculation is as follows: Where N is the number of output confidence scores, and T is the total number of all data.
8. The method according to claim 1, characterized in that, Step six includes: By using historical data to fine-tune the unknown parameters of the BRB model, the optimization problem of the BRB model can be expressed as: Where n represents the sample size, y i Let θ represent the observed values related to unknown external disturbances, and θ be the vector of unknown parameters to be optimized. The objective function is the Huber loss function, which is a robust loss function that exhibits squared error when the error is small and absolute error when the error is large, thus providing better robustness in handling outliers. The Huber loss function is defined as follows: in, This is used to control the switching of the loss function between squared error and absolute error; During the optimization process, the model's parameter update strategy includes updating the confidence level, rule weights, fuzzy set parameters, and attribute weights. The specific update formulas are as follows: Confidence Update: Rule weight Update: Attribute weight Update: in, It is the learning rate; This is the gradient to be calculated; Gradient calculation is crucial for parameter updates, and the specific calculation formula is as follows: Subsequent confidence gradient It can be calculated as follows: in: Regular weight gradient It can be calculated as follows: in: Attribute weight gradient It can be calculated as follows: in: The calculation formula is the same as above; To maintain the interpretability of the BRB model, the parameters need to be constrained after each update: 。