CN105699043A - Method for improving measuring stability and precision of wind tunnel sensor - Google Patents
Method for improving measuring stability and precision of wind tunnel sensor Download PDFInfo
- Publication number
- CN105699043A CN105699043A CN201610230341.7A CN201610230341A CN105699043A CN 105699043 A CN105699043 A CN 105699043A CN 201610230341 A CN201610230341 A CN 201610230341A CN 105699043 A CN105699043 A CN 105699043A
- Authority
- CN
- China
- Prior art keywords
- sensor
- value
- wind tunnel
- array
- precision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M9/00—Aerodynamic testing; Arrangements in or on wind tunnels
- G01M9/02—Wind tunnels
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M9/00—Aerodynamic testing; Arrangements in or on wind tunnels
- G01M9/06—Measuring arrangements specially adapted for aerodynamic testing
Landscapes
- Physics & Mathematics (AREA)
- Fluid Mechanics (AREA)
- General Physics & Mathematics (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The invention provides a method for improving the measuring stability and precision of wind tunnel sensors. Under a static state, a plurality sets of standard values are input to a plurality of sensors which are in same measuring range and of a same type, and corresponding measurement value sets are obtained. Program determines whether the measured values are in a measuring precision scope required by a wind tunnel test. A neural network is trained by using absolute values of the difference between the multiple sets of measurement values as an output of the neural network, so as to establish a neural network difference prediction model. Under a test state, predicted difference data are obtained through the neural network difference prediction model and compared with real-time difference data. A voting method is used to veto the values exceeding the predicted results of the neural network model, and sum and average the measurement values which accord with the prediction standards to obtain the optimal measurement value of sensors. In this way, the measuring stability and precision of sensors are improved.
Description
Technical field
The invention belongs to aerospace industry aerodynamic wind-tunnel technique field, particularly relate to a kind of method improving wind tunnel sensors measurement stability and precision。
Background technology
During wind tunnel test, sensor is the instrument and equipment measuring the important tests data such as dummy vehicle power, moment, pressure, the stability of himself and measurement precision directly influence the reliability of test data, accuracy and test efficiency: on the one hand, wind tunnel sensors has a high stability and high accurancy and precision is to ensure that the important prerequisite of test signals of wind tunnel real-time precision measurment and control, is the key factor affecting wind tunnel test data reliability and precision;On the other hand, the stability of wind tunnel sensors is also the key factor affecting wind tunnel test efficiency。The shortened with the R&D cycle is improved constantly along with advanced aircraft manufacture claim, the requirement of the precision of sensor in wind tunnel system, stability is also more and more higher。But, when wind tunnel test, the as easy as rolling off a log impact being subject to the factors such as self performance, ambient temperature, electromagnetic interference of the precision of sensor, stability, self stability, precision are reduced, thus causing the reliability of test data obtained by general measuring method, precision not to reach the requirement that advanced aircraft is developed。
Summary of the invention
It is an object of the invention to provide a kind of method improving wind tunnel sensors measurement stability and precision。The method is first when static state, multiple sensors to same range, with type input many group standard values and obtain the measured value array of its correspondence, program judges that measured value is whether within the scope of the certainty of measurement that wind tunnel test requires subsequently, if in accuracy rating, using the measured value input as train samples data, neutral net is trained by the absolute value of the mutual difference between corresponding many groups measured value as the output of neutral net, thus setting up the sensor neutral net difference forecast model of solidification。At the test condition, gather same range, many groups of real-time measured values with the multiple sensor of type, by neutral net difference forecast model, obtain prediction difference data, then it is compared judgement with calculating real-time difference data, carry out voting method, the value exceeding Neural Network model predictive result is vetoed, the measured value meeting prediction standard is carried out summation be averaged, obtain sensor measurement optimal value, thus improving sensor measurement stability and precision。
For achieving the above object, the present invention adopts the following technical scheme that
A kind of method improving wind tunnel sensors measurement stability and precision, comprises the following steps:
Step one: for different wind-tunnel objects to be measured, select the corresponding sensor that can measure the same range of its numerical value, same model, form three groups of (or more than three groups) sensors on hardware and object to be measured is acquired simultaneously。
Step 2: when static state, input multiple standard value Standard=(Std1 to the same group of sensor with range, Std2, ..., Stdi) (Stdi represents i-th standard value) respectively obtain the measured value Measure=(Mea_i1 that each sensor is corresponding, Mea_i2 ..., Mea_ij) measured value obtained corresponding to jth standard value of i-th sensor (Mea_ij represent)。
Step 3: whether all meet the wind tunnel test certainty of measurement requirement of measured physical quantity according to the data in standard value Standard and actual measured value Measure computation and measurement value array Measure。By meet the measured value that accuracy rating requires be reassembled into array Netinput=(Netinput_i1, Netinput_i2 ..., Netinput_ij) measurement data of jth standard value of i-th sensor (Netinput_ij represent)。
Step 4: using the Netinput input as train samples data, neutral net is trained by the absolute value array of the mutual difference between sensor correspondence measured value each under same standard value as the output of neutral net, thus setting up the sensor neutral net difference forecast model of solidification。
Step 5: under test conditions, at a time, each sensor will collect the measured value Test (Test1 of object to be measured simultaneously, Test2, ..., Testj) (Testj represents jth sensor measured value at this moment), using the Test input as sensor neutral net difference forecast model, then network will automatically generate prediction output array Forecast (Forecast_12, Forecast_13, ..., Forecast_j-1, j) (Forecast_j-1, j represents the absolute value of the mutual difference between-1 sensor of jth and the jth measurement value sensor of neural network prediction)。
Step 6: calculate each sensor absolute value Realout (Realout_12 in the mutual difference of the measured value of synchronization, Realout_13, ..., Realout_j-1, j) absolute value of the mutual difference between-1 sensor of jth and jth measurement value sensor (Realout_j-1, the j represent during actual tests)。Neural network prediction is exported array Forecast and actual array Realout multilevel iudge, judged result according to each group of data carries out logic voting, if Forecast_ij < Realout_ij and Forecast_ik is < Realout_ik(i, j, k is Any Digit) then think that i-th sensor measured value occurs that error is vetoed, according to voting method result, satisfactory data recombination is created the measurement value sensor array Real (Real1 in this moment, Real2, ..., Realj) (Realj represents jth and meets the sensor values that voting method requires), undertaken suing for peace by Truth data group Real and be averaged the value Realtest obtaining this moment object to be measured;
In step one, the sensor of three groups or above range of the same race, labeled rating of the same race need to be increased, in step 6, adopt comparative neural network characteristic parameter forecast model output array and export array in real time, carry out logical judgment voting, after voting method, true value is carried out summation and is averaged。
In sum, owing to have employed technique scheme, the invention has the beneficial effects as follows:
True value, by increasing the sensor of the same range organizing labeled rating of the same race, after utilizing the standard value of standard device output performance parameter, via carrying out true value anticipation after the sensor test examined and determine, is used for building neural network characteristics parametric prediction model more by this programme;When wind tunnel test, using the status monitoring value of the sensor of sign performance of wind tunnel parameter that collects as the input of model, the neural network characteristics parametric prediction model set up is utilized to obtain the high-precision angle value of measurement of characteristic parameter in the way of Automatic Program decides by vote true value。When solving traditional single sensor experiences failure, blowing is scrapped or measured value precision is low when secondary data, and investigation cycle fault time is long, have a strong impact on quality and the progress of test, adopt new measurement strategies and data processing method, ensure the stability of measurement device and the precision of data, reach the requirement that advanced aircraft is developed。
Accompanying drawing explanation
Examples of the present invention will be described by way of reference to the accompanying drawings, wherein:
Fig. 1 is flow chart of the invention process;
Fig. 2 is the basic structure schematic diagram of neutral net。
Detailed description of the invention
This raising wind tunnel sensors that the present invention proposes measures the method for stability and precision, using the sensor of multiple same range, ad eundem precision in measurement strategies, data process and make use of establishment neural network characteristics parametric prediction model and measure monitoring calculation difference and the voting method in later stage in real time。Neutral net is a kind of imitation animal nerve network behavior feature, and Corpus--based Method carries out the algorithm mathematics model of parallel information process。Based on this theory, we are by arranging scientific and reasonable standard feature parameter experiment condition, the data set being allowed to condition in accuracy rating passes through neural network model, then the neural network characteristics parametric prediction model trained is put in wind tunnel test, put to the vote by the output array of the real-time monitoring calculation difference array of the data in later stage and forecast model, reached to improve the purpose of sensor accuracy class and stability。
Being illustrated in figure 1 the process chart (for the measurement of stagnation pressure in wind tunnel test) of an embodiment of the inventive method, this flow process includes following step:
Step one: the working pressure range according to wind-tunnel stagnation pressure, rationally selects the pressure transducer 3 of suitable same range, same to precision, same model。
Step 2: total pressure measurement gas circuit is linked into from same in-site measurement position 3 sensors of identical range。
Step 3: when static state, is added group normal pressure to same group of sensor of each range and is recorded measured value A, B, C of correspondence by sensor, (A, B, C represent the measured value that each sensor is corresponding under many group normal pressures respectively)。
Step 4: judge that measured value A, B, C are whether in accuracy rating, if in accuracy rating, it can be used as the input of train samples data, by corresponding three groups of numerical value array | A-B |, | A-C |, | B-C | (| A-B |, | A-C |, | B-C | represents the threshold value that the output valve of respective sensor exports after neural metwork training respectively) as neutral net output, neutral net is trained, thus setting up the neural network model (its network knot journey figure is as shown in Figure 2) of solidification。
Step 5: if during wind tunnel test, the measured value with three sensors of range is A ', B ', C ', it can be used as neutral net to input, and then can obtain its prediction output | A '-B ' | ' by well-established neutral net error prediction model, | A '-C ' | ', | B '-C ' | '。
Step 6: by neural network forecast result | A '-B ' | ', | A '-C ' | ', | the A '-B ' | that | B '-C ' | ' is actual with output, | A '-C ' |, measured value contrasts | B '-C ' |, data are made decisions by " three select two " the voting method of implementation, put to the vote vetoed exceeding the value that network model predicts the outcome。If | A '-B ' | < | A '-B ' | ', | A '-C ' | < | A '-C ' | ', | B '-C ' | < | B '-C ' | ', then A ', B ', C ' all meet required precision。If | A '-B ' |<| A '-B ' | ', | A '-C ' |>| A '-C ' | ', | B '-C ' |>| B '-C ' | ', then A ', B ' meet required precision。If | A '-B ' |>| A '-B ' | ', | A '-C ' |>| A '-C ' | ', | B '-C ' |<| B '-C ' | ', then B ', C ' meet required precision。If | A '-B ' |>| A '-B ' | ', | A '-C ' |<| A '-C ' | ', | B '-C ' |>| B '-C ' | ', then A ', C ' meet required precision。
Step 7: the measurement actual value meeting required precision after " three select two " voting method differentiates is carried out summation and averages, obtains the result after total pressure measurement optimizes, participates in iteration and the calculating of follow-up Flow Field in Wind Tunnel parameter。
The invention is not limited in aforesaid detailed description of the invention。The present invention expands to any new feature disclosed in this manual or any new combination, and the step of the arbitrary new method disclosed or process or any new combination。
Claims (4)
1. one kind is improved the method that wind tunnel sensors measures stability and precision, it is characterised in that comprise the following steps:
Step one: for different wind-tunnel objects to be measured, select the corresponding sensor that can measure the same range of its numerical value, same model, object to be measured is acquired simultaneously;
Step 2: when static state, inputs multiple standard values and respectively obtains the measured value that each sensor is corresponding to the same group of sensor with range;
Step 3: whether all meet the wind tunnel test certainty of measurement requirement of measured physical quantity according to the data in standard value and actual measured value computation and measurement value array, is reassembled into array by the measured value meeting accuracy rating requirement;
Step 4: using the input as train samples data of the array that is reassembled into, neutral net is trained by the absolute value array of the mutual difference between sensor correspondence measured value each under same standard value as the output of neutral net, thus setting up the sensor neutral net difference forecast model of solidification;
Step 5: under test conditions, at a time, each sensor is using the input as sensor neutral net difference forecast model of the measured value that collects object to be measured simultaneously, then network will automatically generate prediction output array;
Step 6: calculate each sensor absolute value in the mutual difference of the measured value of synchronization, adopt comparative neural network characteristic parameter forecast model output array and export array in real time, carry out logical judgment voting, after voting method, true value is carried out summation and is averaged。
2. a kind of method improving wind tunnel sensors measurement stability and precision according to claim 1, it is characterised in that described sensor includes at least three groups。
3. a kind of method improving wind tunnel sensors measurement stability and precision according to claim 2, it is characterised in that the sensor of each group is range of the same race, labeled rating of the same race。
4. a kind of method improving wind tunnel sensors measurement stability and precision according to claim 1, it is characterized in that in step 6, if the voting of described logical judgment is particularly as follows: some neural network prediction exports the number absolute value less than actual measured value, then think that this sensor measured value occurs that error is vetoed, according to voting method result, satisfactory data recombination creates the measurement value sensor array in this moment, and being undertaken suing for peace by measured value array is averaged the value obtaining this moment object to be measured。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610230341.7A CN105699043B (en) | 2016-04-14 | 2016-04-14 | A kind of wind tunnel sensors that improve measure stability and the method for precision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610230341.7A CN105699043B (en) | 2016-04-14 | 2016-04-14 | A kind of wind tunnel sensors that improve measure stability and the method for precision |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105699043A true CN105699043A (en) | 2016-06-22 |
CN105699043B CN105699043B (en) | 2018-06-08 |
Family
ID=56216179
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610230341.7A Active CN105699043B (en) | 2016-04-14 | 2016-04-14 | A kind of wind tunnel sensors that improve measure stability and the method for precision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105699043B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114486164A (en) * | 2022-04-18 | 2022-05-13 | 中国空气动力研究与发展中心空天技术研究所 | Three-hole probe gradient flow field test calibration method based on neural network |
CN114993604A (en) * | 2022-05-24 | 2022-09-02 | 中国科学院力学研究所 | Wind tunnel balance static calibration and measurement method based on deep learning |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0868722A (en) * | 1994-08-31 | 1996-03-12 | Nippon Telegr & Teleph Corp <Ntt> | Method for automatically analyzing light path characteristics |
CN102393881A (en) * | 2011-09-08 | 2012-03-28 | 杭州微松环境科技有限公司 | High-accuracy detection method of real-time temperature data fusion of multiple sensors |
CN103063233A (en) * | 2012-12-28 | 2013-04-24 | 徐州工程学院 | Method for reducing measuring errors by adopting a plurality of sensors |
CN103557884A (en) * | 2013-09-27 | 2014-02-05 | 杭州银江智慧城市技术集团有限公司 | Multi-sensor data fusion early warning method for monitoring electric transmission line tower |
CN103940453A (en) * | 2014-04-15 | 2014-07-23 | 东华大学 | Method for improving sensor measuring precision |
CN104715282A (en) * | 2015-02-13 | 2015-06-17 | 浙江工业大学 | Data prediction method based on improved PSO-BP neural network |
-
2016
- 2016-04-14 CN CN201610230341.7A patent/CN105699043B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0868722A (en) * | 1994-08-31 | 1996-03-12 | Nippon Telegr & Teleph Corp <Ntt> | Method for automatically analyzing light path characteristics |
CN102393881A (en) * | 2011-09-08 | 2012-03-28 | 杭州微松环境科技有限公司 | High-accuracy detection method of real-time temperature data fusion of multiple sensors |
CN103063233A (en) * | 2012-12-28 | 2013-04-24 | 徐州工程学院 | Method for reducing measuring errors by adopting a plurality of sensors |
CN103557884A (en) * | 2013-09-27 | 2014-02-05 | 杭州银江智慧城市技术集团有限公司 | Multi-sensor data fusion early warning method for monitoring electric transmission line tower |
CN103940453A (en) * | 2014-04-15 | 2014-07-23 | 东华大学 | Method for improving sensor measuring precision |
CN104715282A (en) * | 2015-02-13 | 2015-06-17 | 浙江工业大学 | Data prediction method based on improved PSO-BP neural network |
Non-Patent Citations (3)
Title |
---|
张永怀: "采用 BP 神经网络及其改进算法改善传感器特性", 《传感器技术学报》 * |
张铁壁: "多传感器数据融合技术在智能压力检测中的应用", 《河北工程技术高等专科学校学报》 * |
赵望达: "基于RBF神经网络提高压力传感器精度的新方法", 《传感器学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114486164A (en) * | 2022-04-18 | 2022-05-13 | 中国空气动力研究与发展中心空天技术研究所 | Three-hole probe gradient flow field test calibration method based on neural network |
CN114486164B (en) * | 2022-04-18 | 2022-06-17 | 中国空气动力研究与发展中心空天技术研究所 | Three-hole probe gradient flow field test calibration method based on neural network |
CN114993604A (en) * | 2022-05-24 | 2022-09-02 | 中国科学院力学研究所 | Wind tunnel balance static calibration and measurement method based on deep learning |
CN114993604B (en) * | 2022-05-24 | 2023-01-17 | 中国科学院力学研究所 | Wind tunnel balance static calibration and measurement method based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN105699043B (en) | 2018-06-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108038300B (en) | Optical fiber state evaluation method based on improved membership function combined with neural network | |
CN108960303B (en) | Unmanned aerial vehicle flight data anomaly detection method based on LSTM | |
CN102340811B (en) | Method for carrying out fault diagnosis on wireless sensor networks | |
CN109034191B (en) | ELM-based one-dimensional telemetry data abnormal interpretation method | |
CN105846780A (en) | Decision tree model-based photovoltaic assembly fault diagnosis method | |
CN108092623B (en) | A kind of photovoltaic array multisensor fault detecting and positioning method | |
CN104035431B (en) | The acquisition methods of kernel functional parameter and system for non-linear process monitoring | |
WO2022021726A1 (en) | Pmu-based power system state estimation performance evaluation method | |
CN111625988A (en) | Bridge health management analysis and prediction system and method based on deep learning | |
CN110555230A (en) | rotary machine residual life prediction method based on integrated GMDH framework | |
CN105606914A (en) | IWO-ELM-based Aviation power converter fault diagnosis method | |
CN111860839A (en) | Shore bridge fault monitoring method based on multi-signal fusion and Adam optimization algorithm | |
WO2021114320A1 (en) | Wastewater treatment process fault monitoring method using oica-rnn fusion model | |
CN115950609B (en) | Bridge deflection anomaly detection method combining correlation analysis and neural network | |
CN107832789A (en) | Characteristic weighing k nearest neighbor method for diagnosing faults based on the conversion of average influence Value Data | |
CN114912339A (en) | Multi-factor temperature prediction method for primary equipment of transformer substation | |
CN110222825B (en) | Cement product specific surface area prediction method and system | |
CN113486950B (en) | Intelligent pipe network water leakage detection method and system | |
CN111047476A (en) | Dam structure safety monitoring accurate prediction method and system based on RBF neural network | |
CN105699043A (en) | Method for improving measuring stability and precision of wind tunnel sensor | |
CN114189047A (en) | False data detection and correction method for active power distribution network state estimation | |
CN110414086A (en) | A kind of combined stress accelerated factor calculation method based on sensitivity | |
CN110705114B (en) | Ventilation fault diagnosis method without training sample | |
CN115659249B (en) | Abnormality detection method for intelligent station measurement control system | |
CN105741184A (en) | Transformer state evaluation method and apparatus |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |