CN117609737A - Method, system, equipment and medium for predicting health state of inertial navigation system - Google Patents
Method, system, equipment and medium for predicting health state of inertial navigation system Download PDFInfo
- Publication number
- CN117609737A CN117609737A CN202410072858.2A CN202410072858A CN117609737A CN 117609737 A CN117609737 A CN 117609737A CN 202410072858 A CN202410072858 A CN 202410072858A CN 117609737 A CN117609737 A CN 117609737A
- Authority
- CN
- China
- Prior art keywords
- data
- current period
- periodic
- inertial navigation
- navigation system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000036541 health Effects 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 68
- 230000000737 periodic effect Effects 0.000 claims abstract description 89
- 238000012360 testing method Methods 0.000 claims abstract description 63
- 230000009467 reduction Effects 0.000 claims abstract description 56
- 238000000342 Monte Carlo simulation Methods 0.000 claims abstract description 14
- 238000010801 machine learning Methods 0.000 claims abstract description 8
- 238000009826 distribution Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 230000003862 health status Effects 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 13
- 238000012706 support-vector machine Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 10
- 238000012847 principal component analysis method Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 239000000523 sample Substances 0.000 description 33
- 238000005070 sampling Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000012423 maintenance Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000013145 classification model Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000013499 data model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003449 preventive effect Effects 0.000 description 2
- 235000015842 Hesperis Nutrition 0.000 description 1
- 235000012633 Iberis amara Nutrition 0.000 description 1
- 238000000692 Student's t-test Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 229910001186 potin Inorganic materials 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012353 t test Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
- G01C25/005—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass initial alignment, calibration or starting-up of inertial devices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Manufacturing & Machinery (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Navigation (AREA)
Abstract
The invention discloses a method, a system, equipment and a medium for predicting the health state of an inertial navigation system, which relate to the field of health management of the inertial navigation system, and the method comprises the following steps: acquiring test data of the inertial navigation system in the current period; the test data is aperiodic data; the method comprises the steps of interpolating missing values in test data of a current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period; performing dimension reduction on the periodic data of the current period to obtain periodic dimension reduction data of the current period; inputting the periodic dimension reduction data of the current period into a health state prediction model to obtain the health state of the inertial navigation system in the current period; wherein, the health state prediction model is constructed based on a machine learning method. The invention can improve the accuracy of the state of health prediction of the inertial navigation system.
Description
Technical Field
The invention relates to the field of health management of inertial navigation systems, in particular to a method, a system, equipment and a medium for predicting the health state of an inertial navigation system.
Background
The inertial navigation system is used as a key component of a control system, plays roles of accurate positioning and attitude determination in a complex dynamic system, and is one of high-precision devices formed by the system. Plays a vital role in the fields of aerospace planes, carrier rockets and the like. The purpose of predicting the health state of the inertial navigation system is to comprehensively use the historical information and the test data and evaluate the overall performance and state of the system. The evaluation has important significance, can effectively evaluate the performance and situation of the system, identify the potential risk of the system, and realize fault diagnosis and preventive maintenance at low cost.
However, in practical applications of inertial navigation systems, the high value health samples available from inertial navigation systems are lacking, limited by the number of tests. Meanwhile, because the test time interval is aperiodic, the periodic continuous health state data of the inertial navigation system cannot be obtained. In the case of small amounts of data, lacking continuous detection data, the evaluation error may accumulate over time, possibly resulting in a decrease or lack of accuracy in evaluating the health status and potential problems of the device. The existence of the above problems may negatively impact the reliability, performance and maintenance policies of the device.
When the health state of the inertial navigation system is predicted, the data detected at equal intervals have better continuity and stability on the sampling frequency. The continuity and stability are beneficial to reducing noise and uncertainty of data and improving stability and accuracy of prediction, so that a more accurate model is built and a prediction effect is improved. Therefore, in order to more accurately evaluate the health status of the device, it is necessary to consider the periodic processing of non-periodic test data to compensate for the limitations caused by the limited test data and the uncertainty of the time interval.
Disclosure of Invention
Based on the above, the embodiment of the invention provides a method, a system, equipment and a medium for predicting the health state of an inertial navigation system, so as to solve the limitation caused by the limited test data and the uncertainty of time intervals, thereby improving the accuracy of the health state prediction of the inertial navigation system.
In order to achieve the above object, the embodiments of the present invention provide the following solutions.
An inertial navigation system health state prediction method, comprising: acquiring test data of the inertial navigation system in the current period; the test data are aperiodic data; the method comprises the steps of interpolating missing values in test data of a current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period; performing dimension reduction on the periodic data of the current period to obtain periodic dimension reduction data of the current period; inputting the periodic dimension reduction data of the current period into a health state prediction model to obtain the health state of the inertial navigation system in the current period; wherein the health state prediction model is constructed based on a machine learning method.
Optionally, the method for determining the health state prediction model includes: acquiring test data and corresponding health states of an inertial navigation system in a history period; interpolating missing values in the test data of the historical period by adopting a Markov Monte Carlo method to obtain periodic data of the historical period; performing dimensionality reduction on the periodic data of the historical period to obtain periodic dimensionality reduction data of the historical period; constructing training data according to the periodic dimensionality reduction data of the historical period and the corresponding health state; and training the support vector machine by adopting the training data, and determining the trained support vector machine as the health state prediction model.
Optionally, interpolating a missing value in the test data of the current period by using a markov monte carlo method to obtain periodic data of the current period, which specifically includes: adopting Mei Teluo wave Litsea-black Huntingth algorithm and a method for estimating potential scale reduction factors to interpolate missing values of any feature in test data of a current period to obtain periodic data of each feature of the current period; a parameter in the test data is used as a feature; and determining the periodic data of all the characteristics of the current period as the final periodic data of the current period.
Optionally, the Mei Teluo bose-stigmine algorithm and the method for estimating the potential scale reduction factor are adopted to interpolate the missing value of any feature in the test data of the current period to obtain the periodic data of each feature of the current period, which specifically comprises the following steps: for any feature in the test data of the current period, the process of interpolating the missing value includes: constructing an initial Markov chain which obeys stable distribution of the features according to the time sequence; determining the acceptance probability of the state transition according to the suggested distribution and the stable distribution; generating a random sample of the missing position according to the acceptance probability; inserting the random sample into the missing position of the initial Markov chain to generate a new Markov chain; judging whether the Markov new chain is converged or not by adopting a method for estimating potential scale reduction factors; if the random sample is converged, inserting the random sample into the test data of the current period to obtain the periodic data of the characteristics of the current period; if the characteristics are not converged, reconstructing an initial Markov chain which is compliant with stable distribution of the characteristics until a converged Markov new chain is generated.
Optionally, performing dimension reduction on the periodic data of the current period to obtain the periodic dimension reduction data of the current period, which specifically includes: and adopting a principal component analysis method to reduce the dimension of the periodic data in the current period to obtain the periodic dimension reduction data in the current period.
Optionally, the test data includes: the four parameter sets are a zero-order item drift coefficient set of the accelerometer, a primary item drift coefficient set of the accelerometer, a zero-order item drift coefficient set of the gyroscope and a primary item drift coefficient set of the gyroscope respectively; each type of parameter set comprises corresponding drift coefficients on different direction axes; the corresponding drift coefficient in one direction axis is used as a parameter.
The invention also provides a system for predicting the health state of the inertial navigation system, which comprises the following steps: the data acquisition module is used for acquiring test data of the inertial navigation system in the current period; the test data are aperiodic data; the interpolation module is used for interpolating the missing value in the test data of the current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period; the dimension reduction module is used for reducing the dimension of the periodic data of the current period to obtain the periodic dimension reduction data of the current period; the health state prediction module is used for inputting the periodic dimension reduction data of the current period into the health state prediction model to obtain the health state of the inertial navigation system in the current period; wherein the health state prediction model is constructed based on a machine learning method.
The invention also provides electronic equipment, which comprises a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the method for predicting the health state of the inertial navigation system.
The invention also provides a computer readable storage medium storing a computer program which when executed by a processor implements the above-described inertial navigation system health state prediction method.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: according to the embodiment of the invention, the aperiodic test data in the inertial navigation system is converted into the periodic data by adopting the Markov Monte Carlo method, the healthy state prediction model constructed by combining the data dimension reduction method and the machine learning method is based on the periodic data, the prediction of the healthy state of the inertial navigation system is realized, the limitation caused by the limited test data and the uncertainty of the time interval is solved, the accuracy of the healthy state prediction of the inertial navigation system can be improved, and the reliability, the performance and the maintenance strategy of the equipment are positively influenced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for predicting the health status of an inertial navigation system according to an embodiment of the present invention;
FIG. 2 is a graph showing the comparison of the periodicity of test data of the zero-order term drift coefficient of the X-axis of a gyroscope according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of variance ratio after PCA dimension reduction according to an embodiment of the present invention;
fig. 4 is a block diagram of an inertial navigation system health status prediction system according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
Example 1
The embodiment provides a method for predicting the health state of an inertial navigation system, aiming at aperiodic test data capable of reflecting the health state of the inertial navigation system, a Mei Teluo Bolus-Black-Potin (Metropolis-Hastings algorithm, M-H) algorithm is adopted to extract a random sample sequence from a probability distribution when direct sampling is difficult, a most likely value is used for interpolating missing values, and periodic data is obtained under the condition of retaining original data, wherein the M-H algorithm is a Markov Monte Carlo method in statistical processing. Second, to enhance the accuracy and reliability of the prediction results, principal component analysis methods (Principal Components Analysis, PCA) are utilized to reduce the dimensionality and reduce redundancy and noise in the data. Finally, a predictive model is learned and constructed by a support vector machine (Support Vector Machine, SVM) according to the dimensionality reduced periodic data so as to predict the health state of the inertial navigation system at the future moment.
The following describes the method for predicting the health status of the inertial navigation system according to the present embodiment in detail.
Referring to fig. 1, the method for predicting the health status of the inertial navigation system according to the present embodiment includes the following steps.
Step 101: acquiring test data of the inertial navigation system in the current period; the test data is non-periodic data.
The accelerometer and the gyroscope are used as important components of the inertial navigation system and are arranged in the XYZ axis direction of the inertial navigation system. The navigation precision of the inertial navigation system is mainly related to zero-order drift coefficients and one-order drift coefficients of the accelerometer and the gyroscope, and the values of the zero-order drift coefficients and the one-order drift coefficients are influenced by various factors, such as uncertainty of equipment manufacturing, change of environmental conditions and the like. Thus, these drift coefficients can be considered as a random process, the values of which change over time.
Thus, the test data of the present embodiment includes: the four parameter sets are a zero-order item drift coefficient set of the accelerometer, a primary item drift coefficient set of the accelerometer, a zero-order item drift coefficient set of the gyroscope and a primary item drift coefficient set of the gyroscope respectively; each type of parameter set comprises corresponding drift coefficients on different direction axes; the corresponding drift coefficient in one direction axis is used as a parameter.
Step 102: the method comprises the following steps of interpolating missing values in test data of a current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period.
(1) Adopting Mei Teluo Bolus-Black-Ting algorithm and method of estimating potential scale reduction factor (Estimated Potential Scale Reduction, EPSR) to interpolate missing value of any feature in the test data of the current period to obtain periodic data of each feature of the current period; one parameter in the test data is taken as a feature.
Specifically, for any feature in the test data of the current period, the process of interpolating the missing value includes: constructing an initial Markov chain which obeys stable distribution of the features according to the time sequence; determining the acceptance probability of the state transition according to the suggested distribution and the stable distribution; generating a random sample of the missing position according to the acceptance probability; inserting the random sample into the missing position of the initial Markov chain to generate a new Markov chain. And judging whether the Markov new chain is converged or not by adopting a method for estimating the potential scale reduction factor. If the random sample is converged, inserting the random sample into the test data of the current period to obtain the periodic data of the characteristics of the current period; if the characteristics are not converged, reconstructing an initial Markov chain which is compliant with stable distribution of the characteristics until a converged Markov new chain is generated.
(2) And determining the periodic data of all the characteristics of the current period as the final periodic data of the current period.
Periodic data is obtained by using a Mei Teluo Bolisi-Black-Tinstas algorithm, model training is performed by using complete data after data periodic interpolation, so that the prediction performance of a model can be improved, but compared with the original data, the interpolated sample may slightly reduce the interpretability and the robustness of the model. In order to verify that the interpolated data can converge on the original data model, a method of estimating a potential scale reduction factor is also employed to determine whether the samples converge on the probability distribution of the original data.
Step 103: and performing dimension reduction on the periodic data of the current period to obtain the periodic dimension reduction data of the current period.
Specifically, in high-dimensional data, potential correlation often exists among dimensions, so that the problems of high data redundancy and strong correlation are caused. Meanwhile, the noise accompanied by high-dimensional data can reduce the efficiency and performance of data analysis, and increase the complexity of problem analysis. One solution to this problem is to pre-process the data by PCA. PCA can effectively find the main direction of data, remove the dimension which does not store much information but carries noise, and greatly improve the prediction performance of the model. PCA can eliminate noise of high-dimensional data while retaining main information of the data, reduce the dimension of the data as much as possible, play a role in avoiding information loss and reducing the calculated amount, and greatly improve the efficiency of data analysis and the accuracy of results.
Therefore, the present embodiment adopts PCA to reduce the dimension of the periodic data of the current period, so as to obtain the periodic dimension reduction data of the current period.
Step 104: and inputting the periodic dimension reduction data of the current period into a health state prediction model to obtain the health state of the inertial navigation system in the current period.
Wherein the health state prediction model is constructed based on a machine learning method.
In step 104, the method for determining the health status prediction model includes the following steps.
(1) Test data and corresponding health states of the inertial navigation system in a history period are obtained.
(2) And interpolating missing values in the test data of the historical period by adopting a Markov Monte Carlo method to obtain periodic data of the historical period. The interpolation process in this step is the same as the interpolation process in step 102, and will not be described here.
(3) And performing dimensionality reduction on the periodic data of the historical period to obtain the periodic dimensionality reduction data of the historical period. The dimension reduction method in this step is the same as the dimension reduction method in step 103, and will not be described here again.
(4) Training data is constructed from the periodic reduced dimension data of the historical period and the corresponding health status.
(5) And training the support vector machine by adopting the training data, and determining the trained support vector machine as the health state prediction model.
The support vector machine model is suitable for processing high-dimensional data with a small sample size. And learning and predicting the health state of the inertial navigation system by using a multi-classification support vector machine, and predicting the health state of the inertial navigation system at the future moment by classifying the health state of the inertial navigation system. By using the established health state prediction model, the health state of the inertial navigation system can be judged, the overall performance of the equipment is further evaluated, potential risks possibly existing are identified, and a corresponding preventive maintenance strategy is formulated.
The following focuses on the process of aperiodic data periodicity, data dimension reduction by using PCA, and health state prediction model establishment.
1. The non-periodic data is periodic.
In inertial navigation system, detection data x of accelerometer, zero order drift coefficient and one order drift coefficient of gyroscope under non-periodic condition (t ) (t=0, 1, 2.) may be approximated as a continuous random process. Due to x in random sequence (t +1) Is only equal to x (t ) Related to other states in the random sequence. Thus, the original non-periodic data can be periodic by the following steps.
1) Setting each feature of inertial navigation system to establish Markov chain [ x ] according to time sequence (0) , x (1) , x (2) ,...]. With the continuous increase of time, after generating a sufficiently long Markov chain, the Markov chain is subjected to stable distributionX is a sample randomly sampled from the markov chain.
2) Setting the suggested distribution asThe meaning of the proposed distribution is that the sampled value is from +.>Transfer to x (t ) Is a function of the probability of (1),is the division state x in the Markov chain (t ) Other states than those described above. The selection of the suggested distribution is subjectively determined by a person according to the characteristics of the data distribution.
3) According to a smooth distributionObtaining the acceptance probability->Is a calculation formula of (2).
（1）
Wherein,for sampling value->Smooth distribution of compliance,/->For the sampling value x (t ) A smooth distribution of compliance.From x for the sampled value (t ) Transfer to->Probability of->Characterization sample value from +.>Transfer to x (t ) Is a probability of (2).
4) Generating random numbers subject to an average distribution。
The process is carried out by the steps of, （2）
5) In the case of an inertial navigation system having I features, the steps 2) to 4) are set to be repeated C times to obtain C random samples, each random sample is inserted into an initial markov chain of each feature, and each feature generates C new markov chains. Wherein,markov new chain generated after inserting sample for ith characteristic original Markov chainThere are S samples, +.>An s-th sample representing the c-th new markov chain in the i-th feature, wherein,. Using EPSR algorithm to judge the admission sample +.>Whether the post model converges. The judgment process is as follows.
(1) Calculating in-chain meanMean->。
（3）
（4）
(2) Each intra-chain variance B and inter-chain variance W are calculated.
（5）
（6）
(3) For the post-inspection square differenceAn estimation is made.
（7）
(4) Calculation of EPSR values。
（8）
Generally when EPSR <1.1, then the target is considered to converge, C random samples are interpolated into the original dataset. Looping steps 2) -5) until the original data is periodic.
2. PCA is used for data dimension reduction.
1) The raw data is normalized.
Let the matrix X be made up of N I-dimensional periodic sample data sets, each row representing a set of I-eigenvalue sample data.
That is to say, （9）
（10）
normalizing the matrix X to obtain a normalized matrix，/>The elements in (2) are shown in the formula (11) and the formula (12).
（11）
（12）
In the method, in the process of the invention,is the average value of the samples S i Is the standard deviation of the sample, x ni Is X i N element of (a)>Is->N-th element of (a) in the above-mentioned sequence.
2) The eigenvalues and eigenvectors are calculated.
Let u be the unit vector, for ease of representation, the matrixX for the nth data of (2) n Representing, let matrix->Is the inner product of the nth row data and u. The total variance of the N sets of sample data after being respectively inner-integrated with u is shown in formula (13).
（13）
Wherein,let +.>Is covariance matrix->U is the corresponding eigenvector, and T represents the transpose.
3) And selecting a main component.
The I eigenvalues of the covariance matrix are set as shown in equation (14).
（14）
Selecting feature vectors of the first M principal components with accumulated variance contribution rate greater than 85% according to feature value ordering。
4) The principal component is calculated.
Feature vectorThe principal component is calculated as a standard orthonormal basis as shown in formula (15).
（15）
Wherein Y is m For the M-th principal component (m=1, 2, M), (u m1 ,u m2 ,...,u mi ) The orthonormal basis representing the mth principal component, the principal component linear expression is shown in equation (16).
（16）
3. And establishing a prediction model.
The health status reference levels are "healthy", "sub-healthy" and "unhealthy", respectively. The multi-classification SVM algorithm is used to convert the multi-classification into a plurality of two-classification. This translates the data model with 3 classifications intoAnd (5) classifying the models. And establishing a prediction model according to the dimensionality reduced data, so as to carry out classified prediction on the new data.
The data set after the main component is extracted is provided with M samples, wherein the L samples are divided into training sets, the rest are used as test sets, and the training samples are shown in a formula (17).
（17）
Wherein L is the number of samples,is->Training samples->Is->Classification labels of individual training samples.
1) And constructing an optimal hyperplane.
Assuming that two classifications of the classification model are p and q, the hyperplane equation is shown in equation (18).
（18）
After substituting all samples and their labels into equation (19), the hyperplane can correctly classify the sample data if the following conditions are satisfied, ω and b are parameters of the model, ω represents the weight, and b represents the bias.
（19）
Support vectorIs the heterogeneous sample closest to the hyperplane among the samples.
The distance d between the ultrasonic probe and the hyperplane is: /> （20）
2) And solving a classification function.
Establishing an objective functionAnd solving for the parameter ω.
（21）
In the formula (21), C is a penalty factor,for relaxation variable, ++>And->Determines the allowable error of sample classification, +.>Is->Relaxation variables corresponding to the training samples, +.>Is->Lagrangian multipliers corresponding to the training samples. And (3) carrying out optimization solution on the formula (21) by using a Lagrangian multiplier method to obtain a new objective function Q (a).
（22）
In the formula (22), Q (a) represents an allowable error,is->Lagrangian multiplier corresponding to each training sample,>is->Lagrangian multiplier corresponding to each training sample,>to remove->The numbers of training samples other than K are kernel functions,>is->Training samples->Is->Classification labels of individual training samples. And solving to obtain the optimal values of the parameters omega and b, and determining a hyperplane equation, thereby establishing a prediction model.
An application example is given below to verify the validity of the above-mentioned inertial navigation system health state prediction method.
The health state of the inertial navigation system is embodied by zero-order drift coefficients and one-order drift coefficients of the gyroscope and the accelerometer. According to the actual situation, 37 sets of drift test data are collected, which have the problems of small sample size and unequal sampling intervals. The original data is obtained by zero-order drift coefficient and one-order drift coefficient of the accelerometer) Zero-order drift coefficient and one-order drift coefficient of gyroscope) Composition is prepared. Since the minimum sampling interval in the original data is one month, the data is interpolated here as periodic detection data with an interval of one month. Wherein K is 0X 、K 0Y 、K 0Z The zero-order drift coefficients, K, of the accelerometer are respectively corresponding to the X axis, the Y axis and the Z axis 1X 、K 1Y 、K 1Z The primary drift coefficients corresponding to the X axis, the Y axis and the Z axis of the accelerometer are D 0X 、D 0Y 、D 0Z The zero-order item drift coefficients corresponding to the X axis, the Y axis and the Z axis of the gyroscope are D 1X 、D 1Y 、D 1Z The primary term drift coefficients corresponding to the X axis, the Y axis and the Z axis of the gyroscope respectively.
Step 1: the original non-periodic data is periodic.
The original 37 groups of drift test data are interpolated into 50 groups of equal period data through an M-H algorithm and an EPSR method.
Taking the test data of the zero-order item drift coefficient of the X axis of the gyroscope as an example, the interpolation effect is shown in fig. 2, and as can be seen from fig. 2, the data after interpolation (periodic data) and the data before interpolation (original non-periodic data) are compared, and the data after periodicity accords with the change trend of the original data. The similarity of the two sets of data under different indicators was compared using a statistical t-test, the results are shown in table 1.
TABLE 1 differential analysis
Taking the zero-order drift coefficient of the X-axis gyroscope as an example. T-test was performed between the two sets of data before and after interpolation to obtain the results of h=0 and p=0.8002. According to conventional practices in statistics, the original hypothesis cannot be rejected, i.e., there is no statistically significant difference between the two sets of data, typically when p > 0.05.
Step 2: PCA is used for data dimension reduction.
And selecting the main components with the accumulated variance contribution rate greater than 85% as the first 3 main components through a PCA algorithm. The 12-dimensional data is converted into 3-dimensional data. Since PCA dimension reduction is a method of maximizing variance based on the theory of minimizing projection errors. Whether the new data subjected to PCA dimension reduction contains main information in the original data or not can be supported to be sample data for establishing a multi-classification support vector machine model.
Through calculation, the projection error after PCA dimension reduction is 4.768 multiplied by 10 -29 . The reconstruction error is close to zero, which indicates that the reduced-dimension data is very close to the original data, and almost no information is lost. In addition, the degree of dispersion of the numerical values can be expressed in terms of variance of the data. As shown in fig. 3, the variance of the reduced-dimension data is analyzed as a proportion of the variance of the original data.
The variance ratio of each principal component after dimension reduction represents the contribution degree of each principal component to the total variance. As can be seen from fig. 3, the variance of the first three principal components is relatively large, and the reduced-dimension data can better retain the information of the original data. In summary, the variance is relatively large and the reconstruction error is small. The reduced-dimension data can be considered to have the characteristics of saving most of information of the original data, being capable of well restoring the structure and the characteristics of the original data and using fewer dimensions to represent key variances of the original data. These features make the reduced-size data more compact and easier to process and analyze.
Step 3: and establishing a prediction model.
And 2, obtaining a data set after PCA dimension reduction, wherein 70% (35 groups) of new data are randomly selected as training samples for training a predictive model based on a multi-classification support vector machine. The remaining 30% (15 groups) of data were used as test samples for evaluating the predictive effect of the predictive model. And respectively predicting the original data without periodicity after dimension reduction and directly predicting to verify the validity of the method.
In practical application, the state of 'unhealthy' detected by the inertial navigation system during the health state detection is more important than the 'healthy' and 'sub-healthy' states. When the validity of the prediction method is checked, setting a positive sample as a sample with a label of 'unhealthy'; negative samples are samples labeled 'healthy' or 'sub-healthy', indicating that the sample does not belong to a healthy state. Since the recall refers to the proportion of the number of samples of the model correctly predicted as positive samples to all the true positive samples, a high recall means that the model can better identify positive samples. F1-score integrates accuracy and recall rate, and not only considers the capturing capability of the model to positive samples, but also considers the prediction accuracy of the model to negative samples. Compared with the accuracy and the recall, the F1-score can comprehensively evaluate the performance of the classification model.
As shown in table 2: when the method for predicting the health state of the inertial navigation system is used, the accuracy of the model is 86.67%. This means that the model predicts a higher proportion of the correct samples to the total number of samples and the prediction results are more accurate. The recall rate was 100.00%, indicating that the model was able to capture all true positive samples completely without missing any one. The F1-score was 100.00%, indicating that the model performed excellent in recognition of positive samples and prediction accuracy of negative samples.
Table 2 sample evaluation results
Example two
In order to execute the corresponding method of the above embodiment to achieve the corresponding functions and technical effects, a system for predicting the health status of an inertial navigation system is provided below.
Referring to fig. 4, the system includes: a data acquisition module 201, configured to acquire test data of a current period of the inertial navigation system; the test data is non-periodic data. The interpolation module 202 is configured to interpolate the missing value in the test data of the current period by using a markov monte carlo method, so as to obtain periodic data of the current period. The dimension reduction module 203 is configured to reduce the dimension of the periodic data in the current period, so as to obtain the periodic dimension reduction data in the current period. The health state prediction module 204 is configured to input the periodic reduced-dimension data of the current period into a health state prediction model to obtain the health state of the inertial navigation system in the current period. Wherein the health state prediction model is constructed based on a machine learning method.
Example III
The embodiment provides an electronic device, including a memory and a processor, where the memory is configured to store a computer program, and the processor runs the computer program to enable the electronic device to execute the method for predicting the health state of an inertial navigation system according to the first embodiment.
Alternatively, the electronic device may be a server.
In addition, the embodiment of the invention also provides a computer readable storage medium, which stores a computer program, and the computer program realizes the method for predicting the health state of the inertial navigation system according to the first embodiment when being executed by a processor.
The above embodiments solve the limitation caused by the limited test data and the uncertainty of the time interval, and can improve the accuracy, stability and efficiency of the state of health prediction of the inertial navigation system, thereby positively affecting the reliability, performance and maintenance strategy of the device.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described in this specification with reference to specific examples, the description of which is only for the purpose of aiding in understanding the method of the present invention and its core ideas; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. A method for predicting the health status of an inertial navigation system, comprising:
acquiring test data of the inertial navigation system in the current period; the test data are aperiodic data;
the method comprises the steps of interpolating missing values in test data of a current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period;
performing dimension reduction on the periodic data of the current period to obtain periodic dimension reduction data of the current period;
inputting the periodic dimension reduction data of the current period into a health state prediction model to obtain the health state of the inertial navigation system in the current period;
wherein the health state prediction model is constructed based on a machine learning method;
the method for obtaining the periodic data of the current period by interpolating the missing value in the test data of the current period by adopting a Markov Monte Carlo method comprises the following steps:
adopting Mei Teluo wave Litsea-black Huntingth algorithm and a method for estimating potential scale reduction factors to interpolate missing values of any feature in test data of a current period to obtain periodic data of each feature of the current period; a parameter in the test data is used as a feature;
and determining the periodic data of all the characteristics of the current period as the final periodic data of the current period.
2. The inertial navigation system health state prediction method according to claim 1, wherein the method of determining the health state prediction model comprises:
acquiring test data and corresponding health states of an inertial navigation system in a history period;
interpolating missing values in the test data of the historical period by adopting a Markov Monte Carlo method to obtain periodic data of the historical period;
performing dimensionality reduction on the periodic data of the historical period to obtain periodic dimensionality reduction data of the historical period;
constructing training data according to the periodic dimensionality reduction data of the historical period and the corresponding health state;
and training the support vector machine by adopting the training data, and determining the trained support vector machine as the health state prediction model.
3. The method for predicting the health status of an inertial navigation system according to claim 1, wherein the method for obtaining the periodic data of each feature in the current period by interpolating the missing value of any feature in the test data in the current period by adopting Mei Teluo bos-nistins algorithm and a method for estimating a potential scale reduction factor specifically comprises:
for any feature in the test data of the current period, the process of interpolating the missing value includes:
constructing an initial Markov chain which obeys stable distribution of the features according to the time sequence;
determining the acceptance probability of the state transition according to the suggested distribution and the stable distribution;
generating a random sample of the missing position according to the acceptance probability;
inserting the random sample into the missing position of the initial Markov chain to generate a new Markov chain;
judging whether the Markov new chain is converged or not by adopting a method for estimating potential scale reduction factors;
if the random sample is converged, inserting the random sample into the test data of the current period to obtain the periodic data of the characteristics of the current period; if the characteristics are not converged, reconstructing an initial Markov chain which is compliant with stable distribution of the characteristics until a converged Markov new chain is generated.
4. The method for predicting the health status of an inertial navigation system according to claim 1, wherein the step of reducing the periodic data of the current period of time to obtain the periodic reduced data of the current period of time comprises:
and adopting a principal component analysis method to reduce the dimension of the periodic data in the current period to obtain the periodic dimension reduction data in the current period.
5. The inertial navigation system health prediction method of claim 1, wherein the test data comprises: the four parameter sets are a zero-order item drift coefficient set of the accelerometer, a primary item drift coefficient set of the accelerometer, a zero-order item drift coefficient set of the gyroscope and a primary item drift coefficient set of the gyroscope respectively; each type of parameter set comprises corresponding drift coefficients on different direction axes; the corresponding drift coefficient in one direction axis is used as a parameter.
6. An inertial navigation system health state prediction system, comprising:
the data acquisition module is used for acquiring test data of the inertial navigation system in the current period; the test data are aperiodic data;
the interpolation module is used for interpolating the missing value in the test data of the current period by adopting a Markov Monte Carlo method to obtain periodic data of the current period;
the dimension reduction module is used for reducing the dimension of the periodic data of the current period to obtain the periodic dimension reduction data of the current period;
the health state prediction module is used for inputting the periodic dimension reduction data of the current period into the health state prediction model to obtain the health state of the inertial navigation system in the current period;
wherein the health state prediction model is constructed based on a machine learning method.
7. An electronic device comprising a memory for storing a computer program and a processor that runs the computer program to cause the electronic device to perform the inertial navigation system health prediction method of any one of claims 1 to 5.
8. A computer readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the inertial navigation system health state prediction method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410072858.2A CN117609737B (en) | 2024-01-18 | 2024-01-18 | Method, system, equipment and medium for predicting health state of inertial navigation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410072858.2A CN117609737B (en) | 2024-01-18 | 2024-01-18 | Method, system, equipment and medium for predicting health state of inertial navigation system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117609737A true CN117609737A (en) | 2024-02-27 |
| CN117609737B CN117609737B (en) | 2024-03-19 |
Family
ID=89948303
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410072858.2A Active CN117609737B (en) | 2024-01-18 | 2024-01-18 | Method, system, equipment and medium for predicting health state of inertial navigation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117609737B (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100024546A1 (en) * | 2007-07-31 | 2010-02-04 | The Boeing Company | Disc resonator integral inertial measurement unit |
| US20110004110A1 (en) * | 2000-05-30 | 2011-01-06 | Vladimir Shusterman | Personalized Monitoring and Healthcare Information Management Using Physiological Basis Functions |
| US20140372063A1 (en) * | 2012-03-06 | 2014-12-18 | Wuhan University | Quick calibration method for inertial measurement unit |
| US20180153444A1 (en) * | 2016-12-05 | 2018-06-07 | Intel Corporation | Body movement tracking |
| WO2018120428A1 (en) * | 2016-12-29 | 2018-07-05 | 平安科技（深圳）有限公司 | Personalized scenario prediction method, apparatus, device and storage medium |
| WO2019047593A1 (en) * | 2017-09-05 | 2019-03-14 | 百度在线网络技术（北京）有限公司 | Method and device for processing automatic driving training data |
| CN109883429A (en) * | 2019-04-15 | 2019-06-14 | 山东建筑大学 | Zero-speed detection method and indoor pedestrian's inertial navigation system based on Hidden Markov Model |
| BR102018003279A2 (en) * | 2018-02-20 | 2019-09-10 | Rosemount Aerospace Inc | inertial unit of measurement and method |
| WO2020087846A1 (en) * | 2018-10-31 | 2020-05-07 | 东南大学 | Navigation method based on iteratively extended kalman filter fusion inertia and monocular vision |
| CN115980667A (en) * | 2022-10-14 | 2023-04-18 | 上海交通大学 | Positioning method based on hidden Markov model and improved Viterbi algorithm |
| CN116629831A (en) * | 2023-05-15 | 2023-08-22 | 南京理工大学 | Switch machine health management method and system based on hidden semi-Markov model |
| CN116933643A (en) * | 2023-07-25 | 2023-10-24 | 泰州历帆科技有限公司 | Intelligent data monitoring method based on partial robust M regression and multiple interpolation |
| CN117269742A (en) * | 2023-08-28 | 2023-12-22 | 中国电力科学研究院有限公司 | Method, device and medium for evaluating health state of circuit breaker in high-altitude environment |
-
2024
- 2024-01-18 CN CN202410072858.2A patent/CN117609737B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110004110A1 (en) * | 2000-05-30 | 2011-01-06 | Vladimir Shusterman | Personalized Monitoring and Healthcare Information Management Using Physiological Basis Functions |
| US20100024546A1 (en) * | 2007-07-31 | 2010-02-04 | The Boeing Company | Disc resonator integral inertial measurement unit |
| US20140372063A1 (en) * | 2012-03-06 | 2014-12-18 | Wuhan University | Quick calibration method for inertial measurement unit |
| US20180153444A1 (en) * | 2016-12-05 | 2018-06-07 | Intel Corporation | Body movement tracking |
| WO2018120428A1 (en) * | 2016-12-29 | 2018-07-05 | 平安科技（深圳）有限公司 | Personalized scenario prediction method, apparatus, device and storage medium |
| WO2019047593A1 (en) * | 2017-09-05 | 2019-03-14 | 百度在线网络技术（北京）有限公司 | Method and device for processing automatic driving training data |
| BR102018003279A2 (en) * | 2018-02-20 | 2019-09-10 | Rosemount Aerospace Inc | inertial unit of measurement and method |
| WO2020087846A1 (en) * | 2018-10-31 | 2020-05-07 | 东南大学 | Navigation method based on iteratively extended kalman filter fusion inertia and monocular vision |
| CN109883429A (en) * | 2019-04-15 | 2019-06-14 | 山东建筑大学 | Zero-speed detection method and indoor pedestrian's inertial navigation system based on Hidden Markov Model |
| CN115980667A (en) * | 2022-10-14 | 2023-04-18 | 上海交通大学 | Positioning method based on hidden Markov model and improved Viterbi algorithm |
| CN116629831A (en) * | 2023-05-15 | 2023-08-22 | 南京理工大学 | Switch machine health management method and system based on hidden semi-Markov model |
| CN116933643A (en) * | 2023-07-25 | 2023-10-24 | 泰州历帆科技有限公司 | Intelligent data monitoring method based on partial robust M regression and multiple interpolation |
| CN117269742A (en) * | 2023-08-28 | 2023-12-22 | 中国电力科学研究院有限公司 | Method, device and medium for evaluating health state of circuit breaker in high-altitude environment |
Non-Patent Citations (2)
| Title |
|---|
| 王刚;陈捷;洪荣晶;王华;: "基于HMM和优化的PF的数控转台精度衰退模型", 振动与冲击, no. 06, 28 March 2018 (2018-03-28) * |
| 赵申坤;姜潮;龙湘云;: "一种基于数据驱动和贝叶斯理论的机械系统剩余寿命预测方法", 机械工程学报, no. 12, 20 June 2018 (2018-06-20) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117609737B (en) | 2024-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3620983B1 (en) | Computer-implemented method, computer program product and system for data analysis | |
| US12001949B2 (en) | Computer-implemented method, computer program product and system for data analysis | |
| US10387768B2 (en) | Enhanced restricted boltzmann machine with prognosibility regularization for prognostics and health assessment | |
| CN110427654B (en) | Landslide prediction model construction method and system based on sensitive state | |
| US8892478B1 (en) | Adaptive model training system and method | |
| US8712929B1 (en) | Dynamic data filtering system and method | |
| JP7126256B2 (en) | Abnormality diagnosis device, abnormality diagnosis method, and program | |
| CN108919059A (en) | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing | |
| JP2001502831A (en) | A method for classifying the statistical dependence of measurable time series | |
| CN114297036A (en) | Data processing method and device, electronic equipment and readable storage medium | |
| CN111612262A (en) | Wind power probability prediction method based on quantile regression | |
| CN115979310A (en) | Inertial navigation system performance degradation evaluation method and system, electronic device and medium | |
| CN117150402A (en) | Power data anomaly detection method and model based on generation type countermeasure network | |
| Khlamov et al. | The astronomical object recognition and its near-zero motion detection in series of images by in situ modeling | |
| Wang et al. | Assessing the Performance Degradation of Lithium‐Ion Batteries Using an Approach Based on Fusion of Multiple Feature Parameters | |
| CN117609737B (en) | Method, system, equipment and medium for predicting health state of inertial navigation system | |
| CN112880726A (en) | Sensor fault diagnosis method and device based on variational modal decomposition sample entropy | |
| CN116757533A (en) | Industrial equipment abnormality detection method and related device | |
| Catanach et al. | Bayesian updating and uncertainty quantification using sequential tempered mcmc with the rank-one modified metropolis algorithm | |
| Lu et al. | Weak monotonicity with trend analysis for unsupervised feature evaluation | |
| CN115375038A (en) | Aircraft engine failure mode identification and service life prediction method | |
| CN114397521A (en) | Fault diagnosis method and system for electronic equipment | |
| Khlamov et al. | Statistical Modeling for the Near-zero Apparent Motion Detection of Objects in Series of Images From data Stream | |
| CN115249059A (en) | Model training and abnormal data analysis method and device and computer storage medium | |
| Huang et al. | Fault prognosis using deep convolutional neural network and bootstrap-based method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |