CN101012913A - Chaos analysis and micro-processor based conduit pipe micro-leakage diagnosing method and device - Google Patents

Abstract

A tube micro-leakage diagnosing method and relative device are based on the chaos analysis and microprocessor, comprising a signal collector, a signal processor and a power module. The inventive method comprises that collecting signal; outputting the signal to the signal processor; the signal processor via blind source analysis extracts micro changeable signal, to find the leakage via its effect on the chaos turbulence mode; when the difference between the real-time chaos turbulence mode and the history mode can meet the property standard, the invention will not process continuously and when the difference is higher than the property standard, the invention finds the existence of leakage, and uses self-adaptive decouple to obtain and amplify the pressure signal and time signal of pressure change. The invention has high processing speed and wide application.

Description

Pipeline tiny leakage diagnostic method and device based on chaos analysis and microprocessor

Technical field:

The invention belongs to fluid accumulating monitor for faults Detection range, particularly a kind of pipeline tiny leakage diagnostic method and device based on chaos analysis and microprocessor.

Technical background:

The fluid accumulating has become an important constituent element in national economy and the daily life, no matter is oil, rock gas, or daily water consumption, gas pipe line or the like.The storing and transporting security of these fluids is not only the key of tremendous economic interests, also is the key of environmental protection and utilization of resources.Because fluid accumulating media is different various, requirement for the data capture of fluid line or container also is not quite similar, cause the security monitoring of FLUID TRANSPORTATION to be limited in scope, be difficult to realize for aggregation of data collection and the processing of supporting security monitoring in the fluid delivery process.

The method and apparatus of present existing fluid accumulating monitoring data collection processing aspect mostly is the single signal in the storage and transport process and gathers and transmit, and is very limited for data type that breaks down and analysing and processing.The method that adopts in the existing patented technology mainly is to be divided into the following aspects, first aspect: the measurement shoutage leak hunting method: be exactly to utilize input quantity sum in the pipeline to equal the principle of pipeline output quantity sum.But in fact, it generally is unbalanced importing and exporting instantaneous flow, if leak when causing very big measurement shoutage, can judge that roughly pipeline has abnormal conditions to take place, but because on-the-spot technological operation, as transfer and can cause measurement shoutage to rise in the processes such as valve, pressurization, its phenomenon with leak similarly, therefore simple employing measurement shoutage leak detecting may cause frequent false alarm.Second aspect: the mixing detection method that pressure loss method combines with the measurement shoutage leak hunting method: this method mainly is to utilize when pipeline takes place to leak, and the leakage point place is owing to inside and outside pressure difference, and fluid runs off rapidly, is accompanied by the decline of pressure.The fluid on leakage point both sides replenishes to the leakage point place owing to there is pressure difference, and this process is upwards downstream transmission successively, is equivalent to the leakage point place and has produced the negative pressure wave of propagating with certain speed.When negative pressure wave was delivered to first and last two stations, the force samples value descended, and then this is called as the pressure flex point constantly constantly, and this point is called as the pressure flex point.Time difference and the pressure-wave propagation speed that propagates into two ends according to suction wave positions then.In existing patent, for example: the patent No. is 200410080221.0, the applying date is on September 28th, 2004, name is called " a kind of liquid pressure conduit leakage detection method and device ", this detecting method and device use the suction wave technology by to the picking up, handle and analyze of noise signal in the pipeline, and utilize cross correlation algorithm to carry out detection and location.But this method has significant limitation.For example in actual conditions, because the permanent use of pilferage technology and pipeline, when making the pipeline generation minute leakage or the pressure loss gradual, significant change does not take place in the pipeline internal pressure along with the pressure loss of leakage point, when the minor variations of fault generation passes to pipeline two side sensers, it is very faint that negative pressure wave signal has become, and makes the force samples of station end not change, and perhaps only takes place gradual.In data analysis, method falls in pressure can not find the flex point of system and the moment that flex point takes place, thereby can not realize monitoring and location to leaking like this.

In addition, handle in fluid accumulating minute leakage and leak detection data collection and analysis, relevant device can not be finished the desired data collection and analysis of leak supervision of multiple situation effectively.The deficiency of existing patent and correlation technique mainly shows:

(1) gather signal, the collection point is comparatively single, single signal that can only a certain type of collecting treatment, versatility is poor.The collection point is single, can not carry out data collection and analysis to multiple spot on a large scale, and application is narrower.

(2) fluid accumulating monitor data analysing and processing is indifferent, can not satisfy monitoring requirement for the various data-signal collecting treatment of the operating mode of complexity, and it is wide to finish monitoring or monitoring range effectively.

(3) Zhuan Zhi data processing speed has much room for improvement.Reasons such as speed bottle-neck during owing to the complexity of instruction process and peripheral communication such as CPU and storage cause the data processing speed of system not high, can not adapt to as image processing etc. the exigent occasion of data processing rate.

Summary of the invention:

At problems of the prior art, the invention provides a kind of pipeline tiny leakage diagnostic method and device based on chaos analysis and microprocessor.

Apparatus of the present invention comprise signal gathering unit, signal processing unit and power module, signal processing unit adopts the master-slave mode dual-core architecture, constitute by embedded microcontroller and DSP digital signal processor (DSP), wherein embedded microcontroller is responsible for the coordination control of the whole device of data as the core control processing device of whole device, and DSP carries out the identification of tiny signal chaos to be handled.Embedded microcontroller links to each other with the HPI port of dsp processor by serial port, finishes coordination control and data communication between embedded microprocessor and the dsp processor; Dsp processor is finished the presence states signal acquisition process by signal gathering unit, handles that through the chaos turbulent flow analysis result is sent to embedded microprocessor; Embedded microcontroller links to each other with the GPS chip by USB interface, function when finishing the time initialization of whole apparatus system and locating the school; Power module is respectively embedded microcontroller, dsp processor power supply, as shown in Figure 1.When the user need show detection and location, apparatus of the present invention also comprised upper-position unit, and the data that this moment, embedded microprocessor transmitted dsp processor are packed and finished detection and location by the upper-position unit that Ethernet interface is uploaded on the network.

The inventive method comprises signals collecting, judge and leak and signal processing, at first utilize the various high-performance sensors and the signal gathering unit that are installed in the pipeline end to carry out signals collecting, and these signals are transferred to signal processing unit, signal processing unit is analyzed by blind source and is extracted small variable signal, influence to the chaos turbulence model determines whether to leak at this minor variations signal, when the model difference between real-time chaos turbulence model and the historical models meets performance index, do not deal with, when model difference during greater than performance index, promptly judge to have to leak to exist, obtain the pressure signal after calculating with scale-up model and the moment signal of variation in pressure by adaptive decoupling then.

Under the prerequisite of safety, optimize operation for guaranteeing pipeline, deeply grasp the dynamic rule of pipeline operating mode all sidedly, in conjunction with dynamic (dynamical) turbulent flow dynamic characteristic of pressure fluid, the inventive method is that set of equation is described the mathematical model that liquid pipeline section transient state flows with motion equation, equation of continuity, energy equation simultaneous.

In the formula: V is a flow rate of liquid, and unit is m/s; X is a distance, and unit is m; T is the time, and unit is s; P is a somewhere absolute pressure in the pipe, and unit is Pa; G is a gravity accleration, and unit is m/s ²θ is the pipeline inclination angle; λ is the flow resistance coefficient; D is an internal diameter of the pipeline, and unit is m; A is a velocity of wave, and unit is m/s; C is the fluid thermal capacitance, and unit is J/Kg ℃; T is a fluid temperature (F.T.), and unit is ℃; K is the fluid line overall coefficient of heat transfer, and unit is W/ (m ²℃); T ₀Be mean ground temperature, unit is ℃; ρ is a fluid density.

In practical engineering application, above-mentioned set of equation through dimensionless abstract following formula, i.e. the mathematical model that flows of liquid pipeline section transient state, this equation is the important foundation of research turbulent flow and space time structure, is a space-time chaos model.

$\frac{\∂A(x,t)}{\∂t}=(1+\mathrm{i\α})\frac{{\∂}^{2}A(x,t)}{{\∂x}^2}+A(x,t)-(1-\mathrm{i\β}){\left|A(x,t)\right|}^{2}A(x,t)=0$

In the formula: (x t) is complex variable to A, and it is an order parameter.α, β is accommodation coefficient.

Based on the adaptive decoupling of neuroid, be from a plurality of states, to extract the output that needs in the inventive method, adopt multilayer feedforward neural network to realize.To import data and pass to the obfuscation layer, calculate the membership function that each input vector is imported each linguistic variable value fuzzy set.

${\mathrm{\μ}}_{A_i}^k\left(x_i^p\right)=\exp [-{\left(\frac{x_i^p-{\mathrm{\α}}_i^k}{{\mathrm{\σ}}_i^k}\right)}^2]$

In the formula: μ _Ai ^k(x _i ^p) for importing x _i ^pCan release the probability of k bar rule; x _i ^pBe i variable in P the sample; α _i ^k, σ _i ^kWhen satisfying k bar rule, the center of i variable and width, the different rule of each input variable correspondence.

Adopt the principle of compensation, for every rule connects two contacts.One is passive fuzzy neuron, and mapping is input to the worst output, for the worst situation is formulated a conservative decision-making, as shown in the formula:

$u^k=\underset{i=1}{\overset{n}{\mathrm{\Π}}}{\mathrm{\μ}}_{A_i}^k\left(x_i^p\right)$

In the formula: n is a neuroid node number.

Another is positive fuzzy neuron, and mapping is input to best output, formulates an optimistic decision-making for best case, is shown below:

$v^k={\left[\underset{i=1}{\overset{n}{\mathrm{\Π}}}{\mathrm{\μ}}_{A_i}^k\left(x_i^p\right)\right]}^{\frac{1}{n}}$

Formulate a compromise relatively decision-making thus, be shown below:

$B^k={\left(u^k\right)}^{1-r}{\left(v^k\right)}^r={\left[\underset{i=1}{\overset{n}{\mathrm{\Π}}}{\mathrm{\μ}}_{A_i}^k\left(x_i^p\right)\right]}^{1-r+\frac{r}{n}}$

In the formula: 0≤r≤1, r is a compensativity.

Carry out the reverse gelatinization and calculate, obtain the exact value of network output, as shown in the formula:

$f\left(x^p\right)=\frac{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{b}^k{\mathrm{\δ}}^k{B}^k}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\δ}}^k{B}^k}$

In the formula: b ^k, δ ^kBe respectively when satisfying k bar rule, the center of output membership function and width are with the training of BP learning algorithm; M is fuzzy former piece variable number.

The present invention be fit to multiple collection signal type, processing rate accurately fast, be applicable to various complicated accumulating operating modes, can carry out analysing and processing to the particularly trickle leakage failure data of leakage failure effectively.

Description of drawings:

Fig. 1 is apparatus of the present invention structural representation;

Fig. 2 is a device circuit schematic diagram in the example,

(a) power module circuitry schematic diagram,

(b) voltage conversion circuit schematic diagram,

(c) one road signal acquisition circuit schematic diagram,

(d) serial port circuit schematic diagram,

(e) dsp processor interface circuit schematic diagram;

Fig. 3 is the software general flow chart;

Fig. 4 is flush bonding processor data-signal collecting flowchart figure;

Fig. 5 is a chaos turbulent flow analysis flow chart;

Fig. 6 is blind source analysis process figure;

Fig. 7 compares block diagram for model;

Wherein 1 is signal processing unit.

Embodiment:

The invention will be further described below in conjunction with accompanying drawing.

Apparatus of the present invention are applied to the oil transport pipeline leakage positioning system in large-scale oil field, comprise signal gathering unit, signal processing unit, power module, signal gathering unit obtains pressure, temperature and flux signal from the pipe ends sensor acquisition, and the core algorithm by flush bonding processor inside is realized the detection of leaking and location and made accurate judgment by the leakage of DSP digital signal processor to the pipeline small flow.Embedded microprocessor is selected ARMS3C44B0X for use in this example, and dsp processor is selected TMS320F2812 for use.

This device physical circuit principle as shown in Figure 2.Terminal 5VDDIO links to each other with 5,6,11,12 pins of TPS767D318 among Fig. 2 (b) among Fig. 2 (a), and terminal 3.3VDDIO, 1.8VDDIO link to each other with 3.3VDDIO, the 1.8VDDIO end of TMS320F2812 among Fig. 2 (e) respectively among Fig. 2 (b).

Signal gathering unit is by pressure, temperature transducer parallel acquisition 6 road pressure, 6 road temperature analog signals in this example, each road signal acquisition circuit is identical, shown in Fig. 2 (c), wherein the WENDU1 terminal links to each other with sensor, and the WENDU2 terminal links to each other with the ADCIN interface of dsp processor among Fig. 2 (e).

Fig. 2 (d) is depicted as the serial port circuit schematic diagram of dsp processor and arm processor, and wherein terminal SCI_OUT, SCI_IN link to each other with SCI_OUT, the SCI_IN interface of dsp processor among the figure (e) respectively, and JPMX links to each other with the HPI interface of microprocessor.

The oil transport pipeline leakage diagnosing method is as follows in this example:

At first utilize the pressure that is installed in the pipeline end, temperature transducer is gathered pressure, temperature signal, these signals are sent to signal processing unit by signal gathering unit, signal processing unit is analyzed by blind source and is extracted small variable signal, influence to the chaos turbulence model determines whether to leak at this minor variations signal, when the model difference between real-time chaos turbulence model and the historical models meets performance index, do not deal with, when model difference during greater than performance index, promptly judge to have to leak to exist, obtain the pressure signal after calculating with scale-up model and the moment signal of variation in pressure by adaptive decoupling then.Its diagnostic procedure is finally realized by the program that embeds in dsp processor and the embedded microprocessor, carries out according to the following steps, as shown in Figure 3:

Step 1: beginning;

Step 2: embedded microprocessor and dsp processor communication control and initialization;

Step 3: define program exit address and initialization I/O equipment, the house dog of initialization simultaneously, interrupt vector, device clock;

Step 4: exception response is set, and the interrupt response address is set and opens interruption;

Step 5: the initialization storage system is the sampled data storage allocation;

Step 6: the sampling interval is set and enables each sampling channel;

Step 7: routine data is carried out signals collecting to embedded microprocessor and DSP utilizes the chaos turbulent flow analysis to handle;

Step 8: packing gathers to the data result, and uploads;

Step 9: finish.

Wherein embedded microprocessor carries out signal acquisition process to routine data and may further comprise the steps, as shown in Figure 4:

Step 1: beginning;

Step 2: wait for and interrupting;

Step 3: judge interrupt requests, be followed successively by and gather the continuous signal interruption, gather discrete signal and interrupt.House dog is interrupted, and GPS interrupts;

Step 4: interrupt if gather continuous signal, then carry out the continuous signal acquisition operations;

Interrupt if gather discrete signal, then carry out the discrete signal acquisition operations;

If house dog is interrupted, then restart sampling;

If GPS interrupts, then reset the time during school;

Do not interrupt if having, then get back to step 2;

Step 5: the data-signal of gathering is carried out software filtering;

Step 6: data result gathers;

Step 7: return.

DSP chaos turbulent flow analysis process may further comprise the steps, as shown in Figure 5:

Step 1: beginning;

Step 2: data are carried out blind source analyze, signal is separated with noise;

Step 3; Set up the model of real time data;

Step 4: compare processing with historical models;

Step 5: judge whether to meet the assessed for performance index,, then carry out adaptive decoupling if meet performance index, otherwise, then turn back to step 2;

Step 6: obtain output information;

Step 7: information is issued localization process;

Step 8: return.

Blind source analytic process may further comprise the steps, as shown in Figure 6:

Step 1: beginning;

Step 2: observation survey data;

Step 2: data are carried out transform handle;

Step 4: the data after handling are set up disjunctive model;

Step 5: cost function is set;

Step 6: regulate the gain of model;

Step 7: carry out signal and separate;

Step 8: return.

Figure 7 shows that relatively block diagram of model.

The disjunctive model that adopts in the analytic process of blind source is as follows:

y ₁(t)＝ω ₁₁x ₁(t)+ω ₁₂x ₂(t-k ₁₂)

y ₂(t)＝ω ₂₂x ₂(t)+ω ₂₁x ₁(t-k ₂₁)

In the formula: { ω _IjAnd { k _IjBe respectively the weights and the time delay of recovery system.X is that observation signal, y are that separation signal, t are that time, k are sampling instant.

The cost function that is provided with is as follows:

F(ω _i，k _i)＝[E{G(y _i)}-E{G(v)}] ² i＝1，2

In the formula: G is non-quadratic function arbitrarily, and v is and y _iGaussian variable with the variance zero-mean.E is a mean square deviation.

Claims (8)

1. pipeline tiny leakage diagnostic method based on chaos analysis and microprocessor, it is characterized in that at first carrying out signals collecting, and these signals are transferred to signal processing unit, signal processing unit is analyzed by blind source and is extracted small variable signal, influence to the chaos turbulence model determines whether to leak at this minor variations signal, when the model difference between real-time chaos turbulence model and the historical models meets performance index, do not deal with, when model difference during greater than performance index, promptly judge to have to leak to exist, obtain the pressure signal after calculating with scale-up model and the moment signal of variation in pressure by adaptive decoupling then.

2. a kind of pipeline tiny leakage diagnostic method based on chaos analysis and microprocessor according to claim 1 is characterized in that described real-time chaos turbulence model is described below:

$\frac{\∂A(x,t)}{\∂t}=(1+\mathrm{i\α})\frac{{\∂}^{2}A(x,t)}{{\∂x}^2}+A(x,t)-(1-\mathrm{i\β}){\left|A(x,t)\right|}^{2}A(x,t)=0$

A in the formula (x t) is complex variable, α, and β is accommodation coefficient.

3. a kind of pipeline tiny leakage diagnostic method according to claim 1 based on chaos analysis and microprocessor, it is characterized in that described adaptive decoupling, be the output that from a plurality of states, extract to need, adopt multilayer feedforward neural network to realize, the exact value of its network output as shown in the formula:

$f\left(x^p\right)=\frac{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{b}^k{\mathrm{\δ}}^k{B}^k}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\δ}}^k{B}^k}$

In the formula: b ^k, δ ^kBe respectively when satisfying k bar rule, the center of output membership function and width are with the training of BP learning algorithm; M is fuzzy former piece variable number.

4. a kind of pipeline tiny leakage diagnostic method based on chaos analysis and microprocessor according to claim 1 is characterized in that described chaos turbulent flow analysis process specifically may further comprise the steps:

Step 1: beginning;

Step 2: data are carried out blind source analyze, signal is separated with noise;

Step 3; Set up the model of real time data;

Step 4: compare processing with historical models;

Step 5: judge whether to meet the assessed for performance index,, then carry out adaptive decoupling if meet performance index, otherwise, then turn back to step 2;

Step 6: obtain output information;

Step 7: information is issued localization process;

Step 8: return.

5. a kind of pipeline tiny leakage diagnostic method based on chaos analysis and microprocessor according to claim 4 is characterized in that blind source analytic process may further comprise the steps in the described step 2:

Step 1: beginning;

Step 2: observation survey data;

Step 3: data are carried out transform handle;

Step 4: the data after handling are set up disjunctive model;

Step 5: cost function is set;

Step 6: regulate the gain of model;

Step 7: carry out signal and separate;

Step 8: return.

6. the device that adopts based on the pipeline tiny leakage diagnostic method of chaos analysis and microprocessor according to claim 1, comprise signal gathering unit, signal processing unit and power module, it is characterized in that described signal processing unit is made of embedded microcontroller and DSP digital signal processor, embedded microcontroller links to each other with the HPI port of DSP digital signal processor by serial port, and signal gathering unit links to each other with embedded microprocessor by DSP digital signal processor; Power module is respectively embedded microcontroller, DSP digital signal processor power supply.

7. the device that adopts based on the pipeline tiny leakage diagnostic method of chaos analysis and microprocessor according to claim 6 is characterized in that also comprising the GPS chip, and the GPS chip links to each other with the USB interface of embedded microcontroller.

8. the device that adopts based on the pipeline tiny leakage diagnostic method of chaos analysis and microprocessor according to claim 6, the user it is characterized in that when need show detection and location, also comprise upper-position unit, this moment, embedded microprocessor linked to each other with upper-position unit by Ethernet interface.

Images