CN104034796A - Real-time processing device and method of internal detection data of pipeline magnetic flux leakage - Google Patents

Abstract

The invention discloses a real-time processing device and method of internal detection data of pipeline magnetic flux leakage, and belongs to the technical field of pipeline detection. The device is installed on an internal pipeline detector and comprises a magnetic flux leakage sensor unit, a signal conditioning module, an A/D conversion module and a central processing unit; the magnetic flux leakage sensor unit comprises a plurality of magnetic flux leakage sensors, and the plurality of magnetic flux leakage sensors are uniformly arranged on the internal pipeline detector along the circumferential direction of a pipeline section; the central processing unit comprises a time sequence control module, a defect data judging module, a defect data characteristic extracting module and a data storage module; the method provided by the invention can be used for quickly identifying abnormal data in defect detection, extracting the data characteristics, recording the characteristics of the abnormal data and related information and analyzing the severe defect position within 30 minutes after the detection of the internal pipeline detector is finished, so that the leakage at the severe defect position is relatively well prevented and disastrous consequences are avoided.

Description

In a kind of pipe leakage, detect data real-time processing device and method

Technical field

The invention belongs to pipe detection technical field, be specifically related to detection data real-time processing device and method in a kind of pipe leakage.

Background technology

Pipeline transportation is a kind of very important means of transportation, and for the longer pipeline of active time, it is extremely necessary that the method by Magnetic Flux Leakage Inspecting exists the position shape of defect and hazard level analysis to it.The differentiation Main Basis fault location of defect shape and hazard level detects the data characteristics of data.Therefore, the feature extraction of abnormality detection data is most important links in whole Magnetic Flux Leakage Inspecting process.

The interior detection of defect of pipeline is mainly to adopt in-pipeline detector to carry out Magnetic Flux Leakage Inspecting, and is stored in the memory storage of in-pipeline detector detecting data, is then detecting the laggard row data processing of end.The real-time of data processing and intelligent poor.Meanwhile, adopt this kind of data storage processing mode needs to be had to the memory storage of a large amount of storage spaces.And power consumption that need to be larger.Due to, oil pipeline is pipeline under the ocean especially, and between general node, distance is longer.The electric energy of in-pipeline detector is comparatively nervous.Therefore, such data storage processing mode is not very perfect.At present, also relatively less for the research of other data storage methods.

Summary of the invention

The deficiency existing for prior art, the invention provides detection data real-time processing device and method in a kind of pipe leakage.

Technical scheme of the present invention:

In pipe leakage, detect a data real-time processing device, be arranged on in-pipeline detector, comprising: leakage field sensor unit, signal condition module, A/D modular converter and CPU (central processing unit);

Described leakage field sensor unit comprises a plurality of leakage field sensors, along pipeline section circumferencial direction, be evenly arranged on in-pipeline detector, each the leakage field sensor in described leakage field sensor unit for detection of pipeline magnetic flux leakage signal output electrical signals to signal condition module;

Described signal condition module is processed with amplifying for the electric signal receiving being carried out to filtering, and filtering and the electric signal amplifying after processing are delivered to A/D modular converter;

Described A/D modular converter is for carrying out analog to digital conversion and the digital electric signal conversion being sent to CPU (central processing unit) to the electric signal receiving from signal condition module;

Described CPU (central processing unit), comprises time-sequence control module, defective data discrimination module, defective data characteristic extracting module and data memory module;

Described time-sequence control module is for controlling the conversion order of each path of A/D modular converter;

Described defective data discrimination module is for receiving the digital signal that A/D modular converter transmits, and abnormal data wherein and the validity of abnormal data are differentiated; Determine the sequence number of each effective anomaly data in the real time data receiving; The effective anomaly data that determine are sent to defective data characteristic extracting module, and the sequence number that effective anomaly data are corresponding is sent to data memory module;

The effective anomaly data that described defective data characteristic extracting module is used for receiving are extracted feature, and the feature of the effective anomaly data that extract is sent to data memory module;

Described data memory module is for storing and export the effective anomaly data characteristics receiving and sequence number thereof.

Adopt detection data real-time processing device in described pipe leakage to carry out detecting the method that data are processed in real time in pipe leakage, comprise the following steps:

Step 1: the interior detection history magnetic flux leakage data of obtaining real-time pipeline detection magnetic flux leakage data and normal pipeline section;

The duct section of take between two girth joints is unit, detects the magnetic leakage signal of the normal pipeline section of multistage, and sets up the interior detection history magnetic flux leakage data record of normal pipeline section; Normal pipeline section refers to the pipeline section that does not have defect to occur;

Step 2: according to the interior detection history magnetic flux leakage data of normal pipeline section, determine abnormal data threshold value;

From the interior detection history magnetic flux leakage data of the normal pipeline section of multistage, determine respectively the maximal value of each normal pipeline section magnetic flux leakage data, try to achieve the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage meanwhile, try to achieve the wherein mean value of one section of normal pipeline section magnetic flux leakage data; Detect one section of defect pipeline section, and try to achieve the mean value of this defect pipeline section magnetic flux leakage data; Ask for the ratio of the mean value of described defect pipeline section magnetic flux leakage data and the mean value of described one section of normal pipeline section magnetic flux leakage data; That is,

$\stackrel{\‾}{X_{\max }}=\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{X}_{\max },\stackrel{\‾}{X}=\frac{1}{N_1}\underset{i=1}{\overset{N_1}{\mathrm{\Σ}}}{X}_i,\stackrel{\‾}{X_1}=\frac{1}{N_2}\underset{i=1}{\overset{N_2}{\mathrm{\Σ}}}{X}_{\mathrm{li}},C=\frac{\stackrel{\‾}{X_1}}{\stackrel{\‾}{X}}$

In formula: n is normal pipeline section quantity; X _maxmaximal value for every section of normal pipeline section magnetic flux leakage data; for the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage; be one section of normal pipeline section magnetic flux leakage data mean value; X _iit is one section of normal pipeline section magnetic flux leakage data point; N ₁be one section of normal pipeline section magnetic flux leakage data point quantity; be one section and have defect pipeline section magnetic flux leakage data mean value; X _1ibe one section and have defect pipeline section magnetic flux leakage data point; N ₂be one section and have defect pipeline section magnetic flux leakage data point quantity; C is the ratio of the mean value of one section of defect pipeline section magnetic flux leakage data and the mean value of one section of normal pipeline section magnetic flux leakage data;

Abnormal data threshold value is

Step 3: according to abnormal data threshold value, isolate the abnormal data in real-time pipeline detection magnetic flux leakage data, and abnormal data and axial location thereof are carried out to record; Axially refer to pipe lengths;

The method of separated described abnormal data is: the real-time pipeline detection magnetic flux leakage data that is greater than abnormal data threshold k is considered as abnormal data;

Step 4: utilize correlation analysis, be adjacent the comparison of data by abnormal data, differentiate the validity of abnormal data; Adjacent data refers to the data that the sensor adjacent with the sensor that abnormal data detected detects in same axial position;

Adopt and calculate the method that abnormal data is adjacent the covariance of data, weigh abnormal data and be adjacent the correlativity between data; Because close together between two adjacent groups sensor, so two adjacent groups signal has certain similarity, if two groups of data have reached certain positive correlation degree, thinks that data acquisition is correct, isolated abnormal data is valid data; Otherwise think that data acquisition is wrong, isolated abnormal data is invalid data;

Step 5: utilize cubic spline interpolation, the effective abnormal data respectively each sensor being recorded in real time carries out cubic spline interpolation, obtains respectively the cubic spline interpolation matched curve of the effective anomaly data that each sensor records in real time, is called for short each curve;

The ordinate of described each curve is abnormal data value, the axial detection position that horizontal ordinate is abnormal data;

Step 6: respectively according to each curve, ask for the eigenwert of the effective anomaly data that each sensor records in real time;

Step 6.1: the axial eigenwert of asking for the effective anomaly data that each sensor records in real time;

Step 6.1.1: determine peak value and the valley of each curve, and calculate the peak-to-valley value of each curve; Peak-to-valley value refers to the poor of peak value and valley; The size of peak-to-valley value is relevant with the degree of depth of defect of pipeline; That is, determine the poor of maximal value, minimum value and the maximal value of the effective anomaly data that each sensor records in real time and minimum value;

Step 6.1.2: determine the abscissa value of minimum valley point and the abscissa value of time low valley point of each curve, and calculate the paddy valley of each curve; Paddy valley refers to the poor of the abscissa value of minimum valley point and the abscissa value of time low valley point; That is, determine the axial location of minimum value and the axial location of inferior minimum value of the effective anomaly data that each sensor records in real time, and the axial location of the axial location of described minimum value and time minimum value is poor; The size of paddy valley is relevant with the length of defect etching;

Step 6.1.3: the area that calculates respectively each curve;

The computing formula of each area under the curve is:

$S=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\{x_i\left(t\right)-\min [x_i\left(t\right)\left]\right\}$

In formula, S is area under the curve, x _i(t) be exceptional data point, min[x _i(t)] be minimum abnormal data;

Step 6.1.4: calculate each curve energy;

The computing formula of each curve energy is: wherein, S _efor curve energy;

Step 6.1.5: utilize the method for wavelet transformation, ask for the flex point spacing on each curve;

Method: because the corner position that will ask for is respectively between crest and two troughs, therefore only respectively the part between the minimum valley point of each curve and two troughs of inferior low valley point is carried out to wavelet transformation, the second derivative that the wavelet basis of choosing is smooth function; At the zero point of the curve after continuous wavelet transform, be the flex point of virgin curve, and the present invention chooses mexican hat wavelet as base small echo, at yardstick a=4, in the scope of a=32, abnormal data is carried out to continuous wavelet transform.According to the transform effect of concrete abnormal signal, choose wavelet scale, during general yardstick a=8, effect is best, tries to achieve the position of each knee point;

