CN117272022A - Detection method of MEMS oscillator - Google Patents
Detection method of MEMS oscillator Download PDFInfo
- Publication number
- CN117272022A CN117272022A CN202311216458.6A CN202311216458A CN117272022A CN 117272022 A CN117272022 A CN 117272022A CN 202311216458 A CN202311216458 A CN 202311216458A CN 117272022 A CN117272022 A CN 117272022A
- Authority
- CN
- China
- Prior art keywords
- mems
- data
- generate
- oscillation
- oscillator
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 59
- 230000010355 oscillation Effects 0.000 claims abstract description 232
- 238000001228 spectrum Methods 0.000 claims abstract description 99
- 238000000034 method Methods 0.000 claims abstract description 79
- 238000012545 processing Methods 0.000 claims abstract description 74
- 230000003595 spectral effect Effects 0.000 claims abstract description 74
- 238000004458 analytical method Methods 0.000 claims abstract description 47
- 230000000116 mitigating effect Effects 0.000 claims abstract description 33
- 238000005457 optimization Methods 0.000 claims abstract description 31
- 238000000547 structure data Methods 0.000 claims abstract description 30
- 238000004088 simulation Methods 0.000 claims abstract description 24
- 238000010276 construction Methods 0.000 claims abstract description 18
- 238000013507 mapping Methods 0.000 claims description 71
- 239000011159 matrix material Substances 0.000 claims description 58
- 230000007613 environmental effect Effects 0.000 claims description 53
- 238000004422 calculation algorithm Methods 0.000 claims description 42
- 230000004044 response Effects 0.000 claims description 36
- 230000003044 adaptive effect Effects 0.000 claims description 33
- 238000010586 diagram Methods 0.000 claims description 33
- 230000006870 function Effects 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 16
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000005516 engineering process Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 230000010354 integration Effects 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 9
- 230000002787 reinforcement Effects 0.000 claims description 9
- 238000012512 characterization method Methods 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 8
- 238000012800 visualization Methods 0.000 claims description 8
- 230000015572 biosynthetic process Effects 0.000 claims description 7
- 238000005305 interferometry Methods 0.000 claims description 7
- 238000003786 synthesis reaction Methods 0.000 claims description 7
- 230000001427 coherent effect Effects 0.000 claims description 5
- 125000004122 cyclic group Chemical group 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 238000010801 machine learning Methods 0.000 abstract description 5
- 206010000117 Abnormal behaviour Diseases 0.000 abstract 1
- 230000006399 behavior Effects 0.000 description 19
- 230000008569 process Effects 0.000 description 16
- 238000012549 training Methods 0.000 description 10
- 230000008859 change Effects 0.000 description 7
- 238000009826 distribution Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 6
- 238000013136 deep learning model Methods 0.000 description 6
- 239000000284 extract Substances 0.000 description 6
- 239000013078 crystal Substances 0.000 description 5
- 230000001687 destabilization Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000001914 filtration Methods 0.000 description 5
- 238000000513 principal component analysis Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000009977 dual effect Effects 0.000 description 4
- 238000012880 independent component analysis Methods 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 230000002452 interceptive effect Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 3
- 238000013500 data storage Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000000368 destabilizing effect Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000011524 similarity measure Methods 0.000 description 3
- 238000009966 trimming Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000005312 nonlinear dynamic Methods 0.000 description 2
- 230000009022 nonlinear effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000007670 refining Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000010937 topological data analysis Methods 0.000 description 1
- 238000012795 verification 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
- 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
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The invention relates to the technical field of machine learning, in particular to a detection method of an MEMS oscillator. The method comprises the following steps: the method comprises the steps of obtaining oscillation signal data of an MEMS oscillator, carrying out oscillation nonlinear spectrum construction and spectral domain topology analysis processing on the oscillation signal data of the MEMS oscillator to generate MEMS dynamic spectrum characteristic structure data, carrying out self-adaptive modal detection and feedback logic construction on the MEMS dynamic spectrum characteristic structure data to generate an MEMS real-time feedback rule set, carrying out environment simulation calibration and oscillation dynamic interference decoupling based on the MEMS real-time feedback rule set to generate an oscillation interference mitigation strategy, and carrying out self-adaptive adjustment optimization on the MEMS oscillator by utilizing the oscillation interference mitigation strategy to generate oscillator self-adjusting parameters. By means of machine learning. The invention can monitor and predict the performance and state of the MEMS oscillator in real time. This helps to find the abnormal behavior of the oscillator early.
Description
Technical Field
The invention relates to the technical field of machine learning, in particular to a detection method of an MEMS oscillator.
Background
MEMS oscillators are miniature oscillators and are commonly used in various electronic devices such as clock circuits, communication devices, inertial navigation systems, and the like. Due to their small size, low power consumption and high frequency, they are widely used in modern electronic products. Therefore, the performance and reliability of the oscillator is critical to the stable operation of the entire electronic system. Machine learning techniques have great potential in the detection methods of MEMS oscillators, which can monitor the performance of the oscillator by analyzing a large amount of sensor data, predict potential problems, optimize the tuning strategy of the oscillator, to ensure the reliability and performance of the device. The MEMS oscillator data may be subject to noise, interference, and uncertainty, which may lead to instability or inaccuracy of the trained model.
Disclosure of Invention
The present invention provides a method for detecting a MEMS oscillator, so as to solve at least one of the above technical problems.
To achieve the above object, the present invention provides a method for detecting a MEMS oscillator, the method comprising the steps of:
step S1: acquiring oscillation signal data of the MEMS oscillator, and carrying out oscillation nonlinear spectrum construction on the oscillation signal data of the MEMS oscillator to generate an MEMS nonlinear oscillation spectrum;
step S2: performing spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data;
step S3: performing adaptive mode detection on the MEMS oscillator through MEMS dynamic spectrum characteristic structure data to generate MEMS oscillation mode detection data;
step S4: performing feedback logic construction based on MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set;
step S5: performing environmental simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data;
step S6: according to the MEMS oscillation calibration data, oscillation dynamic interference decoupling is carried out so as to generate an oscillation interference alleviation strategy;
step S7: and carrying out self-adaptive adjustment optimization on the MEMS oscillator by utilizing an oscillation interference mitigation strategy so as to generate oscillator self-tuning parameters.
The invention provides a detection method of an MEMS oscillator, which is used for acquiring oscillation signal data of the MEMS oscillator, carrying out oscillation nonlinear revealing processing on the oscillation signal data of the MEMS oscillator so as to generate an MEMS nonlinear oscillation map, and deeply resolving the oscillation signal data of the MEMS oscillator through the oscillation nonlinear revealing processing. This allows the nonlinear characteristics of the oscillator to be revealed and is not limited to fundamental frequency and amplitude analysis alone. This deep resolution helps to understand the complex oscillation behavior of MEMS oscillators, including nonlinear effects and instability phenomena. And carrying out spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data, wherein the nonlinear dynamic spectrum characteristic structure data of the MEMS oscillator can be used for detecting the instability condition of the oscillator. By analyzing the variation trend of the spectrum characteristics, possible instability modes can be found in advance, so that measures are taken to avoid system faults or reduce damage risks, adaptive mode identification is carried out on MEMS dynamic spectrum characteristic structure data by deep learning to generate MEMS oscillation mode detection data, and the step can be used for adaptively identifying the mode of the MEMS oscillator by the deep learning technology. This means that it is able to capture with high accuracy the vibration behaviour of the oscillator in different modes, including vibration frequency, vibration mode and amplitude. This helps to more fully understand the dynamic performance of the oscillator. And carrying out feedback logic construction based on the MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set, and realizing real-time oscillator performance monitoring and adjustment by constructing feedback logic based on the MEMS oscillation mode detection data. This helps to ensure that the MEMS oscillator always maintains optimal performance under different operating conditions, thereby improving the real-time response and adaptability of the system. And performing environment simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data, and performing environment simulation calibration based on the MEMS real-time feedback rule set to enable the MEMS oscillator to show better adaptability under different environment conditions. This means that the oscillator can maintain stability and performance in various application scenarios without frequent recalibration or adjustment. And according to the MEMS oscillation calibration data, the oscillation dynamic interference decoupling is carried out to generate an oscillation interference mitigation strategy, and the oscillator is easily affected by external interference under specific working conditions, so that instability is caused. The interference decoupling strategy of the step helps to reduce this risk, ensures that the oscillator can maintain stable oscillation performance under different environments, and thus enhances the stability of the system. The self-adaptive adjustment optimization is carried out on the MEMS oscillator by utilizing the oscillation interference alleviation strategy so as to generate the self-adjusting parameters of the oscillator, and the performance stability of the MEMS oscillator can be obviously improved by applying the oscillation interference alleviation strategy. By detecting and responding to oscillation disturbances in real time, this strategy helps to maintain the stability of the oscillator under different operating conditions, thereby reducing the risk of performance fluctuations and system failures.
Drawings
FIG. 1 is a schematic flow chart of steps of a method for detecting a MEMS oscillator according to the present invention;
FIG. 2 is a detailed implementation step flow diagram of step S1;
fig. 3 is a detailed implementation step flow diagram of step S2.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the application provides a detection method of a MEMS oscillator. The execution subject of the detection method of the MEMS oscillator includes, but is not limited to, the system being mounted with: mechanical devices, data processing platforms, cloud server nodes, network transmission devices, etc. may be considered general purpose computing nodes of the present application. The data processing platform includes, but is not limited to: at least one of an audio management system, an image management system and an information management system.
Referring to fig. 1 to 3, the present invention provides a method for detecting a MEMS oscillator, the method comprising the following steps:
step S1: acquiring oscillation signal data of the MEMS oscillator, and carrying out oscillation nonlinear spectrum construction on the oscillation signal data of the MEMS oscillator to generate an MEMS nonlinear oscillation spectrum;
step S2: performing spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data;
Step S3: performing adaptive mode detection on the MEMS oscillator through MEMS dynamic spectrum characteristic structure data to generate MEMS oscillation mode detection data;
step S4: performing feedback logic construction based on MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set;
step S5: performing environmental simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data;
step S6: according to the MEMS oscillation calibration data, oscillation dynamic interference decoupling is carried out so as to generate an oscillation interference alleviation strategy;
step S7: and carrying out self-adaptive adjustment optimization on the MEMS oscillator by utilizing an oscillation interference mitigation strategy so as to generate oscillator self-tuning parameters.
