CN100370385C - Method for identifying boiler drum level model - Google Patents

Abstract

The present invention relates to a method for identifying water level models of a boiler steam drum in the technical field of computer application. The present invention is additionally provided with an offset relay identifying module added with hysteresis. A periodic oscillation limit cycle of the system is analyzed under the relay characteristics function, and four parameter identifying formulas of the water level of the steam drum are obtained. The present invention provides a novel on-line and accurate relay identifying method aiming to the common water level problem of the steam drum in the chemical process. Under the conditions of no need of priori information of any time lag, static gain, etc., according to accurate parse expression obtained from time domain analysis, the present invention accurately calculates four parameters to be identified, and thus, the obtained time lag model with reverse response is matched with the physical system.

Description

The method of identifying boiler drum level model

Technical field

What the present invention relates to is the method in a kind of Computer Applied Technology field, specifically is a kind of method of identifying boiler drum level model.

Background technology

The disturbance that influences boiler drum level mainly contains four sources: the one, and the disturbance of feedwater aspect; The 2nd, the variation of steam load; The 3rd, the variation of fuel quantity; The 4th, the variation of drum pressure.Wherein, the factor that influences the steam water-level variation mainly is feedwater flow, steam flow and fuel quantity, selected steam water-level is as output variable y, feedwater flow is as control variable u, the dynamic perfromance of steam water-level object under the discharge effect then, be made up of integral element, reverse response link, first order inertial loop and time lag link, its transfer function model G (s) can be expressed as

$G\left(s\right)=\frac{Y\left(s\right)}{U\left(s\right)}=\frac{K_p(-{\mathrm{\τ}}_{1}s+1)}{({-\mathrm{\τ}}_{2}s+1)s}{e}^{-\mathrm{\θs}}$

Here K _pBe static gain, τ ₁And τ ₂Be time constant, θ is a time lag.Current outstanding problem is how accurately to pick out this four model parameters.Employing is during based on the discrimination method (for example step response method) of routine, can not pick out all model parameters simultaneously, need estimation time lag and static gain size in advance, and identification precision is not high, particularly being subjected to outside noise easily disturbs, the controller that this out of true identification deviation can cause designing in view of the above lost efficacy, and serious meeting causes that drum breaks or explodes, and this is very unfavorable and harmful to actual production.Therefore accurately the above-mentioned single order integration of identification reverse response time-lag process is a very difficult and problem demanding prompt solution in the chemical industry practice, must adopt advanced identification technique to solve.

 str  m, K.J. wait 1984 and proposed the relay feedback identification method the earliest, this method can ensure the stable closed loop oscillatory response of controlled process, thereby acquisition process critical message easily, therefore adjusting for the PID controller parameter provides the basis, has been widely used in recent years in the industrial PID controller parameter self-adjusting system.The relay feedback identification method of many improvement and expansion is suggested in succession.

Through the literature search of prior art is found, at stablizing Object with Time Delay, Wang, Q.G. is at document Low-order modeling from relay feedback, and (the low order modeling method based on relay feedback is published in Industrial ﹠amp; Engineering Chemistry Research, chemical engineering industry and engineering research, 1997,36 (2), 375-381.) biased relay of the stagnant ring of middle first Application band obtains more accurate identification result.But because the dynamic perfromance of boiler drum level object is more special, existing integration and time lag link must be considered the pairing reverse response link of " false water level " phenomenon again.Wang, the method for Q.G. only is applicable to that identification comprises the stable Object with Time Delay of three parameters.Luyben, W.L. document Identification and tuning of integrating processes with deadtime andinverse response (have time lag and inverse response the integration object identification and adjust, be published in Industrial ﹠amp; Engineering Chemistry Research, chemical engineering industry and engineering research, 2003,42,3030-3035.) in a kind of step response identification method has been proposed, but this method not only needs to know in advance time lag, and is having under the situation of interference noise identification precision very poor.How designing relay feedback tests on-line identification fast and effectively and goes out four model parameters, how to improve identification precision, how to overcome external interference and model mismatch problem, all do not solved for the control difficult problem of system completely always thereby alleviate the identification error band.

Summary of the invention

The objective of the invention is in the boiler drum level control system common in the chemical process, the deficiency of existing identification technique, a kind of method of identifying boiler drum level model is proposed, can be under the situation of prior imformations such as time lag and static gain, obtain accurate analytical expression according to time-domain analysis to the periodic oscillation limit cycle, thereby accurately calculate parameter to be identified, reverse response Time-Delay model and real system are complementary, and then provide the basis for designing steam water-level PID controller.

The present invention is achieved by the following technical solutions, specifically may further comprise the steps:

Step 1: biased relay output valve μ ₊, μ _-Ring ε known parameters is delivered among the storage unit RAM with stagnating.

Step 2: by the relay identification test, host computer writes down the input-output function u and the y of identification process, reads the mean value of a plurality of stable oscillation stationary vibration limit cycle cycle parameters, i.e. cycle P _U1, P _U2, amplitude A _p, A _d, the time interval

With

, each parameter is delivered among the storage unit RAM.

Step 3: call the algorithm routine that weaves in advance and resolve four parameters that pick out system model, the specific algorithm step is as follows:

A) read μ in the internal memory ₊, μ _-, ε, P _U1, P _U2,

With

Substitution equation (1) formula is calculated timeconstant to be identified ₂

${-\mathrm{\τ}}_2\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{{-P}_{u1}/{\mathrm{\τ}}_2})}{-{\mathrm{\μ}}_{\_}(1-e^{{-P}_{u2}/{\mathrm{\τ}}_2})}\right)={\tilde{t}}_{\min }-{\tilde{t}}_{\max }---\left(1\right)$

B) with τ ₂The A in the internal memory is read in substitution system of equations (2), (3) and (4) _pAnd A _d, calculate timeconstant to be identified ₁

$\mathrm{\θ}={\tilde{t}}_{\min }+{\mathrm{\τ}}_2\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{{-P}_{u2}/{\mathrm{\τ}}_2})}\right)---\left(2\right)$

$K_p=\frac{A_p-A_d}{({\mathrm{\μ}}_{+}{P}_{u1}+{2\mathrm{\μ\τ}}_1)-{\mathrm{\τ}}_2[\mathrm{\μ}\_\ln \left(\frac{-\mathrm{\μ}\_(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{-P_{u1}/{\mathrm{\τ}}_2})}\right)-{\mathrm{\μ}}_{+}\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{-P_{u2}/{\mathrm{\τ}}_2})}\right)]}---\left(3\right)$

$-K_p(\mathrm{\μ}\_P_{u2}+2\mathrm{\μ\θ})-2K_p\mathrm{\μ}({\mathrm{\τ}}_1+{\mathrm{\τ}}_2)[1-\frac{e^{\mathrm{\θ}/{\mathrm{\τ}}_2}}{1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2}}(e^{-P_{u1}/{\mathrm{\τ}}_2}+e^{-P_{u2}/{\mathrm{\τ}}_2}-{2e}^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})]=2\mathrm{\ϵ}---\left(4\right)$

C) with τ ₁And τ ₂Substitution equation (2) calculates parameter θ to be identified.

D) with τ ₁And τ ₂Substitution equation (3) calculates parameter K to be identified _p

Above various in,

K _p---object static gain, τ ₁---time constant of object one, τ ₂---time constant of object two,

θ---object time lag, μ ₊---relay output high level, μ _----the relay output low level,

The stagnant ring of ε---biased relay, P _U1---holding time of relay output high level,

P _U2---holding time of relay output low level, A _p---the positive peak of object output oscillating curve,

A _d---the negative peak of object output oscillating curve,

---jump to the time interval that object reaches positive peak from relay,

---jump to the time interval that object reaches negative peak from relay.

The present invention adds the biasing relay identification module that band stagnates and encircles in existing chemical industry boiler drum level control system, by the periodic oscillation limit cycle of system under the relay characteristics effect analyzed, obtain the identification formula of four parameters in the steam water-level model.

The present invention can directly move enforcement on existing industrial computer, on-line identification goes out the boiler drum level model.Application process of the present invention can roughly be divided into two stages, and the first, at the boiler drum level debug controller parameter initial stage, under the relay effect, whole closed-loop system enters the stable oscillation stationary vibration process.In this process, note relay and object output array respectively, and be kept among the RAM, by simple data analysis, obtain the measurement information of a plurality of keys.The second, after a plurality of critical quantity measurement informations of oscillation period have deposited the industrial computer real-time data base in, call the program parsing that weaves in advance and pick out the system model parameter.Innovation draws this algorithm on the basis of accurately time-domain analysis just.The a complete set of adjustment process of the present invention can be finished on industrial computer Simulation Control interface.

In actual Chemical Engineering Process Control field by using the inventive method, only need the experiment of single relay identification accurately to obtain four parameters according to above analytic method, thereby easy and simple to handlely and can disturb by effectively overcoming noise rapidly, thereby overcome the major defect of traditional discrimination method.The discrimination method that the present invention provides can be widely used in the identification of the reverse response production run of band time lag in the industries such as the energy, metallurgy, petrochemical industry, thermal power generation, weaving.

Description of drawings

Fig. 1 is relay identification theory of constitution figure of the present invention;

Fig. 2 uses the characteristic of the stagnant ring of band biased relay for the present invention;

Fig. 3 is system and relay characteristics output waveform;

Fig. 4 is industrial computer Simulation Control interface;

Fig. 5 is the identification algorithm program flow diagram;

Fig. 6 is that specific embodiment is used relay identification method of the present invention and existing other discrimination method identification result comparison diagrams.

Embodiment

Below in conjunction with drawings and Examples technical scheme of the present invention is further described.

Embodiment:

For a chemical industry boiler steam water-level object, as follows in the dynamic perfromance of forward direction feedwater flow under the discharge effect:

$0.6\frac{-0.4s+1}{(s+1)s}{e}^{-0.1s}$

Use the discrimination method that the present invention provides, the first step: set up a relay identification system according to structural drawing shown in the accompanying drawing 1; And the biased relay correlation parameter delivered among the storage unit RAM.

μ ₊＝1.2，μ _-＝-0.8，ε＝0.2

Second step: by the relay identification test, host computer writes down the input-output function u and the y of identification process, reads the mean value of 10 stable oscillation stationary vibration limit cycle cycle parameters, i.e. cycle P _U1=2.8804, P _U2=4.3206, amplitude A _p=0.7952, A _d=-0.5838, the time interval ${\tilde{t}}_{\min }=1.2959$ With ${\tilde{t}}_{\max }=0.9348$ Also deliver among the storage unit RAM.

The 3rd step; Call the algorithm routine that weaves in advance and resolve four parameters that pick out system model:

A) read μ in the internal memory ₊, μ _-, ε, P _U1, P _U2,

With

Substitution equation (1) formula is calculated timeconstant to be identified ₂=1.0000.

B) with τ ₂The A in the internal memory is read in substitution system of equations (2), (3) and (4) _pAnd A _d, calculate timeconstant to be identified ₁=0.3999.

C) with τ ₁And τ ₂Substitution equation (2) calculates parameter θ to be identified=0.1002.

D) with τ ₁And τ ₂Substitution equation (3) calculates parameter K to be identified _p=0.6000.

What more than set forth is the accurate identification effect that a embodiment that the present invention provides shows.It may be noted that the present invention is not only limited to the foregoing description, the present invention is directed in the industrial process general reverse response time-lag process model and provided the discrimination method of relay feedback, so be applicable to the various production runes that this dynamic perfromance is arranged.

Claims (2)

1. the method for an identifying boiler drum level model is characterized in that, comprises the steps:

Step 1: biased relay output valve μ ₊, μ _-Ring ε known parameters is delivered among the storage unit RAM with stagnating;

Step 2: by the relay identification test, host computer writes down the input-output function u and the y of identification process, reads the mean value of a plurality of stable oscillation stationary vibration limit cycle cycle parameters, i.e. cycle P _U1, P _U2, amplitude A _p, A _d, the time interval

With

Each parameter is delivered among the storage unit RAM;

Step 3: call the algorithm routine that weaves in advance and resolve four parameters that pick out system model, the specific algorithm step is as follows:

A) read μ in the internal memory ₊, μ _-, ε, P _U1, P _U2, With

Substitution equation (1) formula is calculated timeconstant to be identified ₂:

${-\mathrm{\τ}}_2\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{-P_{u1}/{\mathrm{\τ}}_2})}{-{\mathrm{\μ}}_{-}(1-e^{{-P}_{u2}/{\mathrm{\τ}}_2})}\right)={\tilde{t}}_{\min }-{\tilde{t}}_{\max }---\left(1\right)$

B) with τ ₂The A in the internal memory is read in substitution system of equations (2), (3) and (4) _pAnd A _d, calculate timeconstant to be identified ₁:

$\mathrm{\θ}={\tilde{t}}_{\min }+{\mathrm{\τ}}_2\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{-P_{u2}/{\mathrm{\τ}}_2})}\right)---\left(2\right)$

$K_p=\frac{A_p-A_d}{({\mathrm{\μ}}_{+}{P}_{u1}+{2\mathrm{\μ\τ}}_1)-{\mathrm{\τ}}_2[{\mathrm{\μ}}_{-}\ln \left(\frac{-{\mathrm{\μ}}_{-}(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{-P_{u1}/{\mathrm{\τ}}_2})}\right)-{\mathrm{\μ}}_{+}\ln \left(\frac{{\mathrm{\μ}}_{+}(1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})}{2\mathrm{\μ}({\mathrm{\τ}}_1/{\mathrm{\τ}}_2+1)(1-e^{-P_{u2}/{\mathrm{\τ}}_2})}\right)]}---\left(3\right)$

$-K_p({\mathrm{\μ}}_{-}{P}_{u2}+2\mathrm{\μ\θ})-{2K}_p\mathrm{\μ}({\mathrm{\τ}}_1+{\mathrm{\τ}}_2)[1-\frac{e^{\mathrm{\θ}/{\mathrm{\τ}}_2}}{1-e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2}}(e^{-P_{u1}/{\mathrm{\τ}}_2}+e^{-P_{u2}/{\mathrm{\τ}}_2}-2e^{-(P_{u1}+P_{u2})/{\mathrm{\τ}}_2})]=2\mathrm{\ϵ}---\left(4\right)$

C) with τ ₁And τ ₂Substitution equation (2) calculates parameter θ to be identified;

D) with τ ₁And τ ₂Substitution equation (3) calculates parameter K to be identified _p

Above various in,

K _p---object static gain, τ ₁---time constant of object one, τ ₂---valve constant two during object,

θ---object time lag, μ ₊---relay output high level, μ _----the relay output low level,

The stagnant ring of ε---biased relay, P _U1---holding time of relay output high level,

P _U2---holding time of relay output low level, A _p---the positive peak of object output oscillating curve,

A _d---the negative peak of object output oscillating curve,

---jump to the time interval that object reaches positive peak from relay,

---jump to the time interval that object reaches negative peak from relay.

2. the method for identifying boiler drum level model according to claim 1 is characterized in that: directly operation is implemented on existing industrial computer, and a complete set of adjustment process is finished on industrial computer Simulation Control interface.

Images