Step 6.1.6: the ratio of asking for ratio, area and the paddy valley of the peak-to-valley value of each curve and ratio, area and the peak-to-valley value of paddy valley;

Step 6.2: the circumferential eigenwert of asking for the effective anomaly data that each sensor records in real time; Circumferentially refer to pipeline section circumferencial direction;

Step 6.2.1: determine circumferential abnormal data threshold value;

Method flow is: the peak value of the effective anomaly data that record when front sensor of take is basic point, near axial location corresponding to this peak value, the peak value of the effective anomaly data that first sensor as front sensor left side or right side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if it is less than the peak value of the effective anomaly data that record when front sensor, the peak value of the effective anomaly data that sensor as front sensor right side or left side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if the peak value of the effective anomaly data that the adjacent sensors of left and right both direction records is all less than the peak value of the effective anomaly data that record when front sensor, using the peak value of the effective anomaly data that record when front sensor as final peak-peak, and carry out record to this peak-peak with when front sensor, using the half value of final peak-peak as circumferential abnormal data threshold value,

Step 6.2.2: determine circumferential abnormal data;

Method is: the sensor that is considered as circumferential abnormal data and correspondence thereof that the peak value of the effective anomaly data that under described peak-peak, the adjacent sensor of left and right both direction of sensor records is greater than circumferential abnormal data threshold value is considered as in pipeline section defect coverage, obtains the number of sensors that circumferential abnormal data and pipeline section defect cover;

And the axial location at the peak value place of the abnormal data that the sensor of pipeline section defect covering is recorded carries out record, based on sensing station difference and detection error, the residing axial location of each peak value has difference, utilize least square method to determine the position of relative error minimum, as this, organize the axial location of circumferential abnormal data peak value;

Step 6.2.3: utilize cubic spline interpolation, respectively circumferential abnormal data is carried out to cubic spline interpolation, obtain the cubic spline interpolation matched curve of circumferential abnormal data, be called for short circumferential curve;

Step 6.2.4: the area of the cubic spline interpolation matched curve of the effective anomaly data that many sensors that run of designing defect covers record in real time and;

Step 6.2.5: take with step 6.1.1 to the identical method of step 6.1.6, repeated execution of steps 6.1.1 is to step 6.1.6, the feature that obtains circumferential curve, comprising: the ratio of the ratio of ratio, area and the peak-to-valley value of peak-to-valley value, paddy valley, area, energy, flex point spacing, peak-to-valley value and paddy valley, area and paddy valley;

Step 7: the eigenwert of the effective anomaly data of output step 6 gained;

Beneficial effect:

Utilizing method of the present invention to carry out identification fast to the abnormal data in defects detection extracts with data characteristics, and only the feature of abnormal data and relevant information are carried out to record, the data volume of storage has significantly and reduces with respect to traditional method that total data is recorded.Traditional data recording mode, owing to having recorded the data volume of super large, generally needs the time of several days even tens days just can obtain complete data analysis report.And adopting said method can go out major defect position by 30 minutes inner analysis after internal detector detects, thereby better prevented that these major defects from leaking, caused catastrophic consequence.

Accompanying drawing explanation

Fig. 1 is the interior structural representation that detects data real-time processing device of the pipe leakage of one embodiment of the present invention;

Fig. 2 is the circuit theory diagrams of the signal condition module of one embodiment of the present invention;

Fig. 3 is the ADS7844 analog to digital converter of one embodiment of the present invention and the interface circuit figure of EP4CE15F17C8 FPGA;

Fig. 4 is the EP4CE15F17C8 FPGA workflow diagram of one embodiment of the present invention;

Fig. 5 is the defective data discrimination module workflow diagram of one embodiment of the present invention;

Fig. 6 is the defective data characteristic extracting module workflow diagram of one embodiment of the present invention;

The curve peak-to-valley value schematic diagram that Fig. 7 (a) is one embodiment of the present invention; (b) be the curve paddy valley schematic diagram of one embodiment of the present invention; (c) be the area under the curve schematic diagram of one embodiment of the present invention; (d) be the knee point spacing schematic diagram of one embodiment of the present invention;

Fig. 8 is the knee point spacing schematic diagram after wavelet transformation of one embodiment of the present invention.

Embodiment

Below in conjunction with accompanying drawing, one embodiment of the present invention are elaborated.

In pipe leakage in present embodiment, detect data real-time processing device, be arranged on in-pipeline detector, as shown in Figure 1, comprising: leakage field sensor unit, signal condition module, A/D modular converter and CPU (central processing unit);

Leakage field sensor unit in present embodiment comprises 96 leakage field sensors, along pipeline section circumferencial direction, be evenly arranged on in-pipeline detector, each leakage field sensor in described leakage field sensor unit all for detection of pipeline magnetic flux leakage signal output electrical signals to signal condition module;

The signal condition module of present embodiment is processed with amplifying for the electric signal receiving being carried out to filtering, and filtering and the electric signal amplifying after processing are delivered to A/D modular converter, the signal condition module of present embodiment, as shown in Figure 2, the signal receiving from Hall element is circuit filtering first after filtering, then the resistance R 2 that is 10K through resistance is connected to the inverting input 7 that model is the operational amplifier of AD824, in-phase input end 8 connects the reference voltage of 2.5V, the output terminal 6 of AD824 operational amplifier connects one end of the resistance R 3 that resistance is 20K, resistance is one end of resistance R 1 of 20K and one end of the capacitor C of 0.01pF 2, the other end of resistance R 3 connects the input end of A/D conversion chip as the output terminal of signal condition module, the inverting input of the other end concatenation operation amplifier of resistance R 1, the other end ground connection of capacitor C 2, the inverting input 7 of AD824 operational amplifier also connects one end of the capacitor C 1 of 100pF, the other end ground connection of capacitor C 1.

Under the control of A/D modular converter in present embodiment for the time-sequence control module in CPU (central processing unit), the pulse electrical signal receiving from signal condition module is carried out to analog to digital conversion and the digital signal conversion is sent to CPU (central processing unit); What the A/D modular converter in present embodiment adopted is that model is the analog to digital converter of ADS7844.

What the CPU (central processing unit) in present embodiment adopted is that model is the FPGA of EP4CE15F17C8, comprises time-sequence control module, defective data discrimination module, defective data characteristic extracting module and data memory module; Time-sequence control module in present embodiment is for controlling the conversion order of each path of A/D modular converter; Data memory module in present embodiment is for storing the abnormal data feature receiving and sequence number thereof.

The interface circuit of EP4CE15F17C8FPGA and ADS7844 analog to digital converter in present embodiment, as shown in Figure 3, ADS7844 analog to digital converter is converted to digital signal by voltage signal, 5 different output terminals of ADS7844 analog to digital converter connect respectively the self-defined I/O mouth of FPGA time-sequence control module, the CS end that is ADS7844 analog to digital converter connects I/O.71 end, the BUSY end of ADS7844 analog to digital converter connects I/O.72 end, the DCLK end of ADS7844 analog to digital converter connects the I/O.73 end of FPGA, the DIN end of ADS7844 analog to digital converter connects I/O.74 end, the DOUT end of ADS7844 analog to digital converter connects I/O.75 end.

The workflow of EP4CE15F17C8 FPGA in present embodiment, as shown in Figure 4, starts from step 401.

In step 402, digital signal in present embodiment after the conversion of ADS7844 analog to digital converter is sent into defective data discrimination module, defective data discrimination module is differentiated the abnormal data in the real time data receiving and the validity of abnormal data, and effectively abnormal data is defect of pipeline data; Determine the sequence number of each effective anomaly data in the real time data receiving; The effective anomaly data of determining are sent to defective data characteristic extracting module, and the sequence number that effective anomaly data are corresponding is sent to data memory module;

In step 403, the defective data characteristic extracting module in present embodiment is extracted feature to the effective anomaly data that receive, and the eigenwert of the effective anomaly data that extract is sent to data memory module;

In step 404, the data memory module in present embodiment is exported the eigenwert of row's ordinal sum effective anomaly data corresponding to the effective anomaly data of its storage;

The workflow of defective data discrimination module, as shown in Figure 5, starts from step 501.

In step 502, obtain the interior detection history magnetic flux leakage data of real-time pipeline detection magnetic flux leakage data and normal pipeline section;

The duct section of take between two girth joints is unit, detects the magnetic leakage signal of the normal pipeline section of multistage, and sets up the interior detection history magnetic flux leakage data record of normal pipeline section; Normal pipeline section refers to the pipeline section that does not have defect to occur;

In step 503, according to the interior detection history magnetic flux leakage data of normal pipeline section, determine abnormal data threshold value;

From the interior detection history magnetic flux leakage data of the normal pipeline section of multistage, determine respectively the maximal value of each normal pipeline section magnetic flux leakage data, try to achieve the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage meanwhile, try to achieve the wherein mean value of one section of normal pipeline section magnetic flux leakage data; Detect one section of spontaneous corrosion pipeline section, and try to achieve the mean value of this spontaneous corrosion pipeline section magnetic flux leakage data; Ask for the ratio of the mean value of described spontaneous corrosion pipeline section magnetic flux leakage data and the mean value of described one section of normal pipeline section magnetic flux leakage data; That is,

$\stackrel{\‾}{X_{\max }}=\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{X}_{\max },\stackrel{\‾}{X}=\frac{1}{N_1}\underset{i=1}{\overset{N_1}{\mathrm{\Σ}}}{X}_i,\stackrel{\‾}{X_1}=\frac{1}{N_2}\underset{i=1}{\overset{N_2}{\mathrm{\Σ}}}{X}_{\mathrm{li}},C=\frac{\stackrel{\‾}{X_1}}{\stackrel{\‾}{X}}$

