CN102353475A - Distributed grating temperature measurement method based on two-time data fusion technology - Google Patents

Abstract

The invention provides a distributed grating temperature measurement method based on a two-time data fusion technology. The method comprises the following steps of: by using a fiber Bragg grating as a temperature measurement element, measuring a temperature value by detecting a reflection signal of the grating; performing analog-to-digital conversion on the measured temperature value, and sampling; performing first-time tail-cutting weighting data fusion on multi-channel data which is sampled by using a microprocessor, and estimating the influence of interference factors such as cross sensitivity on the measuring accuracy of the grating; replacing the maximum value and the minimum value in the original data by using a tail-cutting weighting data fusion result, and thus obtaining a group of new data; and performing second-time fusion weighting minimum mean square error fusion on the new data, and thus obtaining a fusion result which serves as a final output result. The method is simple in implementation; and by adoption of the method, the problems of difficulty in elimination of the cross sensitivity and low measurement accuracy in the conventional grating measurement are solved, and the robustness of a measurement system is improved.

Description

Distributed grating temperature measuring method based on twice data fusion technology

Technical field

The present invention relates to a kind of multisensor field of measuring technique, particularly a kind of grating temperature sensor measuring method and data processing technique.

Background technology

Grating sensor because of its do not receive the influence of electromagnetic interference (EMI), strong and be convenient to embed and be widely used in fields such as heavy construction, oil field, colliery in the testee to the corrosive atmosphere adaptive faculty.Because grating is convenient to be connected to form distributed sensing system with optical fiber, is applicable to large-scale environment measurement and monitoring.Grating temperature is measured because the heat resistance of grating is good, and distributed optical grating measuring system is convenient to lay and adaptable across the occasion of oil pipe, oil tank and natural gas line.But the grating temperature sensing receives the influence of cross sensitivity easily; Be that grating is simultaneously to temperature and stress sensitive and make measurement data inaccurate; Traditional solution is eliminated the influence of cross sensitivity through compensation; Promptly will make improvements from measurement structure, this can cause measuring system complex structure and cost higher, poor practicability.Grating sensing networking simultaneously causes measurement structure complicated, and maintenance cost is high, the data processing difficulty.Especially in distributed sensing, cross sensitivity causes the measured value difference of each grating in the grating temperature Measurement Network bigger, is difficult to obtain a temperature value accurately, and need compensate each measuring unit, and this can cause the rapid rising of cost.Because the temperature field is distributed as continuous distribution; Its rate of change is less so can adopt the data fusion technology that the grating sensing data are handled, and promptly can save system cost in the grating temperature measuring system and can also eliminate the influence of cross sensitivity simultaneously and improve and measure accuracy thereby the data fusion technology is dissolved into.

The data fusion technology is the focus of multisensor measurement and the research of sensor network field of measurement.Generally speaking improving the accuracy of sensing measurement data can use two kinds of methods to realize: the one, improve transducer sensitivity; The 2nd, adopt the data fusion technology.Because sensor is subject to the influence of surrounding environment, and high sensor receives external interference particularly serious, be unable to do without sensor Data Fusion so improve the sensing measurement data precision at present.The sensor measurement precision is determined by transducer sensitivity and sensor antijamming capability, and high sensitivity causes antijamming capability to descend usually, so high sensor is made restriction to environment for use usually.Various countries utilize data fusion to carry out the multisensor Measurement and Data Processing at present; Make full use of a plurality of sensor resources; Through reasonable domination and use to multisensor and observation data thereof; Multisensor redundant or complementary information on room and time, make up according to certain algorithmic rule, explain to obtain the consistent of measurand.The at present widely used algorithm that is used to improve multisensor measurement data precision comprises weighted least square algorithm, weighted least mean square ERROR ALGORITHM and Kalman filter method; Wherein weighted least square algorithm and weighted least mean square ERROR ALGORITHM need multisensor measurement data accord with normal distribution just can reach the optimum fusion effect; And Kalman filter method fusion accuracy depends on measuring the precision of the Kalman Filtering Model of setting up; Be difficult to accurate modeling and the interference of actual measurement environment is numerous, so the versatility of Kalman filter method is poor.In actual measurement, find; Often concentrated near the maximal value and minimum value in the sensing data by the sensing data of serious interference; A data fusion is difficult to reach the effect of optimum, if but just can use weighted least square algorithm, weighted least mean square ERROR ALGORITHM and Kalman filter method to carry out data fusion these part rejectings that receive to disturb in the data of serious interference.Rejecting the way of disturbing in the data promptly is to reject maximal value and minimum value, and remaining data is carried out data fusion, to obtain an estimated value, at this moment can replace former maximal value, minimum value to carry out data fusion again one time with estimated value.

In sum, above grating temperature measuring system exists cross sensitivity to influence measuring accuracy and penalty method is eliminated the problem that cross sensitivity can cause system architecture to complicate and cost rises, and is difficult to reach the optimum fusion result and directly carry out data fusion.

Summary of the invention

The present invention seeks to solve the grating sensor that exists in the prior art and be subject to disturb and cause the inaccurate problem of measurement result, provide a kind of based on twice technological distributed grating temperature measuring method of data fusion.

The concrete steps of the distributed grating temperature measuring method based on twice data fusion technology provided by the invention are:

1st, the light signal that wideband light source is sent is earlier after resolve into n road light signal behind optoisolator and the multistage photo-coupler, n is greater than 2;

2nd, go up n road light signal after the step decomposes again respectively through after the distributed fiber grating temperature sensor reflection; Each gets into one 2 * 2 photo-coupler and is divided into the identical light signal of two-way; Wherein lead up to the hypotenuse wave filter; Another road does not process, and two ways of optical signals is surveyed with photodetector more simultaneously, and converts light signal into current signal;

3rd, the current signal after the 2nd conversion that obtains of step is nursed one's health, current signal is adjusted to the voltage signal that is fit to the data acquisition scope;

4th, the electric signal after the 3rd step reason is carried out data acquisition, the data that analog electrical signal converted to digital signal and store storage comprise distributed fiber grating temperature sensor numbering and corresponding digital signal;

5th, data fusion for the first time

The data of 5.1st, at first the 4th step being stored sort by the temperature value size, obtain one group of raw data, write down the sensor number at temperature maximal value and minimum value place simultaneously;

5.2nd, the raw data that adopts back-end crop weighted data blending algorithm that the 5.1st step was obtained is carried out the data fusion first time; The measured temperature of the temperature maximal value that replaces being write down with fusion results then and the sensor number correspondence at minimum value place forms one group of new data;

6th, data fusion for the second time

6.1st, the one group of new data that the 5th step was formed after the data fusion through the first time adopts the weighted least mean square error to carry out the data fusion second time, gets fusion results to the end;

6.2nd, the fusion results that obtains with the 6.1st step is exported as the final measured value of system.

7th, output unit displays temperature.

Advantage of the present invention and beneficial effect:

The present invention proposes a kind of distributed grating temperature measuring method that merges based on secondary; Use twice data anastomosing algorithm in this method; Data fusion mainly is used for overcoming the measured deviation that cross sensitivity causes for the first time; For the second time data fusion is used to handle the measurement noise of accord with normal distribution, can improve the accuracy of measurement of distributed grating sensing through twice data fusion.

This method is not only applicable to distributed grating temperature measuring system, is equally applicable to other distributed measurement systems, as long as sensing data demonstrates tangible tendentiousness, promptly obeys skewness and distributes, and this system and method can be used for improving the accuracy of measurement.

This invention realizes simple, and is not high to System Hardware Requirement, and when in the system break down in a certain road, can effectively overcome this fault, and promptly the robustness of system is enhanced.

Description of drawings

Fig. 1 is that system of the present invention forms schematic diagram.

Fig. 2 is the process flow diagram of data fusion part among the present invention.

Fig. 3 is of the present invention in order to distributed grating temperature sensing measurement network system schematic diagram.

Among the figure, 1 isolator, 2 first coupling mechanisms, 3 second coupling mechanisms (totally two 21,22); 4 the 3rd coupling mechanisms (totally six, 31 to 36), (totally six of 5 optical fiber bragg grating FBGs; Be FBG the the 1st to FBG the 6th), 6 the 4th coupling mechanisms (totally six, 41 to 46); 7 hypotenuse wave filters (totally six), 8 and 9 photodetectors (totally ten two), 10 signal conditioning circuits (totally ten two).

Below in conjunction with accompanying drawing the present invention is further specified.But the invention is not restricted to given example.

Embodiment

Embodiment

As shown in Figure 3, the 1 the tunnel to be example explanation, this example is in enclosed environment, to measure.

One, distributed grating temperature is measured: the light that sends from wideband light source gets into first coupling mechanism 2 through isolator 1 back; Get into Fiber Bragg Grating FBG 5 through second coupling mechanism 3 and the 3rd coupling mechanism 4 again; Here grating receive factors such as temperature influence with a part of light reflected back the 3rd coupling mechanism 4; These reflected light are divided into the identical light signal of two-way through the 4th coupling mechanism 6 again, wherein lead up to hypotenuse wave filter 7, and another road does not process.

Two, photodetection: two ways of optical signals is surveyed with photodetector 8 and 9 respectively simultaneously, and the signal that passes through hypotenuse wave filter 7 this moment obtains current signal and does after photodetector 9 is surveyed

The current signal that obtains after photodetector 8 is surveyed through the signal of hypotenuse filtering does

Wherein A is the hypotenuse filter slope, and reflection peak is λ, λ ₀Centre wavelength when not influenced by measuring amount, Δ λ is the reflection peak bandwidth, I ₀Be the light intensity of Fiber Bragg Grating FBG 5 incident lights, R is the reflectivity of Fiber Bragg Grating FBG 5.