The invention provides a detection method of an MEMS oscillator, which is used for acquiring oscillation signal data of the MEMS oscillator, carrying out oscillation nonlinear revealing processing on the oscillation signal data of the MEMS oscillator so as to generate an MEMS nonlinear oscillation map, and deeply resolving the oscillation signal data of the MEMS oscillator through the oscillation nonlinear revealing processing. This allows the nonlinear characteristics of the oscillator to be revealed and is not limited to fundamental frequency and amplitude analysis alone. This deep resolution helps to understand the complex oscillation behavior of MEMS oscillators, including nonlinear effects and instability phenomena. And carrying out spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data, wherein the nonlinear dynamic spectrum characteristic structure data of the MEMS oscillator can be used for detecting the instability condition of the oscillator. By analyzing the variation trend of the spectrum characteristics, possible instability modes can be found in advance, so that measures are taken to avoid system faults or reduce damage risks, adaptive mode identification is carried out on MEMS dynamic spectrum characteristic structure data by deep learning to generate MEMS oscillation mode detection data, and the step can be used for adaptively identifying the mode of the MEMS oscillator by the deep learning technology. This means that it is able to capture with high accuracy the vibration behaviour of the oscillator in different modes, including vibration frequency, vibration mode and amplitude. This helps to more fully understand the dynamic performance of the oscillator. And carrying out feedback logic construction based on the MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set, and realizing real-time oscillator performance monitoring and adjustment by constructing feedback logic based on the MEMS oscillation mode detection data. This helps to ensure that the MEMS oscillator always maintains optimal performance under different operating conditions, thereby improving the real-time response and adaptability of the system. And performing environment simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data, and performing environment simulation calibration based on the MEMS real-time feedback rule set to enable the MEMS oscillator to show better adaptability under different environment conditions. This means that the oscillator can maintain stability and performance in various application scenarios without frequent recalibration or adjustment. And according to the MEMS oscillation calibration data, the oscillation dynamic interference decoupling is carried out to generate an oscillation interference mitigation strategy, and the oscillator is easily affected by external interference under specific working conditions, so that instability is caused. The interference decoupling strategy of the step helps to reduce this risk, ensures that the oscillator can maintain stable oscillation performance under different environments, and thus enhances the stability of the system. The self-adaptive adjustment optimization is carried out on the MEMS oscillator by utilizing the oscillation interference alleviation strategy so as to generate the self-adjusting parameters of the oscillator, and the performance stability of the MEMS oscillator can be obviously improved by applying the oscillation interference alleviation strategy. By detecting and responding to oscillation disturbances in real time, this strategy helps to maintain the stability of the oscillator under different operating conditions, thereby reducing the risk of performance fluctuations and system failures.
In the embodiment of the present invention, as described with reference to fig. 1, the step flow diagram of a method for detecting a MEMS oscillator according to the present invention is shown, and in this example, the method for detecting a MEMS oscillator includes the following steps:
step S1: acquiring oscillation signal data of the MEMS oscillator, and carrying out oscillation nonlinear spectrum construction on the oscillation signal data of the MEMS oscillator to generate an MEMS nonlinear oscillation spectrum;
in the embodiment of the invention, the oscillation signal data of the MEMS oscillator is obtained, the dynamic instability interferometry is carried out on the oscillation signal data of the MEMS oscillator to generate MEMS instability signal frequency band data, and the entropy coding nonlinear disclosure is carried out on the MEMS instability signal frequency band data to generate a MEMS nonlinear entropy characteristic diagram.
Step S2: performing spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data;
in the embodiment of the invention, a time spectrum domain enhancement processing is performed through a MEMS nonlinear oscillation map to generate MEMS enhancement time spectrum data, the MEMS enhancement time spectrum data is calculated by utilizing a spectrum domain feature integral function to generate MEMS spectrum domain feature data, feature network topology mapping is performed based on the MEMS spectrum domain feature data to generate a MEMS spectrum domain feature network map, an MEMS oscillator is obtained, and in combination with the MEMS spectrum domain feature network map, dynamic topology structure analysis is performed on the MEMS oscillator to generate an oscillator structure topology map, and feature projection and structuring are performed based on the oscillator structure topology map to generate MEMS dynamic spectrum feature structure data.
Step S3: performing adaptive mode detection on the MEMS oscillator through MEMS dynamic spectrum characteristic structure data to generate MEMS oscillation mode detection data;
in the embodiment of the invention, the MEMS dynamic spectrum characteristic structural data is calculated by utilizing a multidimensional spectrum characteristic popular conversion formula to generate MEMS spectrum characteristic manifold data, the MEMS spectrum characteristic manifold data is subjected to self-adaptive modal coding to generate MEMS modal characteristic data, the MEMS modal characteristic data is subjected to time sequence integration processing by applying a cyclic neural network to generate MEMS time sequence modal characteristic data, self-adaptive pooling processing is performed on the basis of the MEMS time sequence modal characteristic data to generate an MESM self-adaptive modal matrix, an MEMS oscillator is acquired, and the MEMS oscillator is subjected to oscillation modal identification through the MESM self-adaptive modal matrix to generate MEMS oscillation modal detection data.
Step S4: performing feedback logic construction based on MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set;
in the embodiment of the invention, data characterization processing is performed based on MEMS oscillation mode detection data to generate heterogeneous mode feature mapping data, high-dimensional feedback path embedding is performed by utilizing the heterogeneous mode feature mapping data to generate high-dimensional feedback characteristic data, feature logic entropy coding is performed on the high-dimensional feedback characteristic data through an information entropy principle to generate logic entropy coding mapping data, feedback topology network construction is performed based on the logic entropy coding mapping data to generate a topology feedback relation matrix, and real-time feedback rule fusion is performed according to the topology feedback relation matrix to generate an MEMS real-time feedback rule set.
Step S5: performing environmental simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data;
in the embodiment of the invention, the MEMS real-time feedback rule set is clustered by utilizing a distributed clustering algorithm to generate MEMS environment characteristic data, virtual environment synthesis is performed based on the MEMS environment characteristic data to generate MEMS environment simulation data, response deviation detection is performed through the MEMS environment simulation data to generate a response deviation report, calibration rule programming is performed in combination with the response deviation report to generate a dynamic calibration rule set, and the MEMS oscillator is precisely calibrated in real time by utilizing the dynamic calibration rule set to generate MEMS oscillation calibration data.
Step S6: and performing oscillation dynamic interference decoupling according to the MEMS oscillation calibration data to generate an oscillation interference mitigation strategy.
In the embodiment of the invention, multidimensional interference identification processing is carried out on MEMS oscillation calibration data to generate oscillation interference identification data, data mode conversion is carried out on the oscillation interference identification data to generate an interference mode index chart, network interference sensing is carried out through the interference mode index chart to generate an interference source data report, self-adaptive interference modeling is carried out by combining the interference mode index chart and the interference source data report to generate a dynamic interference prediction model, and real-time interference mitigation processing is carried out on the MEMS oscillator based on the dynamic interference prediction model to generate an oscillation interference mitigation strategy.
Step S7: and carrying out self-adaptive adjustment optimization on the MEMS oscillator by utilizing an oscillation interference mitigation strategy so as to generate oscillator self-tuning parameters.
In the embodiment of the invention, the oscillation interference alleviation strategy is subjected to deep feature extraction to generate an oscillation feature matrix, the oscillation feature matrix is subjected to strategy learning optimization processing by adopting a deep reinforcement learning model to generate an oscillation optimization strategy, and the MEMS oscillator is subjected to parameter self-adaptive fine adjustment by utilizing the oscillation optimization strategy to generate the self-adjusting parameter of the oscillator.
Preferably, step S1 comprises the following steps;
step S11: acquiring oscillation signal data of the MEMS oscillator, and performing dynamic instability interferometry on the oscillation signal data of the MEMS oscillator to generate MEMS instability signal frequency band data;
step S12: performing entropy coding nonlinear disclosure on MEMS unstable signal frequency band data to generate an MEMS nonlinear entropy characteristic diagram;
step S13: performing double-frequency coherent enhancement processing on the MEMS nonlinear entropy feature map by adopting a double-frequency technology to generate MEMS double-frequency enhancement mapping data;
step S14: and performing three-dimensional mapping visualization based on the MEMS dual-frequency enhanced mapping data to generate an MEMS nonlinear oscillation map.
According to the invention, the LED crystal grains are obtained, the LED crystal grains comprise LED crystal grain current and light intensity, a target frequency band of the work of the LED crystal grains is selected, frequency spectrum filtering adjustment is performed to generate LED crystal grain frequency spectrum signal data, and the oscillation signal data of the MEMS oscillator are accurately obtained through modes such as a sensor, digital sampling and the like. The method is helpful to ensure the high quality of the acquired signals, thereby improving the accuracy of subsequent analysis, and the vibration behavior of the MEMS oscillator can be monitored in real time and the change of the interference pattern can be recorded by adopting a dynamic instability interferometry technology. This helps to capture the destabilization phenomena of the oscillator, including frequency drift, phase changes, etc. And performing entropy coding nonlinear disclosure on the MEMS unstable signal frequency band data to generate an MEMS nonlinear entropy characteristic diagram, wherein the application of a nonlinear disclosure algorithm is beneficial to capturing nonlinear characteristics and structures in the MEMS unstable signal frequency band data. These features may include periodicity, noise, emergencies, etc., which are important for fault detection and analysis, and by entropy encoding, the original MEMS destabilizing signal band data may be compressed into a more compact symbol sequence, thereby reducing the cost of data storage and transmission. This helps to efficiently manage large amounts of data under limited resource conditions. And performing double-frequency coherence enhancement processing on the MEMS nonlinear entropy feature map by adopting a double-frequency technology to generate MEMS double-frequency enhancement mapping data, and extracting finer information from the nonlinear entropy feature map of the MEMS oscillator by adopting the double-frequency coherence enhancement technology. The method comprises the step of carrying out coherence analysis on different frequency components, further refining the nonlinear characteristics in the characteristic diagram, enabling the nonlinear characteristics to be more expressive, and strengthening the nonlinear characteristics in a double-frequency range through double-frequency coherence enhancement processing. This means that the nonlinear behaviour of the oscillator will be more pronounced and clearly manifest in different frequency ranges, helping to understand its working mechanism more fully. And carrying out three-dimensional mapping visualization based on the MEMS dual-frequency enhancement mapping data to generate an MEMS nonlinear oscillation map, and presenting the nonlinear characteristic of the MEMS oscillator through the three-dimensional mapping visualization. This allows the nonlinear oscillation behavior to be visualized in an intuitive way without the need for extensive mathematical analysis. This is of great value to engineers, researchers and decision makers, as they can quickly learn about the vibration mode of the oscillator, the destabilization situation and other important features.
As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S1 in fig. 1 is shown, where step S1 includes:
step S11: acquiring oscillation signal data of the MEMS oscillator, and performing dynamic instability interferometry on the oscillation signal data of the MEMS oscillator to generate MEMS instability signal frequency band data;
in the present example, first, in this step, oscillation signal data is collected from the MEMS oscillator. The data may include information such as amplitude, phase, time series, etc. Methods of acquiring the oscillating signal data may include sensor measurements, signal amplification, and digitized sampling, among others. Next, the oscillation signal data is analyzed using dynamic destabilization interferometry. This technique is generally based on the principle of optical interference, using devices such as interferometers to monitor the vibration behaviour of the oscillator. Specific operations include introducing an oscillating signal into an interferometer and recording changes in the interference pattern, followed by analysis of the oscillating signal data using dynamic destabilization interferometry techniques. This technique is generally based on the principle of optical interference, using devices such as interferometers to monitor the vibration behaviour of the oscillator. Specific operations include introducing an oscillating signal into an interferometer and recording changes in the interference pattern.
Step S12: performing entropy coding nonlinear disclosure on MEMS unstable signal frequency band data to generate an MEMS nonlinear entropy characteristic diagram;
in the embodiment of the invention, the acquired MEMS destabilizing signal frequency band data needs to be subjected to detailed data analysis. This includes statistical analysis, frequency domain analysis, time domain analysis, etc. to understand the distribution and characteristics of the data, and to select the entropy coding algorithm suitable for the MEMS destabilizing signal frequency band data. For example, huffman coding, an algorithm commonly used for data compression, may be selected. The algorithm should be selected based on the distribution of the data and the compression efficiency. And performing entropy coding on the analyzed frequency band data. This involves mapping data to a sequence of symbols to reduce the volume of the data while retaining important information. The encoded data will have a higher data density suitable for subsequent processing. And applying a nonlinear feature revealing algorithm to the entropy coded data. This may include applying nonlinear mapping, wavelet transformation, or adaptive filtering methods to capture nonlinear patterns and structures in the data, and finally constructing a MEMS nonlinear entropy characterization map using the revealed nonlinear features. This feature map is a matrix of pixels, where each pixel represents the distribution in space of the nonlinear features of the data. The nonlinear disclosure of the entropy coding is realized through the steps, so that a MEMS nonlinear entropy characteristic diagram is generated.
Step S13: performing double-frequency coherent enhancement processing on the MEMS nonlinear entropy feature map by adopting a double-frequency technology to generate MEMS double-frequency enhancement mapping data;
in the embodiment of the present invention, first, the MEMS nonlinear entropy feature map obtained in step S12 is used as input data. The signature contains information about the nonlinear characteristics of the MEMS oscillator, expressed in an entropy encoded manner. In this step, a suitable dual frequency technique is selected to process the non-linear entropy feature map. The dual-frequency technique may be a signal processing method for enhancing the performance of nonlinear characteristics. Wherein the selection is to use dual frequency coherent processing. And performing double-frequency coherence enhancement processing on the nonlinear entropy characteristic diagram. This can be accomplished by the steps of: the nonlinear entropy characteristic diagram is converted into a frequency domain, and a Fourier transform or wavelet transform method and the like are generally used, so that a double-frequency coherent analysis method is adopted to carry out coherence analysis on different frequency components in frequency domain data. This helps to identify non-linear characteristics in different frequency ranges, and a dual-frequency coherence analysis method is used to perform coherence analysis on different frequency components in the frequency domain data. This helps to identify non-linear characteristics over different frequency ranges. And generating MEMS dual-frequency enhancement mapping data through dual-frequency coherence enhancement processing. This data contains an enhanced representation of the nonlinear characteristics over the dual frequency range, helping to more accurately analyze the nonlinear behavior of the MEMS oscillator. The double-frequency coherence enhancement processing is realized through the steps so as to generate MEMS double-frequency enhancement mapping data.
Step S14: and performing three-dimensional mapping visualization based on the MEMS dual-frequency enhanced mapping data to generate an MEMS nonlinear oscillation map.
In the embodiment of the invention, the MEMS dual-frequency enhancement maps data. These data include nonlinear characteristics of the oscillating signal that are processed by a dual frequency technique to enhance their performance. This dataset contains information about the nonlinear behavior of the MEMS oscillator, selecting the appropriate three-dimensional mapping visualization tool or library. This may include the use of tools such as Matplotlib. And mapping the MEMS dual-frequency enhancement mapping data into a three-dimensional space. This mapping process may involve the following steps: if the dimension of the MEMS dual-frequency enhancement map data is high, dimension reduction techniques, such as principal component analysis, can be used to reduce the data to three dimensions or lower. Visual parameters such as coordinate axis range, color mapping, illumination effect and the like are set to ensure that a visual map can accurately reflect the nonlinear characteristics of the MEMS oscillator, and the processed MEMS dual-frequency enhanced mapping data is visualized into a three-dimensional map by using a selected three-dimensional mapping visualization tool. This profile will exhibit nonlinear oscillation behavior of the MEMS oscillator and in an intuitive manner, the generated three-dimensional profile is analyzed to identify characteristics and instability modes of the oscillator. The three-dimensional mapping visualization is realized through the steps so as to generate the MEMS nonlinear oscillation map.
Preferably, step S2 comprises the steps of:
step S21: performing time spectrum domain enhancement processing through the MEMS nonlinear oscillation spectrum to generate MEMS enhancement time spectrum data;
step S22: calculating the spectrum data during the MEMS enhancement by utilizing the spectrum domain feature comprehensive function to generate MEMS spectrum domain feature data;
step S23: performing feature network topology mapping based on the MEMS spectral domain feature data to generate an MEMS spectral domain feature network map;
step S24: acquiring an MEMS oscillator, and carrying out dynamic topological structure analysis on the MEMS oscillator by combining with the MEMS spectral domain characteristic network diagram so as to generate an oscillator structural topological diagram;
step S25: and performing characteristic projection and structuring based on the topological graph of the oscillator structure to generate MEMS dynamic spectrum characteristic structure data.
The invention carries out time spectrum domain enhancement processing through the MEMS nonlinear oscillation spectrum to generate MEMS enhanced time spectrum data, and obtains high-resolution time spectrum data through processing the original oscillation signal data. This helps to more clearly capture the frequency characteristics of the oscillator, including instantaneous frequency and amplitude variations, improves understanding of the oscillator behaviour, and spectral domain enhancement techniques such as wavelet packet transforms or Gabor filters can highlight specific frequency components of the oscillating signal. In this way, the key features in the oscillating signal are more pronounced, including the periodicity and destabilization modes of the frequency components. This is critical for analyzing and identifying the characteristics of the oscillator. And performing feature network topology mapping based on the MEMS spectral domain feature data to generate an MEMS spectral domain feature network graph, and mapping correlation between the spectral domain features to edges in the network graph. The weights of these edges reflect the strength of correlation between features, enabling the user to quickly identify and understand which features have a strong correlation between them, providing clues for further analysis and improvement. And acquiring the MEMS oscillator, and carrying out dynamic topological structure analysis on the MEMS oscillator by combining with the MEMS spectral domain characteristic network diagram to generate an oscillator structure topological diagram, wherein the correlation and the mutual influence between nodes in the oscillator can be known through the dynamic topological structure analysis, and the complex structure and the operation mechanism of the oscillator are disclosed. This is of great importance for the design, optimization and fault diagnosis of the oscillator, and by analysing the topology, bottleneck or critical nodes in the oscillator structure can be identified. This information can be used to optimize the performance of the oscillator, improving the key performance indicators of oscillation frequency, stability and output power. And performing characteristic projection and structuring based on the oscillator structure topological graph to generate MEMS dynamic spectrum characteristic structure data, and correlating the spectrum domain characteristics of the MEMS oscillator with the structure topological graph. Through this association, the dynamic characteristics of the oscillator, including the interrelationship between the features and the structure of the oscillator, can be more fully understood, with multiple benefits using a graph roll-up neural network (GCN) as the feature projection method. The GCN can effectively capture the relationships between nodes and thus can better model interactions between elements inside the oscillator. This helps to improve understanding and modeling accuracy of the oscillator behavior.
As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:
step S21: performing time spectrum domain enhancement processing through the MEMS nonlinear oscillation spectrum to generate MEMS enhancement time spectrum data;
in the embodiment of the invention, first, raw oscillation signal data is acquired from the MEMS oscillator, which can be acquired by a sensor or the like. The raw data comprises an oscillating signal in the time domain. The raw data is divided into small segments and a window function is applied to reduce the effects of spectral leakage. Common window functions include hanning windows, hamming windows, and the like. Each windowed signal segment is fourier transformed and converted to the frequency domain. This will result in data of the frequency domain representation, including amplitude and phase information, in the frequency domain, employing specific spectral domain enhancement techniques, such as wavelet packet transforms or Gabor filters, etc. The techniques can emphasize specific frequency components of the oscillating signal, improve resolution of the time spectrum, and extract nonlinear entropy features from the enhanced time spectrum data. This may be achieved by applying a non-linear analysis method, such as SampEn (Sample Entropy). And extracting nonlinear entropy features from the enhanced time-frequency spectrum data. This may be achieved by applying a non-linear analysis method, such as SampEn (Sample Entropy). The time-spectrum domain enhancement processing is realized through the steps to generate MEMS enhanced time-spectrum data.
Step S22: calculating the spectrum data during the MEMS enhancement by utilizing the spectrum domain feature comprehensive function to generate MEMS spectrum domain feature data;
preferably, the spectral domain feature synthesis function in step S22 is as follows:
wherein F refers to MEMS spectral domain feature data, x i Refers to the representation of the original data in the time-frequency spectrum domain, n refers to the total number of data points, delta phi i Refers to the offset of the ith data point relative to the average value, T refers to the duration of the whole data, θ refers to the spectral domain angle factor, Δψ i Refers to the phase difference of the ith data point, and λ refers to the spectral domain intensity factor.
The invention creates a spectrum domain characteristic comprehensive function, which is mainly used for calculating MEMS spectrum domain characteristic data F, wherein the characteristic data reflects important information of original data in a time spectrum domain. Specifically, it extracts key features about frequency, phase, intensity, offset from the average value, and the like from the time-frequency domain data. θ, λ, Δphi in the formula i 、Δψ i The isoparametric introduces a multidimensional analysis allowing for more detailed and comprehensive feature extraction of the time-spectrum data. This helps to better understand the time-frequency characteristics of the raw data. By synthesizing the spectral data during analysis, the formula provides a more accurate analysis of oscillator performance. This helps make more informed decisions in oscillator design and optimization to meet the needs of a particular application. S in the formula in(θx i +λΔψ i ) Nonlinear operations are introduced to help extract nonlinear features from the raw data. This allows mining very subtle patterns and information in the raw data that may not be captured in the traditional linear approach, delta phi i The parameter represents the offset of each data point from the average. These offsets can be used to model the relative changes in the data to better understand the fluctuations in the raw data. In the formulaRepresenting the differential operation on the spectral data, and helping to extract spectral features. These features can be used to analyze the frequency distribution of the oscillator.
Step S23: performing feature network topology mapping based on the MEMS spectral domain feature data to generate an MEMS spectral domain feature network map;
in the embodiment of the invention, an innovative feature network topology mapping method is adopted, and the method aims at mapping spectral domain feature data of the MEMS oscillator into a network structure so as to better understand the relation among features. We have chosen an algorithm based on complex network analysis that can identify complex correlations between features, creating a null network graph in which nodes represent the spectral domain features of the MEMS oscillator. Each feature will become a node in the network graph. And establishing the edge relation according to the correlation between the spectral domain features. We will employ a specific similarity measure that will map the correlation to the edges in the network graph. The weights of the edges are distributed according to the correlation strength between the features, and by establishing the correlation edges between the features, we will obtain a complete feature network diagram reflecting the topology structure between the features of the MEMS oscillator spectrum domain. This graph can be regarded as a visual representation of complex relationships between features. Finally, the generated network map is saved as a MEMS spectral domain signature network map, which will be used for subsequent analysis to reveal complex relationships and structures between oscillator signatures. The characteristic network topology mapping is realized through the steps so as to generate the MEMS spectral domain characteristic network diagram.
Step S24: and acquiring the MEMS oscillator, and carrying out dynamic topological structure analysis on the MEMS oscillator by combining the MEMS spectral domain characteristic network diagram so as to generate an oscillator structural topological diagram.
In an embodiment of the invention, a MEMS oscillator to be analyzed is obtained. This may be achieved by a standard MEMS oscillator acquisition device or sensor. The MEMS spectral domain signature network map generated in step S23 is combined with the data of the actual MEMS oscillator. This can be accomplished by mapping the oscillator data to the characteristic network nodes to construct a comprehensive dataset, and using specialized topology analysis algorithms, such as analysis methods based on complex network theory, to analyze the dynamic topology of the MEMS oscillator in detail. This process includes the steps of: determining nodes in the MEMS oscillator, wherein the nodes can represent key structural elements or characteristics, determining connecting edges between the nodes, representing interrelationships between the nodes, extracting topological characteristics such as the degree, the centrality and the like of the nodes to measure the importance of the nodes in a network, and analyzing the evolution of the nodes and the edges in time to obtain dynamic topological information. And generating a structural topological graph of the oscillator according to the result of the dynamic topological analysis. The graph will include information of node, edge and topology characteristics reflecting the dynamic evolution process of the oscillator structure. The dynamic topology analysis is realized through the steps to generate the topological diagram of the oscillator structure.
Step S25: and performing characteristic projection and structuring based on the topological graph of the oscillator structure to generate MEMS dynamic spectrum characteristic structure data.
In the embodiment of the invention, an advanced characteristic projection method is selected to correlate the spectral domain characteristics of the MEMS oscillator with the structural topological graph. We use a graph roll-up neural network (Graph Convolutional Network, GCN) as the feature projection method. GCN is a deep learning method for graph data that can effectively capture the relationships between nodes. The spectral domain signature data of the oscillator is combined with the structural topology map data into a joint dataset. Each node represents a characteristic or topological element of the oscillator, and the GCN is applied to the combined dataset. Specifically, we learn the feature propagation between nodes using GCNs. This involves the following steps: the GCN layer is applied to the federated data set to perform feature propagation on the graph. Each GCN layer combines the features of the node with the features of its neighboring nodes to generate a new feature representation, and applies the GCN layer to the joint dataset to perform feature propagation on the graph. Each GCN layer combines the features of the node with the features of its neighboring nodes to generate a new feature representation, which we have obtained by multi-layer stacking of GCNs, where each node represents a feature or topological element of the oscillator and information about the propagation of the feature is contained between these nodes. Finally, MEMS dynamic spectral feature structure data is extracted from the structured feature representation output by the GCN. This data contains information about the topology of the characteristics of the oscillator. The characteristic projection and structuring can be realized through the steps so as to generate MEMS dynamic spectrum characteristic structure data.
Preferably, step S3 comprises the steps of:
step S31: calculating MEMS dynamic spectrum feature structure data by utilizing a multidimensional spectrum feature popular conversion formula to generate MEMS spectrum feature manifold data;
step S32: performing adaptive mode coding on MEMS spectral feature manifold data to generate MEMS mode feature data;
step S33: performing time sequence integration processing on the MEMS modal feature data by using a cyclic neural network to generate MEMS time sequence modal feature data;
step S34: performing self-adaptive pooling processing based on MEMS time sequence modal characteristic data to generate an MESM self-adaptive modal matrix;
step S35: and acquiring the MEMS oscillator, and performing oscillation mode detection on the MEMS oscillator through the MESM self-adaptive mode matrix to generate MEMS oscillation mode detection data.
The invention generates MEMS modal characteristic data by carrying out self-adaptive modal coding on MEMS spectral characteristic manifold data, and associates the spectral domain characteristics of the MEMS oscillator with a structural topological graph. Through this association, the dynamic characteristics of the oscillator, including the interrelationship between the features and the structure of the oscillator, can be more fully understood, with multiple benefits using a graph roll-up neural network (GCN) as the feature projection method. The GCN can effectively capture the relationships between nodes and thus can better model interactions between elements inside the oscillator. This helps to improve understanding and modeling accuracy of the oscillator behavior. And performing time sequence integration processing on the MEMS modal feature data by using the cyclic neural network to generate MEMS time sequence modal feature data, wherein the time sequence integration processing enables the features of each mode to be combined with the features of the front and rear time. This integration helps to correlate information between different modalities, providing a more comprehensive view. This may reveal interactions and associations between different modalities, helping to better understand the complexity of the oscillator, RNN is a powerful neural network structure with excellent capabilities in processing time series data. By applying RNNs to MEMS modal feature data, one of the benefits is the ability to capture timing dependencies between modal features. This helps to understand the dynamic behavior and response of the MEMS oscillator, thereby better analyzing and predicting its performance. And carrying out self-adaptive pooling processing based on the MEMS time sequence modal characteristic data to generate an MESM self-adaptive modal matrix, wherein the self-adaptive pooling method is beneficial to extracting important information from the time sequence modal characteristic data and refining the characteristic description of the oscillator. By calculating the adaptive weights, important features of the oscillator within different time windows are highlighted, thereby better capturing the dynamic properties of the oscillator. The MEMS oscillator is obtained, the MEMS oscillator is subjected to oscillation mode identification through the MESM self-adaptive mode matrix so as to generate MEMS oscillation mode detection data, and the MEMS oscillator is subjected to oscillation mode identification through the MESM self-adaptive mode matrix, so that the mode identification efficiency of the MEMS oscillator can be remarkably improved. This facilitates accurate determination of the modal characteristics of the oscillator in a shorter time, thereby facilitating faster tuning and improvement of the oscillator performance.
Step S31: calculating MEMS dynamic spectrum feature structure data by utilizing a multidimensional spectrum feature popular conversion formula to generate MEMS spectrum feature manifold data;
preferably, the multi-dimensional spectral feature popular transformation formula in step S31 is as follows:
wherein G refers to MEMS spectral feature manifold structure data, S refers to input spectral features, R refers to a reference function for adding nonlinear characteristics, m refers to the total number of data points in the input MEMS dynamic spectral feature structure data, gamma refers to an exponential function, t refers to a time variable, S j Refers to the jth data point in the spectral signature, delta refers to the manifold weight constant, and beta refers to a nonlinear adjustment factor that adjusts the intensity of the spectral signature.
The invention creates a multi-dimensional spectral feature epidemic transformation formula for generating MEMS spectral feature manifold structural data G. These manifold data contain important information about the dynamics of the MEMS oscillator. In particular, they reflect the complex structure of the oscillator spectral features, including information in terms of frequency distribution, amplitude, phase, etc. Integration of multi-dimensional spectral features in a formula, these spectral features s j Are data points in the MEMS dynamic spectral feature data. By integrating these features, the generated manifold data more fully reflects the characteristics of the oscillator, including a plurality of frequency components and phase information. The parameters gamma, delta and beta represent the index of the exponential function, the manifold weight and the nonlinear adjustment factor, respectively. By adjusting the values of these parameters, the shape, strength, and strength of the nonlinear characteristics of the manifold can be flexibly controlled. The adjustment can be self-adaptive according to different oscillator behaviors, so that the method is more applicable, and the intensity and angle of manifold are affected. The selection of these parameters can be adjusted according to specific analysis requirements to ensure that the generated MEMS spectral signature manifold data accurately reflects the dynamic characteristics of the oscillator. The beta parameter in the formula acts as a nonlinear adjustment factor that can be used to adjust the nonlinear intensity of the spectral features. This allows the method to be adapted to the specific requirements of the oscillator.
Step S32: performing adaptive mode coding on MEMS spectral feature manifold data to generate MEMS mode feature data;
in the embodiment of the present invention, the MEMS spectral signature manifold data obtained from step S31 will be used as input data, and an efficient adaptive mode coding algorithm, such as an auto encoder (Autoencoder), is used. An automatic encoder is a deep learning model that contains encoder and decoder parts. The encoder encodes the input data into a potential representation and the decoder restores the potential representation to the input data. The goal of such algorithms is to learn a compact, high-dimensional representation of features to minimize the information loss of the input data. The MEMS spectral feature manifold data is standardized to ensure that the data has zero mean and unit variance, the network architecture of the adaptive modal encoder is designed, including the hierarchical structure of the encoder and decoder, the number of neurons, etc., and a large number of MEMS spectral feature manifold data is used to train the adaptive modal encoder. During training, the model learns the effective feature representation by minimizing reconstruction errors. By passing the MEMS spectral signature manifold data through the encoder, a potential signature representation is generated, which is MEMS modal signature data, including neuron activation values for a layer in the encoder network, signature values for each modality. The adaptive modal coding is realized through the steps so as to generate MEMS modal characteristic data.
Step S33: performing time sequence integration processing on the MEMS modal feature data by using a cyclic neural network to generate MEMS time sequence modal feature data;
in the embodiment of the invention, the generated MEMS modal characteristic data is prepared. These data may include feature values for individual modalities, with a Recurrent Neural Network (RNN) applied to process the modality feature data. In particular, we can choose to use a suitable RNN architecture, such as long short term memory network (LSTM). A key feature of RNNs is their ability to capture timing information. We apply RNNs to the modality feature data in order to combine the features of each modality with the features of its time of day. This will help to build timing relationships between the modality features, providing sufficient data for the RNN model to train. This includes inputting modality characteristic data and corresponding timing information, and corresponding targets, if any. The goal of training is to enable the RNN model to learn and capture the time-series dependencies between modality features. Once the RNN model is trained, we can use it to process new modality feature data. The RNN will generate time-sequential modality characteristic data for each modality, which contains time-sequential information of modality characteristics. The time sequence integration processing is realized through the steps so as to generate MEMS time sequence modal characteristic data.
Step S34: performing self-adaptive pooling processing based on MEMS time sequence modal characteristic data to generate an MESM self-adaptive modal matrix;
in the embodiment of the present invention, the generated MEMS timing mode feature data is obtained from the previous step S33. The self-adaptive pooling is an effective feature fusion method for extracting important information from time sequence modal feature data and generating an MESM self-adaptive modal matrix. The following are the detailed steps: the time sequence modal characteristic data is divided into non-overlapping time windows. Each time window contains a period of time sequence data, so that the whole period of the oscillator is covered, and the time sequence modal characteristic data is divided into non-overlapping time windows. Each time window contains a period of time series data that ensures that the entire period of the oscillator is covered, and adaptive weights are calculated for the modal characteristics within each time window. This may be calculated based on the importance of the feature or other factors. The adaptive weights help better capture the characteristics of the oscillator and calculate the adaptive weights for the modal characteristics within each time window. This may be calculated based on the importance of the feature or other factors. The adaptive weights help to better capture the characteristics of the oscillator. And combining the fusion results of the time windows into a MESM self-adaptive modal matrix. This matrix contains characterization of the oscillator in different time windows to better reflect the dynamic properties of the oscillator. The self-adaptive pooling processing is realized through the steps so as to generate the MESM self-adaptive modal matrix.
Step S35: and acquiring the MEMS oscillator, and performing oscillation mode detection on the MEMS oscillator through the MESM self-adaptive mode matrix to generate MEMS oscillation mode detection data.
In the embodiment of the invention, the MEMS oscillator to be identified is obtained. This can be acquired by physical devices or from a data store, and the generated MESM adaptive modal matrix must be prepared before oscillation mode identification can take place. This matrix contains important information of the modal characteristics. The specific steps for identifying the oscillation mode comprise: first, the data collected from the MEMS oscillator needs to be preprocessed, including denoising, filtering, normalization, etc. This ensures the quality and consistency of the input data, and first, requires pre-processing, including denoising, filtering, normalization, etc., of the data acquired from the MEMS oscillator. This ensures the quality and consistency of the input data, by applying a matching algorithm, the features extracted from the oscillator data are matched to the modal features in the MESM adaptive modal matrix. This may involve similarity calculation between modality features and generation of a matching score, and making modality identification decisions based on the results of modality matching. This may be a threshold-based decision, e.g. if the match score is above a certain threshold, the oscillation mode is considered to have been successfully detected. And generating MEMS oscillation mode detection data according to the mode detection decision result, and recording the data comprising the oscillation mode information and the detection result and the state information (such as oscillation frequency, oscillation amplitude and the like) of the oscillator. By the steps, the identification of the oscillation mode is realized, so that MEMS oscillation mode detection data are generated.
Preferably, step S4 comprises the steps of:
step S41: performing data characterization processing based on the MEMS oscillation mode detection data to generate heterogeneous mode feature mapping data;
step S42: embedding a high-dimensional feedback path by utilizing the heterogeneous modal feature mapping data to generate high-dimensional feedback characteristic data;
step S43: performing characteristic logic entropy coding on the high-dimensional feedback characteristic data through an information entropy principle to generate logic entropy coding mapping data;
step S44: performing feedback topology network construction based on logic entropy coding mapping data to generate a topology feedback relation matrix;
step S45: and carrying out real-time feedback rule fusion according to the topological feedback relation matrix to generate the MEMS real-time feedback rule set.
According to the invention, the data characterization processing is performed on the basis of the MEMS oscillation mode detection data so as to generate the heterogeneous mode feature mapping data, and the original and possibly huge MEMS oscillation mode detection data is converted into the more compact and concise heterogeneous mode feature mapping data through feature extraction. This process has significant benefits in terms of data storage, transmission, and processing, saving computing and storage resources. The original, possibly huge, MEMS oscillation mode detection data is converted into more compact and concise heterogeneous mode feature mapping data through feature extraction. This process has significant benefits in terms of data storage, transmission, and processing, saving computing and storage resources. The high-dimensional feedback path embedding is performed by utilizing the heterogeneous modal feature mapping data to generate high-dimensional feedback characteristic data, and the feature richness of the data is increased through the high-dimensional feedback path embedding of the heterogeneous modal feature mapping data. This means that the generated high-dimensional feedback characteristic data contains more detailed information about the MEMS oscillation modes, including correlations of the modes. The high-dimensional feedback characteristic data is subjected to characteristic logic entropy coding through the information entropy principle to generate logic entropy coding mapping data, and the complexity of the data can be quantized through calculating the information entropy of the high-dimensional feedback characteristic data, so that the feedback characteristic of the MEMS oscillator can be more comprehensively understood. Logical entropy encoded mapping data presents this complexity in an intuitive way, and high-dimensional feedback characteristic data may contain multiple characteristics, each of which may vary in different dimensions and ranges. The information entropy calculation provides a single numerical representation for each property, converting the multi-dimensional property data into a more compact form. This helps reduce the dimensionality and complexity of the data, making the data easier to understand and analyze. The step of constructing the feedback topological network based on the logic entropy coding mapping data to generate a topological feedback relation matrix is beneficial to establishing a topological relation model among internal oscillating units of the MEMS oscillator. This model can help researchers and engineers better understand the manner of connection and interaction between oscillating units. This is critical to design, analyze, and optimize the performance of the MEMS oscillator. And carrying out real-time feedback rule fusion according to the topological feedback relation matrix to generate an MEMS real-time feedback rule set, and ensuring that the generated real-time feedback rule set has consistency among different components by analyzing the topological feedback relation matrix and executing rule fusion. This helps to avoid inconsistent feedback signals in the system, thereby improving the stability and reliability of the system and enabling detection and resolution of potential conflicts between feedback rules. By merging repeated rules or adjusting rules to resolve conflicts, the system is better able to cope with complex feedback situations.
Step S41: performing data characterization processing based on the MEMS oscillation mode detection data to generate heterogeneous mode feature mapping data;
in the embodiment of the invention, MEMS oscillation mode detection data are acquired. These data include the vibration characteristics of the MEMS oscillator in different modes. The data may include information on modal frequencies, vibration amplitudes, phases, etc. And converting the original modal identification data into more characteristic feature data by adopting a feature extraction method. This may include the following operations: frequency domain analysis technologies such as Fourier transform or wavelet transform are used for extracting vibration characteristics under different frequencies, statistical characteristics of modal identification data such as mean value, variance, skewness and kurtosis are calculated, a most characteristic feature subset is selected according to the needs of specific tasks, and the extracted feature data is mapped to a heterogeneous modal feature space. This may include combining feature data in different modalities into one unified feature vector for subsequent processing to map extracted feature data to heterogeneous modality feature space. This may include combining the feature data in different modes into a unified feature vector for subsequent processing, and generating heterogeneous mode feature map data by the selected algorithm, including a more compact and characterizable description of MEMS oscillation mode detection data. The data characterization processing is realized through the steps so as to generate heterogeneous modal feature mapping data.
Step S42: embedding a high-dimensional feedback path by utilizing the heterogeneous modal feature mapping data to generate high-dimensional feedback characteristic data;
in the embodiment of the invention, heterogeneous mode characteristic mapping data are obtained, and the data contain information about MEMS oscillation modes. Mapping data into a high-dimensional space, and selecting a high-dimensional feedback path embedding algorithm. In this example, we employ a t-distributed random neighborhood embedding (t-SNE) algorithm. Depending on the requirements of the chosen algorithm, we configure the relevant parameters: embedding dimensions: we set the dimensions of the target high-dimensional space to determine the representation dimensions of the data in the new space. Embedding dimensions: we set the dimensions of the target high-dimensional space to determine the representation dimensions of the data in the new space. By applying the t-SNE algorithm, we map the heterogeneous modal feature map data to a high-dimensional space. The process involves the steps of calculating a similarity matrix between heterogeneous modal feature map data to capture the relationship between data points, mapping the data points into a high-dimensional space using the similarity matrix, wherein each dimension represents the position of a data point in the new space, and implementing a high-dimensional feedback path embedding through the steps to generate high-dimensional feedback characteristic data comprising information of the MEMS oscillator feedback characteristics, key characteristic extraction data.
Step S43: performing characteristic logic entropy coding on the high-dimensional feedback characteristic data through an information entropy principle to generate logic entropy coding mapping data;
in the embodiment of the invention, high-dimensional feedback characteristic data are acquired, and the data comprise information reflecting the feedback characteristic of the MEMS oscillator. Each characteristic may have a different dimension and range. And carrying out information entropy calculation on each piece of high-dimensional feedback characteristic data. Information entropy is an indicator for measuring uncertainty and complexity of data. The specific steps for calculating the information entropy are as follows: and dividing each high-dimensional characteristic data into a plurality of sections or intervals, counting the number of data points in each interval, and calculating the information entropy of each interval, wherein the calculation of the information entropy does not need to go deep into formula details and only needs to be estimated according to the frequency distribution of the interval. And generating logic entropy coding mapping data by taking the calculated information entropy value as a characteristic. These data will represent the complexity and uncertainty of the high-dimensional feedback characteristic data in the form of information entropy values. The characteristic logical entropy coding is realized through the steps so as to generate logical entropy coding mapping data.
Step S44: performing feedback topology network construction based on logic entropy coding mapping data to generate a topology feedback relation matrix;
In an embodiment of the invention, logic entropy coding mapping data is obtained, wherein the data comprises characteristic information about the MEMS oscillator. First, it is necessary to define the topology of the MEMS oscillator, i.e., the connection relationship between the individual oscillation units. This may include the definition of nodes (oscillating units) and edges (connection relations). And constructing a topological feedback relation matrix according to the logic entropy coding mapping data. The method comprises the following specific steps: and representing the oscillation units in the MEMS oscillator as nodes of a matrix, and calculating the edge weights among different oscillation units according to the characteristic information in the logic entropy coding mapping data. This may include a similarity measure between features, such as euclidean distance or correlation coefficient, and edge weights between different oscillating units are calculated from the feature information in the logical entropy encoded mapping data. This may include a similarity measure between features, such as euclidean distance or correlation coefficient, and the generated topological feedback relationship matrix may be optimized or dimensionality reduced as desired. This may include pruning, feature selection or dimension reduction algorithms to reduce the complexity of the matrix. Through the steps, the feedback topology network construction is realized to generate a topology feedback relation matrix, wherein the feedback relation matrix comprises feedback relations among internal oscillation units of the MEMS oscillator.
Step S45: and carrying out real-time feedback rule fusion according to the topological feedback relation matrix to generate the MEMS real-time feedback rule set.
In the embodiment of the invention, a topological feedback relation matrix is acquired, and a proper algorithm is selected to execute feedback rule fusion. The algorithm needs to be able to integrate different feedback rules into a consistent set of real-time feedback rules based on the topological feedback relationship matrix. One common approach is to use graph theory algorithms, such as shortest path algorithms or topological ordering. And analyzing the topological feedback relation matrix, determining which components have feedback relation and types of the feedback relation (positive feedback, negative feedback and the like), and fusing different feedback rules according to the analyzed topological feedback relation matrix. This may include merging repeated rules, resolving potential conflicts or inconsistencies, and ensuring consistency and validity of the generated real-time feedback rule set, which is generated from the fused rules, including processing feedback relationship data between different components in the MEMS oscillator.
Preferably, step S5 comprises the steps of:
step S51: performing environmental characteristic clustering on the MEMS real-time feedback rule set by using a distributed clustering algorithm to generate MEMS environmental characteristic data;
Step S52: virtual environment synthesis is carried out based on the MEMS environment characteristic data so as to generate MEMS environment simulation data;
step S53: performing response deviation detection through MEMS environment simulation data to generate a response deviation report;
step S54: performing calibration rule programming in combination with the response deviation report to generate a dynamic calibration rule set;
step S55: and carrying out real-time precise calibration on the MEMS oscillator by using the dynamic calibration rule set to generate MEMS oscillation calibration data.
The invention performs environmental characteristic clustering on the MEMS real-time feedback rule set by using a distributed clustering algorithm to generate MEMS environmental characteristic data, and extracts the characteristic information related to the environment from the MEMS real-time feedback rule set. Such characteristic information may include environmental parameters such as temperature, humidity, pressure, etc., and oscillator feedback rules. Through a clustering algorithm, the features can be effectively discovered and identified, and a basis is provided for subsequent analysis and decision. And extracting the characteristic information related to the environment from the MEMS real-time feedback rule set. Such characteristic information may include environmental parameters such as temperature, humidity, pressure, etc., and oscillator feedback rules. Through a clustering algorithm, the features can be effectively discovered and identified, and a basis is provided for subsequent analysis and decision. Virtual environment synthesis is performed based on the MEMS environmental feature data to generate MEMS environmental simulation data, and step S52 can simulate various environmental conditions in the real world, including temperature, humidity, pressure, etc., by extracting statistical information of the MEMS environmental feature data and setting virtual environment parameters according to the information. This helps to more fully evaluate the performance and stability of the MEMS oscillator without the need for costly and time consuming testing in different environments. And performing response deviation detection through the MEMS environment simulation data to generate a response deviation report, and analyzing the difference between the MEMS environment simulation data and actual MEMS oscillator feedback rule data to aim at detecting the response deviation of the oscillator. This helps to ensure the working quality of the oscillator, discover and deal with potential problems in time, improve product quality and reliability, and respond to deviation detection so that the performance of the oscillator can be monitored in real time. Once a response deviation occurs, the system can react quickly and corrective action can be taken to ensure that the oscillator will function properly under a variety of environmental conditions. Calibration rules are formulated in conjunction with the response deviation report to generate a dynamic calibration rule set, the generation of which enables calibration to be performed under real-time environmental conditions. The performance parameters of the oscillator can be adjusted in real time according to the current environmental conditions so as to ensure the stability and performance of the oscillator under different conditions, and the establishment of calibration rules aims at reducing the response deviation of the oscillator. Through an automatic calibration strategy, the performance parameters of the oscillator can be finely adjusted in real time, so that response deviation is reduced, and accuracy and stability of the oscillator are improved.
Step S51: performing environmental characteristic clustering on the MEMS real-time feedback rule set by using a distributed clustering algorithm to generate MEMS environmental characteristic data;
in the embodiment of the invention, firstly, related data is acquired from the MEMS real-time feedback rule set. This includes feedback rules for the oscillator and information about the environment. In order to realize the clustering of the environmental characteristics, an applicable distributed clustering algorithm is selected. In this step we will employ a density-based clustering algorithm (e.g., DBSCAN-density-based spatial clustering applying noise) to find data points with similar environmental characteristics. And carrying out feature extraction on the data acquired from the MEMS feedback rule set. This may include extracting environmental related features from feedback rules, such as temperature, humidity, pressure, etc., using DBSCAN or other selected distributed clustering algorithms to partition the data points into different clusters. Each cluster represents a similar set of environmental features. For each cluster, statistical information of its environmental features, such as mean, variance, etc., is calculated. These statistics will be part of the generated MEMS environmental characteristic data. The environmental characteristic clustering is realized through the steps, so that MEMS environmental characteristic data including environmental parameters and oscillator feedback rules are generated;
Step S52: virtual environment synthesis is carried out based on the MEMS environment characteristic data so as to generate MEMS environment simulation data;
in the embodiment of the present invention, statistical information of each environmental feature, such as mean, variance, etc., is extracted from the MEMS environmental feature data obtained in step S51. Parameters of the virtual environment, including but not limited to temperature, humidity, pressure, etc., are then set based on these statistics. An appropriate virtual environment model is selected to simulate the behavior of the MEMS oscillator under different environmental conditions. This may include linear models, nonlinear models, physical simulation models, and the like. And generating MEMS environment simulation data according to the set virtual environment parameters and the selected environment model. The specific generation process comprises the following steps: setting initial conditions such as initial position, speed and the like according to statistical information of environmental characteristics, simulating the response of the MEMS oscillator in the environment at certain time step intervals, calculating the response of the MEMS oscillator at different time steps according to a selected environment model, including displacement, speed, acceleration and the like, and recording oscillator response data at each time step, including oscillator state and environmental conditions. The resulting data includes response data of the MEMS oscillator under simulated environmental conditions, including performance data of the MEMS oscillator under different environmental conditions.
Step S53: response bias detection is performed by the MEMS environmental simulation data to generate a response bias report.
In the embodiment of the invention, firstly, a proper data sample is selected from MEMS environment simulation data and actual MEMS oscillator feedback rule data, the quality and the integrity of the data are ensured, and the characteristic extraction is carried out on the selected data sample. This includes measurements of key parameters of the oscillator, such as amplitude, frequency, phase, etc., as well as characteristics related to environmental conditions, such as temperature, humidity, etc. These features will be used for subsequent analysis. Under the condition of no deviation, the actual MEMS feedback rule data is used for establishing a reference model which reflects the oscillator behavior under the normal working state, and an advanced anomaly detection algorithm such as an isolated Forest (Isolation Forest) or a statistical-based method is adopted for comparing the difference between the simulation data and the reference model. This will help identify potential response deviations, identifying points of difference between the simulated data and the actual data. These difference points represent the response deviation of the oscillator in the simulated environment. The detected anomalies or response deviations are analyzed in detail and a response deviation report is generated. The report should include the nature of the deviation, the time of occurrence, the possible reasons, and an evaluation of the effect on the oscillator performance. The response deviation detection is realized through the steps, so that a response deviation report is generated.
Step S54: performing calibration rule programming in combination with the response deviation report to generate a dynamic calibration rule set;
in the embodiment of the invention, the following information and data need to be prepared before entering the calibration rule preparation: response bias report: this is generated in the previous step S53, and contains the correlation information between the environmental characteristics and the oscillator response, the environmental characteristics data: this is generated in step S51 for identifying different environmental conditions, MEMS oscillator performance parameters: the method comprises the steps of calibrating rule templates by using parameters such as oscillation frequency, oscillation amplitude, phase and the like: the rule templates defined in advance include adjustment strategies for calibration parameters. And automatically generating a calibration rule by using a selected algorithm according to the information in the response deviation report and the environmental characteristic data. The generation of rules covers the following aspects: determining oscillator performance parameters, such as oscillation frequency or phase, to be adjusted, formulating an adjustment strategy for the calibration parameters so as to reduce response bias under specific environmental conditions, establishing a mapping relationship between the environmental features and the calibration parameters, so as to select an appropriate calibration strategy according to the environmental conditions, and integrating the generated calibration rules into a dynamic calibration rule set. This rule set contains calibration strategies for different environmental conditions for real-time application to the MEMS oscillator. Calibration rule programming is achieved through the above steps to generate a dynamic calibration rule set.
Step S55: and carrying out real-time precise calibration on the MEMS oscillator by using the dynamic calibration rule set to generate MEMS oscillation calibration data.
In the embodiment of the invention, the environmental condition of the MEMS oscillator needs to be continuously monitored before real-time calibration. This includes real-time measurement of temperature, humidity, pressure, etc. parameters, retrieving calibration rules matching the current environmental conditions from a dynamic calibration rule set. This requires the selection of appropriate rules based on the monitored environmental parameters, and the adjustment of relevant parameters of the MEMS oscillator, such as frequency, amplitude, phase, etc., based on the selected calibration rules. These adjustments are to maintain the performance of the oscillator in the current environment, and the performance of the MEMS oscillator is continuously monitored during calibration. The method comprises the steps of recording performance indexes such as frequency, amplitude and the like of an oscillator output signal, and feeding back the performance indexes to the adjustment process of the calibration parameters according to real-time performance monitoring results. This ensures that the calibration is dynamic and can be adjusted as environmental conditions change, recording the results of each calibration, including the setting of calibration parameters and actual performance data. This will generate MEMS oscillation calibration data, including oscillator parameter data, calibration time stamps, and calibration rule data. The MEMS oscillator is precisely calibrated in real time through the steps, so that MEMS oscillation calibration data are generated.
Preferably, step S6 comprises the steps of:
step S61: performing multidimensional interference identification processing on the MEMS oscillation calibration data to generate oscillation interference identification data, and performing data mode conversion on the oscillation interference identification data to generate an interference mode indicator diagram;
step S62: network interference sensing is carried out through an interference mode indicator diagram so as to generate an interference source data report;
step S63: performing self-adaptive interference modeling by combining the interference pattern index map and the interference source data report to generate a dynamic interference prediction model;
step S64: and carrying out real-time interference mitigation processing on the MEMS oscillator based on the dynamic interference prediction model so as to generate an oscillation interference mitigation strategy.
The invention is beneficial to identifying and separating various interference components in the MEMS oscillation calibration data through multidimensional data processing technology such as Principal Component Analysis (PCA) and Independent Component Analysis (ICA). This includes electromagnetic noise, mechanical vibrations, etc. that may affect the performance of the oscillator. Through effective identification and separation, the accuracy and stability of oscillation calibration data are expected to be improved. The network interference sensing is performed through the interference pattern indicator diagram to generate an interference source data report, the step is beneficial to realizing efficient network maintenance, and a network administrator can quickly take measures to restore the normal operation of the network by timely identifying and positioning the interference source. This helps to reduce network downtime, improve network availability and stability, and step S62 allows the operator to learn the nature and impact of the disturbance by detailed analysis and classification of the disturbance signature. This helps to take targeted interference mitigation measures that reduce the adverse impact of interference on system performance. The core goal of this step is to generate a dynamic interference prediction model that can accurately predict the various disturbances that may occur in the system. The method is favorable for timely identifying and coping with different types of interference, so that the stability and performance of the system are improved, the constructed interference prediction model has self-adaptability, and parameters and algorithms can be adjusted according to the change of the interference mode in the historical data. This means that the model can adapt to the changes of different working environments and interference sources, and high prediction accuracy is maintained. The method has the main beneficial effects that the real-time interference prediction is realized. By building a dynamic disturbance prediction model, the system is able to continuously monitor the operating state of the MEMS oscillator and predict potential disturbance events. In this way, the system can detect possible problems earlier, and help to take timely measures to alleviate the interference, thereby improving the reliability and stability of the system.
Step S61: performing multidimensional interference identification processing on the MEMS oscillation calibration data to generate oscillation interference identification data, and performing data mode conversion on the oscillation interference identification data to generate an interference mode indicator diagram;
in the embodiment of the invention, first, oscillation calibration data is obtained from the MEMS oscillator, and the data includes parameters such as amplitude, frequency, phase and the like of the oscillator. These data are typically recorded in time series. The oscillation calibration data is subjected to interference recognition using multidimensional data processing techniques such as Principal Component Analysis (PCA) and Independent Component Analysis (ICA). These techniques may be used to separate out interfering components in the oscillating signal, such as electromagnetic noise, mechanical vibrations, etc., and extract interfering components in the oscillation calibration data based on the results of the interference identification. These components may be unstable parts of the oscillator, which need to be further analyzed and processed, using specific algorithms, such as wavelet transforms or fourier transforms, to convert the interference identification data from the time domain to the frequency domain. This helps to better represent the interference pattern, mapping the frequency domain data onto a graph or image to generate an interference pattern index map. This graph reflects the distribution and characteristics of interference patterns of different frequencies and amplitudes, which can be used for subsequent analysis. The multi-dimensional interference identification processing is realized through the steps to generate oscillation interference identification data, and the data mode conversion is carried out on the oscillation interference identification data to generate an interference mode indicator diagram.
Step S62: network interference sensing is carried out through an interference mode indicator diagram so as to generate an interference source data report;
in the embodiment of the invention, firstly, a generated interference pattern index chart is obtained, and the chart comprises the instability condition, the interference characteristic and the time domain and frequency domain information of an oscillator. And then, carrying out detailed analysis on the interference pattern index graph by adopting a time domain and frequency domain analysis method so as to identify the characteristics of frequency, intensity, duration and the like of interference, and extracting specific interference characteristics based on an analysis result. This may include information on the main frequency content, power spectral density, signal strength, phase change, etc. The possible interference source locations are located using signal processing techniques or multi-sensor fusion methods based on the interference characteristics. This may involve doppler radar, directional antennas, physical measurements, etc. techniques for classifying and identifying the type of interference using machine learning algorithms, such as Support Vector Machines (SVMs) or deep learning neural networks. This requires training a classifier in advance to distinguish between different types of interference, such as electromagnetic interference, mechanical vibration interference, etc., and generating an interference source data report based on the results of the analysis. This report should include the type, location, characteristics, possible impact range and suggested interference mitigation measures of the interfering source. The generation of reports may employ automated report generation algorithms to ensure consistency and timeliness. Network interference sensing is achieved through the steps to generate an interference source data report.
Step S63: performing self-adaptive interference modeling by combining the interference pattern index map and the interference source data report to generate a dynamic interference prediction model;
in the embodiment of the invention, an appropriate adaptive interference modeling algorithm, such as an adaptive filter, a time sequence analysis method or a neural network model, is selected. The algorithm should be selected to extract useful features from the interference pattern index map and the interference source data report based on the characteristics of the interference pattern index map and the interference source data report and the nature of the problem. These characteristics may include spectral characteristics, time domain characteristics, amplitude, phase, etc. information of the interferer. Feature extraction helps build the input for adaptive disturbance modeling. And constructing a dynamic interference prediction model based on the extracted features and the selected adaptive interference modeling algorithm. Parameters of the model should be adjusted according to the historical data to adapt to the changes of different interference modes, and the built model is trained and verified by using the historical data. This includes fitting of models, parameter optimization and performance assessment. The performance of the model can be evaluated by methods such as cross-validation and the like, and a model updating strategy is formulated so as to maintain the adaptability of the model. This may include periodically retraining the model or triggering a model update when a change in interference pattern is detected, and generating a dynamic interference prediction model after model construction and verification is completed. The model should include adaptive parameters and algorithms for use in real-time interference prediction. The adaptive interference modeling is realized through the steps so as to generate a dynamic interference prediction model.
Step S64: and carrying out real-time interference mitigation processing on the MEMS oscillator based on the dynamic interference prediction model so as to generate an oscillation interference mitigation strategy.
In the embodiment of the invention, the actual running state of the MEMS oscillator is continuously monitored by using the established dynamic interference prediction model. By observing the real-time oscillation signal of the oscillator and the external environment changes, a model is used to predict possible disturbance events. Dynamic interference prediction models typically include parameters that may need to be dynamically adjusted according to actual operating conditions to improve prediction accuracy. For example, adaptive filter techniques, such as Kalman filtering, may be employed to correct predictions in real time, and once a possible interference event is detected, an interference mitigation strategy for the current situation is generated based on the output of the dynamic interference prediction model. This strategy may include the following aspects of processing: according to the model output, the operating parameters of the MEMS oscillator are adjusted, if necessary, the MEMS oscillator is switched to a standby mode to maintain its normal operation while suspending or isolating the mode subject to interference, the operation of the MEMS oscillator is adjusted in real time to eliminate the interference using a feedback control algorithm, such as PID control, digital signal processing techniques, such as filters, are applied to filter the oscillating signal to remove or reduce the effects of the interfering signal. The real-time interference mitigation process is implemented through the above steps to generate an oscillating interference mitigation strategy.
Preferably, step S7 comprises the steps of:
step S71: deep feature extraction is carried out on the oscillation interference alleviation strategy to generate an oscillation feature matrix;
step S72: performing strategy learning optimization processing on the oscillation feature matrix by adopting a deep reinforcement learning model to generate an oscillation optimization strategy;
step S73: and carrying out parameter self-adaptive fine adjustment on the MEMS oscillator by utilizing an oscillation optimization strategy so as to generate oscillator self-adjusting parameters.
According to the method, the oscillation feature matrix is generated by carrying out depth feature extraction on the oscillation interference alleviation strategy, and abundant oscillation features including various features such as frequency spectrum, time domain, amplitude and phase can be extracted from the MEMS oscillator operation record through the deep learning model. The characteristics can comprehensively describe the performance and effect of the oscillation interference mitigation strategy, the oscillation characteristic matrix comprises key characteristics of the oscillation interference mitigation strategy, and the characteristics can better reflect the running state and the environmental condition of the oscillator through extraction of a deep learning model. This helps to improve the effectiveness and performance of the oscillating interference mitigation strategy. And performing strategy learning optimization processing on the oscillation feature matrix by adopting a deep reinforcement learning model to generate an oscillation optimization strategy, and realizing intelligent optimization on the oscillator strategy by adopting the deep reinforcement learning model. This means that the performance and stability of the oscillator can be significantly improved, so that a higher quality oscillation signal can be provided for various applications, the deep reinforcement learning model can be automatically adapted to different oscillation characteristic situations, and the optimal oscillation interference mitigation strategy can be selected according to the current state. The self-adaption enables the oscillator to keep stability under different working conditions without manual adjustment, and the main advantage of the step is that the method adopts an oscillation optimization strategy, so that the MEMS oscillator can automatically adjust parameters in the running process to adapt to the continuously changing working conditions and requirements. Such adaptive tuning can ensure that the oscillator provides optimal performance in different operating environments without manual intervention, and can significantly improve the performance of the oscillator by adaptively tuning the oscillator. Performance metrics include frequency stability, harmonic distortion, phase noise, etc. These performance improvements help ensure that the oscillator provides a more accurate and reliable output signal in practical applications. The adaptive trimming enables the oscillator to respond quickly to changes in the external environment. When external conditions change, the oscillator can quickly adjust its own parameters to maintain optimal performance, which is critical for applications requiring a high degree of stability.
Step S71: deep feature extraction is carried out on the oscillation interference alleviation strategy to generate an oscillation feature matrix;
in the embodiment of the invention, the related data of the oscillation interference mitigation strategy is obtained from the operation record of the MEMS oscillator. Such data includes oscillator status, environmental parameters, interference information, etc. It is ensured that the dataset contains enough samples and features, in which step the features to be extracted are selected. This may include spectral features, time domain features, amplitude features, phase features, etc. The characteristics are selected taking into account their correlation with the oscillating interference mitigation strategy. An appropriate deep learning model is selected for feature extraction. One common approach is to use Convolutional Neural Networks (CNNs) to extract features with information content from the raw data. For example, multiple convolution layers and pooling layers may be used to capture spatial features in the data, and the oscillation interference mitigation strategy data is input into a deep learning model. The model will gradually extract features, from simple features to more abstract features. These features are represented and transformed in different levels, organizing features extracted from the deep learning model into a feature matrix. Each row represents a data sample and each column represents a feature. This feature matrix will serve as input for subsequent oscillation interference mitigation strategy learning. The method can realize deep feature extraction on the oscillation interference mitigation strategy so as to generate an oscillation feature matrix.
Step S72: performing strategy learning optimization processing on the oscillation feature matrix by adopting a deep reinforcement learning model to generate an oscillation optimization strategy;
in the embodiment of the invention, a deep reinforcement learning model suitable for the oscillation optimization problem is selected. We have chosen Deep Q-Network (DQN) as an example, which model has performed well in many reinforcement learning tasks. The DQN combines a deep neural network and a Q learning algorithm to define a neural network architecture of the DQN, including an input layer, a hidden layer, and an output layer. For the input of the oscillation feature matrix, a Convolutional Neural Network (CNN) or a fully-connected neural network may be employed, with the appropriate hierarchy and activation functions being selected based on the characteristics of the data, and the goal of the DQN is to learn a Q-value function that represents the expected jackpot for taking action in a given state. In this step, a specific definition of the Q-value function is required to ensure that it can describe the performance of the oscillator strategy. And taking the oscillation characteristic matrix as input, and starting the strategy learning training process of the DQN. During training, the model will learn to choose the best oscillation interference mitigation strategy under different oscillation feature scenarios, and to stabilize the training, an experience playback buffer is employed for storing past states, actions, rewards, and experiences for the next state. The model is trained by randomly sampling from the buffer zone to reduce the data correlation, and a target network is adopted to stabilize the training due to the problem of the change of the target Q value in the DQN training. The parameters of the target network are updated at intervals to reduce fluctuations in the target Q value. During training, the DQN optimizes the oscillation strategy by maximizing the jackpot signal. The model maximizes the expected rewards by adjusting the oscillation interference mitigation strategy to improve oscillator performance and, after training is completed, the DQN will generate an optimized oscillation strategy.
Step S73: and carrying out parameter self-adaptive fine adjustment on the MEMS oscillator by utilizing an oscillation optimization strategy so as to generate oscillator self-adjusting parameters.
In the embodiment of the invention, an initial state needs to be established for the parameters of the oscillator before starting the adaptive trimming. These parameters may include frequency, amplitude, phase, etc. Ensuring that the initial values of these parameters are within acceptable operating ranges, the parameters of the oscillator are fine-tuned step by step according to the advice in the strategy, using the oscillation optimization strategy generated in the previous step S72 as a guide. This may include incrementing or decrementing parameter values to better adapt to the optimization strategy, and evaluating the performance of the oscillator after each round of trimming. This can be done by monitoring the output signal of the oscillator and calculating the performance index. Performance metrics may include frequency stability, harmonic distortion, phase noise, etc., comparing the tuned oscillator performance to initial performance and optimization objectives. If the performance does not reach the expected optimization goal, continuing the fine tuning process, formulating a condition for stopping fine tuning, for example, when the performance of the oscillator reaches a satisfactory level, or fine tuning reaches a predetermined number of iterations, in each step of the adaptive fine tuning, updating the parameters of the oscillator according to the result of the strategically directed fine tuning, and generating the self-tuning parameters of the oscillator after completing the adaptive fine tuning process.
The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of detecting a MEMS oscillator, comprising the steps of:
step S1: acquiring oscillation signal data of the MEMS oscillator, and carrying out oscillation nonlinear spectrum construction on the oscillation signal data of the MEMS oscillator to generate an MEMS nonlinear oscillation spectrum;
step S2: performing spectral domain topology analysis processing on the MEMS nonlinear oscillation spectrum to generate MEMS dynamic spectrum characteristic structure data;
step S3: performing adaptive mode detection on the MEMS oscillator through MEMS dynamic spectrum characteristic structure data to generate MEMS oscillation mode detection data;
step S4: performing feedback logic construction based on MEMS oscillation mode detection data to generate an MEMS real-time feedback rule set;
Step S5: performing environmental simulation calibration based on the MEMS real-time feedback rule set to generate MEMS oscillation calibration data;
step S6: according to the MEMS oscillation calibration data, oscillation dynamic interference decoupling is carried out so as to generate an oscillation interference alleviation strategy;
step S7: and carrying out self-adaptive adjustment optimization on the MEMS oscillator by utilizing an oscillation interference mitigation strategy so as to generate oscillator self-tuning parameters.
2. The method according to claim 1, wherein the specific steps of step S1 are:
step S11: acquiring oscillation signal data of the MEMS oscillator, and performing dynamic instability interferometry on the oscillation signal data of the MEMS oscillator to generate MEMS instability signal frequency band data;
step S12: performing entropy coding nonlinear disclosure on MEMS unstable signal frequency band data to generate an MEMS nonlinear entropy characteristic diagram;
step S13: performing double-frequency coherent enhancement processing on the MEMS nonlinear entropy feature map by adopting a double-frequency technology to generate MEMS double-frequency enhancement mapping data;
step S14: and performing three-dimensional mapping visualization based on the MEMS dual-frequency enhanced mapping data to generate an MEMS nonlinear oscillation map.
3. The method according to claim 1, wherein the specific steps of step S2 are:
Step S21: performing time spectrum domain enhancement processing through the MEMS nonlinear oscillation spectrum to generate MEMS enhancement time spectrum data;
step S22: calculating the spectrum data during the MEMS enhancement by utilizing the spectrum domain feature comprehensive function to generate MEMS spectrum domain feature data;
step S23: performing feature network topology mapping based on the MEMS spectral domain feature data to generate an MEMS spectral domain feature network map;
step S24: acquiring an MEMS oscillator, and carrying out dynamic topological structure analysis on the MEMS oscillator by combining with the MEMS spectral domain characteristic network diagram so as to generate an oscillator structural topological diagram;
step S25: and performing characteristic projection and structuring based on the topological graph of the oscillator structure to generate MEMS dynamic spectrum characteristic structure data.
4. A method according to claim 3, wherein the spectral domain feature synthesis function in step S22 is specifically:
wherein F refers to MEMS spectral domain feature data, x i Refers to the representation of the original data in the time-frequency spectrum domain, n refers to the total number of data points, delta phi i Refers to the offset of the ith data point relative to the average value, T refers to the duration of the whole data, θ refers to the spectral domain angle factor, Δψ i Refers to the phase difference of the ith data point, and λ refers to the spectral domain intensity factor.
5. The method according to claim 4, wherein the specific steps of step S3 are:
step S31: calculating MEMS dynamic spectrum feature structure data by utilizing a multidimensional spectrum feature popular conversion formula to generate MEMS spectrum feature manifold data;
step S32: performing adaptive mode coding on MEMS spectral feature manifold data to generate MEMS mode feature data;
step S33: performing time sequence integration processing on the MEMS modal feature data by using a cyclic neural network to generate MEMS time sequence modal feature data;
step S34: performing self-adaptive pooling processing based on MEMS time sequence modal characteristic data to generate an MESM self-adaptive modal matrix;
step S35: and acquiring the MEMS oscillator, and performing oscillation mode detection on the MEMS oscillator through the MESM self-adaptive mode matrix to generate MEMS oscillation mode detection data.
6. The method according to claim 5, wherein the multi-dimensional spectral feature popularity transformation formula in step S31 is specifically:
wherein G refers to MEMS spectral feature manifold structure data, S refers to input spectral features, and R refers to increaseA reference function of nonlinear characteristics, m refers to the total number of data points in the input MEMS dynamic spectrum characteristic structure data, gamma refers to an exponential function, t refers to a time variable, s j Refers to the jth data point in the spectral signature, delta refers to the manifold weight constant, and beta refers to a nonlinear adjustment factor that adjusts the intensity of the spectral signature.
7. The method according to claim 1, wherein the specific step of step S4 is:
step S41: performing data characterization processing based on the MEMS oscillation mode detection data to generate heterogeneous mode feature mapping data;
step S42: embedding a high-dimensional feedback path by utilizing the heterogeneous modal feature mapping data to generate high-dimensional feedback characteristic data;
step S43: performing characteristic logic entropy coding on the high-dimensional feedback characteristic data through an information entropy principle to generate logic entropy coding mapping data;
step S44: performing feedback topology network construction based on logic entropy coding mapping data to generate a topology feedback relation matrix;
step S45: and carrying out real-time feedback rule fusion according to the topological feedback relation matrix to generate the MEMS real-time feedback rule set.
8. The method according to claim 1, wherein the specific step of step S5 is:
step S51: performing environmental characteristic clustering on the MEMS real-time feedback rule set by using a distributed clustering algorithm to generate MEMS environmental characteristic data;
Step S52: virtual environment synthesis is carried out based on the MEMS environment characteristic data so as to generate MEMS environment simulation data;
step S53: performing response deviation detection through MEMS environment simulation data to generate a response deviation report;
step S54: performing calibration rule programming in combination with the response deviation report to generate a dynamic calibration rule set;
step S55: and carrying out real-time precise calibration on the MEMS oscillator by using the dynamic calibration rule set to generate MEMS oscillation calibration data.
9. The method according to claim 1, wherein the specific step of step S6 is:
step S61: performing multidimensional interference identification processing on the MEMS oscillation calibration data to generate oscillation interference identification data, and performing data mode conversion on the oscillation interference identification data to generate an interference mode indicator diagram;
step S62: network interference sensing is carried out through an interference mode indicator diagram so as to generate an interference source data report;
step S63: performing self-adaptive interference modeling by combining the interference pattern index map and the interference source data report to generate a dynamic interference prediction model;
step S64: and carrying out real-time interference mitigation processing on the MEMS oscillator based on the dynamic interference prediction model so as to generate an oscillation interference mitigation strategy.
10. The method according to claim 1, wherein the specific step of step S7 is:
step S71: deep feature extraction is carried out on the oscillation interference alleviation strategy to generate an oscillation feature matrix;
step S72: performing strategy learning optimization processing on the oscillation feature matrix by adopting a deep reinforcement learning model to generate an oscillation optimization strategy;
step S73: and carrying out parameter self-adaptive fine adjustment on the MEMS oscillator by utilizing an oscillation optimization strategy so as to generate oscillator self-adjusting parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311216458.6A CN117272022A (en) | 2023-09-19 | 2023-09-19 | Detection method of MEMS oscillator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311216458.6A CN117272022A (en) | 2023-09-19 | 2023-09-19 | Detection method of MEMS oscillator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117272022A true CN117272022A (en) | 2023-12-22 |
Family
ID=89220828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311216458.6A Pending CN117272022A (en) | 2023-09-19 | 2023-09-19 | Detection method of MEMS oscillator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117272022A (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1864073A (en) * | 2003-10-09 | 2006-11-15 | 纳特维斯公司 | System and method for characterizing a sample by low-frequency spectra |
| US20090222144A1 (en) * | 2008-03-04 | 2009-09-03 | Washington State University | Systems and methods for electromechanical oscillation monitoring |
| CN102934158A (en) * | 2010-01-29 | 2013-02-13 | 循环逻辑有限责任公司 | Learning and auditory scene analysis in gradient frequency nonlinear oscillator networks |
| CN102937668A (en) * | 2012-11-08 | 2013-02-20 | 电子科技大学 | Electric system low-frequency oscillation detection method |
| CN104684841A (en) * | 2012-06-13 | 2015-06-03 | 普渡研究基金会 | Microelectromechanical system and methods of use |
| CN109542089A (en) * | 2018-12-21 | 2019-03-29 | 浙江大学 | A kind of industrial process nonlinear detection method of oscillations based on improvement variation mode decomposition |
| CN112435356A (en) * | 2020-11-04 | 2021-03-02 | 南京航天工业科技有限公司 | ETC interference signal identification method and detection system |
| CN112751345A (en) * | 2020-12-30 | 2021-05-04 | 电子科技大学 | LSTM and phase trajectory based electric power system low-frequency oscillation mode identification method |
| CN113532409A (en) * | 2021-06-18 | 2021-10-22 | 北京时代民芯科技有限公司 | High-precision digital MEMS gyroscope control system and method |
| CN113569464A (en) * | 2021-06-21 | 2021-10-29 | 国网山东省电力公司电力科学研究院 | Wind turbine generator oscillation mode prediction method and device based on deep learning network and multi-task learning strategy |
| CN114417929A (en) * | 2022-01-21 | 2022-04-29 | 重庆大学 | GIS equipment nonlinear vibration behavior analysis method and system |
| WO2022141623A1 (en) * | 2021-01-04 | 2022-07-07 | 中南大学 | Train compartment vibration monitoring method, vibration signal feature library establishment method and vibration signal feature library application method |
| CN115441441A (en) * | 2022-09-21 | 2022-12-06 | 四川大学 | Subsynchronous/supersynchronous oscillation detection method based on oscillation component elliptical track |
| CN115912349A (en) * | 2022-11-25 | 2023-04-04 | 四川大学 | Broadband oscillation disturbance source positioning method based on compressed sensing and graph convolution neural network |
-
2023
- 2023-09-19 CN CN202311216458.6A patent/CN117272022A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1864073A (en) * | 2003-10-09 | 2006-11-15 | 纳特维斯公司 | System and method for characterizing a sample by low-frequency spectra |
| US20090222144A1 (en) * | 2008-03-04 | 2009-09-03 | Washington State University | Systems and methods for electromechanical oscillation monitoring |
| CN102934158A (en) * | 2010-01-29 | 2013-02-13 | 循环逻辑有限责任公司 | Learning and auditory scene analysis in gradient frequency nonlinear oscillator networks |
| CN104684841A (en) * | 2012-06-13 | 2015-06-03 | 普渡研究基金会 | Microelectromechanical system and methods of use |
| CN102937668A (en) * | 2012-11-08 | 2013-02-20 | 电子科技大学 | Electric system low-frequency oscillation detection method |
| CN109542089A (en) * | 2018-12-21 | 2019-03-29 | 浙江大学 | A kind of industrial process nonlinear detection method of oscillations based on improvement variation mode decomposition |
| CN112435356A (en) * | 2020-11-04 | 2021-03-02 | 南京航天工业科技有限公司 | ETC interference signal identification method and detection system |
| CN112751345A (en) * | 2020-12-30 | 2021-05-04 | 电子科技大学 | LSTM and phase trajectory based electric power system low-frequency oscillation mode identification method |
| WO2022141623A1 (en) * | 2021-01-04 | 2022-07-07 | 中南大学 | Train compartment vibration monitoring method, vibration signal feature library establishment method and vibration signal feature library application method |
| CN113532409A (en) * | 2021-06-18 | 2021-10-22 | 北京时代民芯科技有限公司 | High-precision digital MEMS gyroscope control system and method |
| CN113569464A (en) * | 2021-06-21 | 2021-10-29 | 国网山东省电力公司电力科学研究院 | Wind turbine generator oscillation mode prediction method and device based on deep learning network and multi-task learning strategy |
| CN114417929A (en) * | 2022-01-21 | 2022-04-29 | 重庆大学 | GIS equipment nonlinear vibration behavior analysis method and system |
| CN115441441A (en) * | 2022-09-21 | 2022-12-06 | 四川大学 | Subsynchronous/supersynchronous oscillation detection method based on oscillation component elliptical track |
| CN115912349A (en) * | 2022-11-25 | 2023-04-04 | 四川大学 | Broadband oscillation disturbance source positioning method based on compressed sensing and graph convolution neural network |
Non-Patent Citations (3)
| Title |
|---|
| VINAYAK PACHKAWADE 等: "MEMS sensor for detection andmeasurement of ultra-fine particles", ENGINEERING RESEARCH EXPRESS, 7 June 2022 (2022-06-07), pages 1 - 18 * |
| WEIWEI ZHANG 等: "Detection of minor damage in structures with guided wave signals and nonlinear oscillator", MEASUREMENT, 31 December 2017 (2017-12-31), pages 1 - 14 * |
| 江敏: "应用于C类振荡器的功率检测比较器", 中国优秀硕士学位论文全文数据库 信息科技辑, no. 02, 15 February 2016 (2016-02-15), pages 135 - 376 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10921759B2 (en) | Computer system and method for monitoring key performance indicators (KPIs) online using time series pattern model | |
| JP6068468B2 (en) | A sequential kernel regression modeling method for prediction and prediction | |
| AU2012284460B2 (en) | System of sequential kernel regression modeling for forecasting and prognostics | |
| AU2012284498B2 (en) | Monitoring method using kernel regression modeling with pattern sequences | |
| US8660980B2 (en) | Monitoring system using kernel regression modeling with pattern sequences | |
| JP2022500745A (en) | Computer implementation methods, computer program products and systems for anomaly detection and / or predictive maintenance | |
| US11269752B1 (en) | System and method for unsupervised anomaly prediction | |
| US8577815B2 (en) | Method and system for concurrent event forecasting | |
| US20190018375A1 (en) | Apparatus and method for event detection and duration determination | |
| US9235208B2 (en) | System of sequential kernel regression modeling for forecasting financial data | |
| KR102320706B1 (en) | Method for setting model threshold of facility monitoring system | |
| CN116380445B (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN116702093B (en) | Marine target positioning method based on big data fusion | |
| CN115769235A (en) | Method and system for providing an alert related to the accuracy of a training function | |
| CN112613233A (en) | Algorithm for discovering environmental monitoring abnormal data based on single-classification support vector machine model | |
| CN117272022A (en) | Detection method of MEMS oscillator | |
| CN115310499B (en) | Industrial equipment fault diagnosis system and method based on data fusion | |
| Xin et al. | Dynamic probabilistic model checking for sensor validation in Industry 4.0 applications | |
| CN111222203A (en) | Method for establishing and predicting service life model of bearing | |
| JP2017539027A (en) | Time-varying profiling engine for physical systems | |
| Auret | Process monitoring and fault diagnosis using random forests | |
| CN116861175A (en) | Operation track correction method based on neural network | |
| CN116587270A (en) | Time rounding method and system based on speed superposition | |
| CN117610601A (en) | Logistics equipment intelligent positioning method and system based on Internet of things | |
| CN116681351A (en) | Quality prediction and root cause analysis system and method for injection molding process |
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 |