In formula: n is normal pipeline section quantity; X _maxmaximal value for every section of normal pipeline section magnetic flux leakage data; for the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage; be one section of normal pipeline section magnetic flux leakage data mean value; X _iit is one section of normal pipeline section magnetic flux leakage data point; N ₁be one section of normal pipeline section magnetic flux leakage data point quantity; for spontaneous corrosion pipeline section magnetic flux leakage data mean value; X _1ifor spontaneous corrosion pipeline section magnetic flux leakage data point; N ₂for spontaneous corrosion pipeline section magnetic flux leakage data point quantity; C is the ratio of the mean value of one section of defect pipeline section magnetic flux leakage data and the mean value of one section of normal pipeline section magnetic flux leakage data;

Abnormal data threshold value is

In step 504: according to abnormal data threshold value, isolate the abnormal data in real-time pipeline detection magnetic flux leakage data, and abnormal data and axial location thereof are carried out to record; Axially refer to pipe lengths;

The method of separated described abnormal data is: the real-time pipeline detection magnetic flux leakage data that is greater than abnormal data threshold k is considered as abnormal data;

In step 505: utilize correlation analysis, be adjacent the comparison of data by abnormal data, differentiate the validity of abnormal data; Adjacent data refers to the data that the sensor adjacent with the sensor that abnormal data detected detects in same axial position;

Adopt and calculate the method that abnormal data is adjacent the covariance of data, weigh abnormal data and be adjacent the correlativity between data; Because close together between two adjacent groups sensor, so two adjacent groups signal has certain similarity, if two groups of data have reached certain positive correlation degree, thinks that data acquisition is correct, isolated abnormal data is valid data; Otherwise think that data acquisition is wrong, isolated abnormal data is invalid data;

COV(x,y)＝E[(x _i-E(x _i))(y _i-E(y _i))] (i＝1,2,…,n)

In formula, COV (x, y) is x, the covariance of two groups of data of y; x _ifor exceptional data point; y _ifor x _iconsecutive number strong point; n ₁for exceptional data point quantity; E (x) is exceptional data point expectation; E (y) is x _iconsecutive number strong point expectation;

What covariance represented is the correlativity of two variablees, if the variation tendency of two variablees is consistent, that is to say that another one is also greater than the expectation value of self if one of them is greater than the expectation value of self, so the covariance between two variablees be exactly on the occasion of.If the variation tendency of two variablees is contrary, one of them is greater than the expectation value of self, and another one is but less than the expectation value of self, and the covariance between two variablees is exactly negative value so.If x and y add up independently, the covariance between the two is exactly 0 so; The validity of the threshold decision abnormal data by covariance coefficient r, the threshold value of the covariance coefficient r that this enforcement originating party formula is determined by experiment is 0.75, the corresponding abnormal data of covariance coefficient that is greater than threshold value 0.75 is effective abnormal data;

$r=\frac{\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{(x_i-\stackrel{\‾}{x})}^2{(y_i-\stackrel{\‾}{y})}^2}}=\frac{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i{y}_i-\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i\·\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y}_i}{\sqrt{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x_i}^2-{\left(\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i\right)}^2}\·\sqrt{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y_i}^2-{\left(\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y}_i\right)}^2}}$

In step 506, utilize cubic spline interpolation, the effective abnormal data respectively each sensor being recorded in real time carries out cubic spline interpolation, obtains respectively the cubic spline interpolation matched curve of the effective anomaly data that each sensor records in real time, is called for short each curve;

The ordinate of described each curve is abnormal data value, the axial detection position that horizontal ordinate is abnormal data;

The workflow of defective data characteristic extracting module, as shown in Figure 6, starts from step 601.

In step 602, ask for the axial eigenwert of the effective anomaly data that each sensor records in real time;

A. determine peak value and the valley of each curve, and calculate the peak-to-valley value Y of each curve _p-p; As shown in Fig. 7 (a), peak-to-valley value Y _p-prefer to the poor of peak value and valley; That is, determine the poor of maximal value, minimum value and the maximal value of the effective anomaly data that each sensor records in real time and minimum value; Peak-to-valley value Y _p-psize relevant with the degree of depth of defect of pipeline;

B. determine the abscissa value of minimum valley point and the abscissa value of time low valley point of each curve, and calculate the paddy valley X of each curve _p-p; As shown in Fig. 7 (b), paddy valley X _p-prefer to the poor of the abscissa value of minimum valley point and the abscissa value of time low valley point; That is, determine the axial location of minimum value and the axial location of inferior minimum value of the effective anomaly data that each sensor records in real time, and the axial location of the axial location of described minimum value and time minimum value is poor; Paddy valley X _p-psize relevant with the length of defect etching;

C. calculate respectively the area of each curve;

As shown in Fig. 7 (c), the computing formula of each area under the curve is:

$S=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\{x_i\left(t\right)-\min [x_i\left(t\right)\left]\right\}$

In formula, S is area under the curve, x _i(t) be exceptional data point, min[x _i(t)] be minimum abnormal data; The size of area under the curve is relevant with the comprehensive condition of the length degree of depth of defect etching;

D. calculate each curve energy;

The computing formula of each curve energy is: wherein, S _efor curve energy; The size of curve energy is relevant with the comprehensive condition of the length degree of depth of defect etching;

E. utilize the method for wavelet transformation, ask for the flex point spacing on each curve;

Method: as shown in Fig. 7 (d), X _k-krepresent flex point spacing, flex point spacing X _k-ksize relevant with the length of defect etching; Because the corner position that will ask for is respectively between crest and two troughs, therefore only respectively the part between two of each curve troughs is carried out to wavelet transformation, the second derivative that the wavelet basis of choosing is smooth function; At the zero point of the curve after continuous wavelet transform, be the flex point of virgin curve, and the present invention chooses mexican hat wavelet as base small echo, at yardstick a=4, in the scope of a=32, abnormal data is carried out to continuous wavelet transform.According to the transform effect of concrete abnormal signal, choose wavelet scale, during general yardstick a=8, effect is best, tries to achieve the position of each knee point, as shown in Figure 8;

F. ask for the peak-to-valley value Y of each curve _p-pwith paddy valley X _p-pratio, area S and peak-to-valley value Y _p-pratio, area S and paddy valley X _p-pratio;

In step 603, ask for the circumferential eigenwert of the effective anomaly data that each sensor records in real time; Circumferentially refer to pipeline section circumferencial direction;

H. determine circumferential abnormal data threshold value;

Method flow is: the peak value of the effective anomaly data that record when front sensor of take is basic point, near axial location corresponding to this peak value, the peak value of the effective anomaly data that first sensor as front sensor left side or right side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if it is less than the peak value of the effective anomaly data that record when front sensor, the peak value of the effective anomaly data that sensor as front sensor right side or left side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if the peak value of the effective anomaly data that the adjacent sensors of left and right both direction records is all less than the peak value of the effective anomaly data that record when front sensor, using the peak value of the effective anomaly data that record when front sensor as final peak-peak, and carry out record to this peak-peak with when front sensor, using the half value of final peak-peak as circumferential abnormal data threshold value,

I. determine circumferential abnormal data;

Method is: the sensor that is considered as circumferential abnormal data and correspondence thereof that the peak value of the effective anomaly data that under described peak-peak, the adjacent sensor of left and right both direction of sensor records is greater than circumferential abnormal data threshold value is considered as in pipeline section defect coverage, obtains the number of sensors that circumferential abnormal data and pipeline section defect cover;

And the axial location at the peak value place of the abnormal data that the sensor of pipeline section defect covering is recorded carries out record, based on sensing station difference and detection error, the residing axial location of each peak value has difference, utilize least square method to determine the position of relative error minimum, as this, organize the axial location of circumferential abnormal data peak value;

J. utilize cubic spline interpolation, respectively circumferential abnormal data is carried out to cubic spline interpolation, obtain the cubic spline interpolation matched curve of circumferential abnormal data, be called for short circumferential curve;

The area of the cubic spline interpolation matched curve of the effective anomaly data that many sensors that K. run of designing defect covers record in real time and;

L. take the method identical with A to F, repeat A to F, the feature that obtains circumferential curve, comprising: the ratio of the ratio of ratio, area and the peak-to-valley value of peak-to-valley value, paddy valley, area, energy, flex point spacing, peak-to-valley value and paddy valley, area and paddy valley;

Although more than described the specific embodiment of the present invention, the those skilled in the art in this area should be appreciated that these only illustrate, and can make various changes or modifications to these embodiments, and not deviate from principle of the present invention and essence.Scope of the present invention is only limited by appended claims.

Claims (4)

1. in pipe leakage, detect a data real-time processing device, it is characterized in that, be arranged on in-pipeline detector, it comprises: leakage field sensor unit, signal condition module, A/D modular converter and CPU (central processing unit);

Described leakage field sensor unit comprises a plurality of leakage field sensors, described a plurality of leakage field sensor is evenly arranged on in-pipeline detector along pipeline section circumferencial direction, each leakage field sensor in described leakage field sensor unit all for detection of pipeline magnetic flux leakage signal output electrical signals to signal condition module;

Described signal condition module is for the electric signal receiving being carried out to filtering and amplify processing, and by filtering with amplify the electric signal after processing and deliver to A/D modular converter;

Described A/D modular converter is for carrying out analog to digital conversion and the digital signal conversion being sent to CPU (central processing unit) to the electric signal receiving from signal condition module;

Described CPU (central processing unit), comprises time-sequence control module, defective data discrimination module, defective data characteristic extracting module and data memory module;

Described time-sequence control module is for controlling the conversion order of each path of A/D modular converter;

Described defective data discrimination module is for receiving the digital signal that A/D modular converter transmits, and abnormal data wherein and the validity of abnormal data are differentiated; Determine the sequence number of each effective anomaly data in the real time data receiving; The effective anomaly data that determine are sent to defective data characteristic extracting module, and the sequence number that effective anomaly data are corresponding is sent to data memory module;

The effective anomaly data that described defective data characteristic extracting module is used for receiving are extracted feature, and the feature of the effective anomaly data that extract is sent to data memory module;

Described data memory module is for storing and export the effective anomaly data characteristics receiving and sequence number thereof.

2. adopt detection data real-time processing device in pipe leakage claimed in claim 1 to carry out detecting the method that data are processed in real time in pipe leakage, it is characterized in that: comprise the following steps:

Step 1: the interior detection history magnetic flux leakage data of obtaining real-time pipeline detection magnetic flux leakage data and normal pipeline section;

Detect the magnetic leakage signal of the normal pipeline section of multistage, and set up the interior detection history magnetic flux leakage data record of normal pipeline section; Normal pipeline section refers to the pipeline section that does not have defect to occur;

Step 2: according to the interior detection history magnetic flux leakage data of normal pipeline section, determine abnormal data threshold value;

Step 3: according to abnormal data threshold value, isolate the abnormal data in real-time pipeline detection magnetic flux leakage data, and abnormal data and axial location thereof are carried out to record; Axially refer to pipe lengths;

Step 4: utilize correlation analysis, be adjacent the comparison of data by abnormal data, differentiate the validity of abnormal data; Adjacent data refers to the data that the sensor adjacent with the sensor that abnormal data detected detects in same axial position;

Adopt and calculate the method that abnormal data is adjacent the covariance of data, weigh abnormal data and be adjacent the correlativity between data; The computing formula of covariance coefficient r is as follows, and the corresponding abnormal data of covariance coefficient that is greater than the threshold value of covariance coefficient r is effective abnormal data; The threshold value of covariance coefficient r is definite by testing;

$r=\frac{\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{(x_i-\stackrel{\‾}{x})}^2{(y_i-\stackrel{\‾}{y})}^2}}=\frac{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i{y}_i-\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i\·\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y}_i}{\sqrt{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x_i}^2-{\left(\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{x}_i\right)}^2}\·\sqrt{n_1\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y_i}^2-{\left(\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{y}_i\right)}^2}},i=\mathrm{1,2},\·\·\·,n_1$

In formula, x _ifor effective anomaly data point; for x _imean value; y _ifor x _iconsecutive number strong point; for x _imean value; n ₁for exceptional data point quantity;

Step 5: utilize cubic spline interpolation, the effective abnormal data respectively each sensor being recorded in real time carries out cubic spline interpolation, obtains respectively the cubic spline interpolation matched curve of the effective anomaly data that each sensor records in real time, is called for short each curve;

The ordinate of described each curve is abnormal data value, the axial detection position that horizontal ordinate is abnormal data;

Step 6: respectively according to each curve, ask for the eigenwert of the effective anomaly data that each sensor records in real time;

Step 6.1: the axial eigenwert of asking for effective anomaly data;

Step 6.1.1: determine peak value and the valley of each curve, and calculate the peak-to-valley value of each curve; Peak-to-valley value refers to the poor of peak value and valley; That is, determine the poor of maximal value, minimum value and the maximal value of the effective anomaly data that each sensor records in real time and minimum value; The size of peak-to-valley value is relevant with the degree of depth of defect of pipeline;

Step 6.1.2: determine the abscissa value of minimum valley point and the abscissa value of time low valley point of each curve, and calculate the paddy valley of each curve; Paddy valley refers to the poor of the abscissa value of minimum valley point and the abscissa value of time low valley point; That is, determine the axial location of minimum value and the axial location of inferior minimum value of the effective anomaly data that each sensor records in real time, and the axial location of the axial location of described minimum value and time minimum value is poor; The size of paddy valley is relevant with the length of defect etching;

Step 6.1.3: the area that calculates respectively each curve;

The computing formula of each area under the curve is:

$S=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\{x_i\left(t\right)-\min [x_i\left(t\right)\left]\right\}$

In formula, S is area under the curve, x _i(t) be exceptional data point, min[x _i(t)] be minimum abnormal data;

Step 6.1.4: calculate each curve energy;

The computing formula of each curve energy is: wherein, S _efor curve energy;

Step 6.1.5: utilize the method for wavelet transformation, ask for the flex point spacing on each curve;

Method: because the corner position that will ask for is respectively between crest and two troughs, therefore only respectively the part between the minimum valley point of each curve and inferior low valley point is carried out to wavelet transformation, the second derivative that the wavelet basis of choosing is smooth function; At the zero point of the curve after continuous wavelet transform, be the flex point of virgin curve;

Step 6.1.6: the ratio of asking for ratio, area and the paddy valley of the peak-to-valley value of each curve and ratio, area and the peak-to-valley value of paddy valley;

Step 6.2: the circumferential eigenwert of asking for effective anomaly data; Circumferentially refer to pipeline section circumferencial direction;

Step 6.2.1: determine circumferential abnormal data threshold value;

Method flow is: the peak value of the effective anomaly data that record when front sensor of take is basic point, near axial location corresponding to this peak value, the peak value of the effective anomaly data that first sensor as front sensor left side or right side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if it is less than the peak value of the effective anomaly data that record when front sensor, the peak value of the effective anomaly data that sensor as front sensor right side or left side next-door neighbour recorded judges, if it is more than or equal to the peak value of the effective anomaly data that record when front sensor, using this peak value as peak-peak, after the same method, the peak value of the effective anomaly data that continuation records follow-up adjacent sensors successively along this direction judges, until after the peak value of the effective anomaly data that record of a sensor be less than the peak value of the effective anomaly data that last sensor records, obtain final peak-peak, and this peak-peak and the corresponding sensor of this peak-peak are carried out to record, if the peak value of the effective anomaly data that record when the adjacent sensors of the left and right both direction of front sensor is all less than the peak value of the effective anomaly data that record when front sensor, using the peak value of the effective anomaly data that record when front sensor as final peak-peak, and carry out record to this peak-peak with when front sensor, using the half value of final peak-peak as circumferential abnormal data threshold value,

Step 6.2.2: determine circumferential abnormal data;

Method is: the sensor that is considered as circumferential abnormal data and correspondence thereof that the peak value of the effective anomaly data that under described peak-peak, the adjacent sensor of left and right both direction of sensor records is greater than circumferential abnormal data threshold value is considered as in pipeline section defect coverage, obtains the number of sensors that circumferential abnormal data and pipeline section defect cover;

And the axial location at the peak value place of the abnormal data that the sensor of pipeline section defect covering is recorded carries out record, based on sensing station difference and detection error, the residing axial location of each peak value has difference, utilize least square method to determine the position of relative error minimum, as the axial location of circumferential abnormal data peak value;

Step 6.2.3: utilize cubic spline interpolation, respectively circumferential abnormal data is carried out to cubic spline interpolation, obtain the cubic spline interpolation matched curve of circumferential abnormal data, be called for short circumferential curve;

Step 6.2.4: according to the result of step 6.1.3, the area of the cubic spline interpolation matched curve of the effective anomaly data that many sensors that run of designing defect covers record in real time and;

Step 6.2.5: take with step 6.1.1 to the identical method of step 6.1.6, repeated execution of steps 6.1.1 is to step 6.1.6, the feature that obtains circumferential curve, comprising: the ratio of the ratio of ratio, area and the peak-to-valley value of peak-to-valley value, paddy valley, area, energy, flex point spacing, peak-to-valley value and paddy valley, area and paddy valley;

Step 7: the eigenwert of the effective anomaly data of output step 6 gained.

3. in pipe leakage according to claim 2, detect the method that data are processed in real time, it is characterized in that: the method that the interior detection history magnetic flux leakage data according to normal pipeline section described in described step 2 is determined abnormal data threshold value is as follows:

From the interior detection history magnetic flux leakage data of the normal pipeline section of multistage, determine respectively the maximal value of each normal pipeline section magnetic flux leakage data, try to achieve the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage meanwhile, try to achieve the wherein mean value of one section of normal pipeline section magnetic flux leakage data; Detect one section of defect pipeline section, and try to achieve the mean value of this defect pipeline section magnetic flux leakage data; Ask for the ratio of the mean value of described defect pipeline section magnetic flux leakage data and the mean value of described one section of normal pipeline section magnetic flux leakage data; That is,

$\stackrel{\‾}{X_{\max }}=\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{X}_{\max },\stackrel{\‾}{X}=\frac{1}{N_1}\underset{i=1}{\overset{N_1}{\mathrm{\Σ}}}{X}_i,\stackrel{\‾}{X_1}=\frac{1}{N_2}\underset{i=1}{\overset{N_2}{\mathrm{\Σ}}}{X}_{\mathrm{li}},C=\frac{\stackrel{\‾}{X_1}}{\stackrel{\‾}{X}}$

In formula: n is normal pipeline section quantity; X _maxmaximal value for every section of normal pipeline section magnetic flux leakage data; for the peaked mean value of the normal pipeline section magnetic flux leakage data of multistage; be one section of normal pipeline section magnetic flux leakage data mean value; X _iit is one section of normal pipeline section magnetic flux leakage data point; N ₁be one section of normal pipeline section magnetic flux leakage data point quantity; be one section and have defect pipeline section magnetic flux leakage data mean value; X _1ibe one section and have defect pipeline section magnetic flux leakage data point; N ₂be one section and have defect pipeline section magnetic flux leakage data point quantity; C is the ratio of the mean value of one section of defect pipeline section magnetic flux leakage data and the mean value of one section of normal pipeline section magnetic flux leakage data;

Abnormal data threshold value is

4. in pipe leakage according to claim 2, detect the method that data are processed in real time, it is characterized in that: the method for isolating the abnormal data in real-time pipeline detection magnetic flux leakage data described in described step 3 is as follows:

The method of separated described abnormal data is: the real-time pipeline detection magnetic flux leakage data that is greater than abnormal data threshold value is considered as abnormal data.