Three, signal conditioning circuit: the electric signal that then detection is obtained carries out signal condition, promptly amplifies through signal conditioning circuit 10 constant amplitude I-V, voltage signal is adjusted in the full width of cloth range of part of data acquisition.

Four, data acquisition and storage: the two paths of signals to obtaining is done division; It is output as

, and wherein A and are constant; So can solve λ, λ and temperature have one-to-one relationship in temperature survey.Carry out data acquisition, quantization encoding then.Data to obtaining are stored, and storage comprises two parts information, and a part is a sensor number, and another part is the temperature value that measures.

Adopt Fig. 3 structure that per second acquisition rate temperature is once carried out 10 times and measure, the result is as shown in table 1.Thereby exist temperature and stress cross sensitivity to influence the accuracy of measurement of temperature in this example.Data unit is ℃ in the table.FBG the 1st to the 6th is six gratings among Fig. 3 in the table.

Table 1

Data fusion describes to be measured as example for the first time.

At first data are sorted, order from big to small is 1.{FBG 2:26.9}2.{FBG 4:25.5}3.{FBG 3:24.6}4.{FBG 1:24.4}5.{FBG 5:24.3}6.{FBG 6:22.1}.

Record max min place number of sensor, the output result of FBG the 2nd is a maximal value, the measurement result of FBG the 6th is a minimum value, so record FBG the 2nd and FBG the 6th.

Measurement data to after sorting is carried out the data fusion first time, adopts back-end crop weighting fusion algorithm, and what notice real participation fusion this moment is the measured value of FBG the 4th, FBG the 3rd, FBG the 1st and FBG the 5th, and back-end crop weighting fusion computing method are:

${\hat{y}}_k=\frac{1}{N(1-2\mathrm{\α})}[(1-r_{\mathrm{\α}})(x_{(N_{\mathrm{\α}}+1)k}+x_{(N-N_{\mathrm{\α}})})+\underset{N_{\mathrm{\α}}+2}{\overset{N-N_{\mathrm{\α}}-1}{\mathrm{\Σ}}}{x}_{\mathrm{jk}}]$

X wherein _JkRepresentative is in j TEMP value of k rank order amount constantly, and α is the ratio that sample accounts for total sample of cutting away, and is 0.34 in this example.N _αBe the integral part of α N, N is a number of sensors, and N is 6 in this example, r _αFraction part for α N.It is 24.2 ℃ that the substitution data computation can get back-end crop weighting fusion result.With 24.2 ℃ of measurement data that replace FBG the 2nd and FBG the 6th, form one group of new measurement data, as shown in table 2.Data unit is ℃ in the table.

Table 2

The data of his-and-hers watches 2 carry out merging the second time again; Adopt the weighted least mean square error to merge; The weighting fusion that promptly power is

; Wherein

is the variance of new data set respective sensor; The result who obtains is 24.3 ℃, is 24.3 ℃ so measure exact value for the first time.

Data in the table 1 are output as table 3 through after the system

Table 3

Table 4

Table 4 is that this method compares with the fusion output variance that directly adopts minimum variance and weighting back-end crop averaging method; Obviously the output variance of this method is less than direct fusion; This shows that the data fluctuations that obtains through this method processing is less; Because the enclosed environment temperature can not have than great fluctuation process in 10 seconds, so this method effectively raises the accuracy of measurement.

Claims (1)

1. distributed grating temperature measuring method based on twice data fusion technology is characterized in that the concrete steps of this method are:

1st, the light signal that wideband light source is sent is earlier after resolve into n road light signal behind optoisolator and the multistage photo-coupler, n is greater than 2;

2nd, go up n road light signal after the step decomposes again respectively through after the distributed fiber grating temperature sensor reflection; Each gets into one 2 * 2 photo-coupler and is divided into the identical light signal of two-way; Wherein lead up to the hypotenuse wave filter; Another road does not process, and two ways of optical signals is surveyed with photodetector more simultaneously, and converts light signal into current signal;

3rd, the current signal after the 2nd conversion that obtains of step is nursed one's health, current signal is adjusted to the voltage signal that is fit to the data acquisition scope;

4th, the electric signal after the 3rd step reason is carried out data acquisition, the data that analog electrical signal converted to digital signal and store storage comprise distributed fiber grating temperature sensor numbering and corresponding digital signal;

5th, data fusion for the first time

The data of 5.1st, at first the 4th step being stored sort by the temperature value size, obtain one group of raw data, write down the sensor number at temperature maximal value and minimum value place simultaneously;

5.2nd, the raw data that adopts back-end crop weighted data blending algorithm that the 5.1st step was obtained is carried out the data fusion first time; The measured temperature of the temperature maximal value that replaces being write down with fusion results then and the sensor number correspondence at minimum value place forms one group of new data;

6th, data fusion for the second time

6.1st, the one group of new data that the 5th step was formed after the data fusion through the first time adopts the weighted least mean square error to carry out the data fusion second time, gets fusion results to the end;

6.2nd, the fusion results that obtains with the 6.1st step is exported as the final measured value of system;

7th, output unit displays temperature.

Images