CN101847004A - Method for performance evaluation and failure diagnosis of coke oven multi-loop control system - Google Patents

Abstract

The invention provides a method for the performance evaluation and the failure diagnosis of a coke oven multi-loop control system; the method for the performance evaluation of the coke oven multi-loop control system adopts a fuzzy comprehensive evaluation method to calculate performance level of a coke oven heating combustion control system; the fuzzy comprehensive evaluation method includes two evaluation factors: improved difference coefficient and process capability index; the method for the failure diagnosis comprises the steps: firstly establishing a failure knowledge base based on the improved difference coefficient and the process capability index of multi-loop control, searching for all the possible failure resources according to the optimal matching principal by regarding the improved difference coefficient and the process capability index as index in case that the performance level of the coke oven heating combustion control system is qualified or below, and embodying these failure sources visually in the form of text. The method enhances automation degree of failure elimination, dramatically lowers labor intensity of staff and raises working efficiency.

Description

The Performance Evaluation of coke oven multi-loop control system and method for diagnosing faults

Technical field

The present invention relates to a kind of Performance Evaluation and method for diagnosing faults of coke oven multi-loop control system.

Technical background

Along with the high speed development of steel and iron industry, coking industry also obtains very fast development, and the design of heating flame control system that is used to control coke oven temperature is more and more perfect.But because long-time running and shortage safeguard that effectively the control performance of coke oven heating flame control system constantly descends, thereby has reduced the economic benefit of coal chemical enterprise, has influenced the mass rate of production and the production cost of coke.

Coke oven is one of industrial furnace very important in the coal chemical industry, is the crucial production equipment of coke, also produces a large amount of coke-oven gas simultaneously, for other production runes provide fuel.But coke oven also is the energy consumption equipment of coke-oven plant's maximum simultaneously, and its coal gas major part of producing is fallen by autophage, so the quality of coke oven heating flame process control has important effect to the reduction of coke-oven plant's production cost and the raising of economic benefit.Existing coke oven heating flame control system generally has good performance at the initial stage of industrial process operation, if but there is not regular maintenance, As time goes on its performance can descend.And then cause problems such as product yield reduction, running cost increase.So need assess the performance of control system by certain methods, the relevant information of control system working condition is provided for the maintainer.Therefore, realize the Performance Evaluation and the fault diagnosis of coke oven heating flame control system are had great importance.

Summary of the invention

The objective of the invention is to propose a kind of Performance Evaluation and method for diagnosing faults of coke oven multi-loop control system, the Performance Evaluation of this coke oven multi-loop control system and method for diagnosing faults can reduce staff's labour intensity, improve fault diagnosis efficient.

Technical solution of the present invention is as follows:

A kind of performance estimating method of coke oven multi-loop control system is characterized in that, adopts fuzzy synthetic appraisement method to calculate the performance rate of coke oven heating flame control system; Described fuzzy synthetic appraisement method has 2 and estimates the factor: improve coefficient of variation and process capability index;

The improvement coefficient of variation is defined as: $V=\frac{\sqrt{\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{(Y_t-Y_{\mathrm{sp}})}^2}{N}}}{Y_{\mathrm{sp}}};$

Wherein, Y _tBe the fire path temperature detected value, Y _SpBe the fire path temperature setting value, V improves coefficient of variation, and N is the number of fire path temperature detected value;

The process capability index definition is: $C_p=\frac{T}{6\mathrm{\σ}};$

Wherein, C _pBe the process capability index of heating flame control system, σ is the population standard deviation of one group of fire path temperature detected value, and T is the technical tolerance amplitude of fire path temperature technological requirement;

Fuzzy grade is divided into high-quality, good, qualified, defective four grades, counts grade 1,2,3 and 4 respectively, adopts trapezoidal subordinate function to determine respectively to estimate the degree of membership of the factor, and fuzzy overall evaluation matrix R is;

$R=\left[\begin{array}{c}{R}_1\\ R_2\end{array}\right]=\left[\begin{array}{c}{\mathrm{\μ}}_{11}{\mathrm{\μ}}_{12}{\mathrm{\μ}}_{13}{\mathrm{\μ}}_{14}\\ {\mathrm{\μ}}_{21}{\mathrm{\μ}}_{22}{\mathrm{\μ}}_{23}{\mathrm{\μ}}_{24}\end{array}\right];$

Wherein, R ₁For improving the degree of membership set of coefficient of variation, R ₂Be the degree of membership set of process capability index, μ _IjBe i and estimate the degree that the factor is under the jurisdiction of j grade, the i value is 1 and 2, and the j value is 1～4;

The threshold value table that improves coefficient of variation and process capability index is as follows:

Improve the coefficient of variation threshold value table

Threshold value	??B a	??B b1	??B b2	??B b3	??B b4	??B c
							Improve coefficient of variation	??0.01	??0.02	??0.05	??0.08	??0.15	??0.2

Process capability index threshold table

Threshold value	??C a	??C b1	??C b2	??C b3	??C b4	??C c
							The process capability index	??0.43	??0.67	??1.0	??1.33	??1.67	??2.0

The evaluation result of coking production run performance is:

E=AoR=(e ₁, e ₂, e ₃, e ₄); Wherein A is a weight vector, A=(α ₁, α ₂), E is the fuzzy overall evaluation result vector, shows the subjection degree of the general status of this evaluation point coke oven heating flame control system to each grade, e ₁Expression coke oven heating flame control system is to the subjection degree of high-quality grade, e ₂Expression coke oven heating flame control system is to the subjection degree of good level, e ₃Expression coke oven heating flame control system is to the subjection degree of qualified grade, and e4 represents the subjection degree of coke oven heating flame control system to defective grade, judges the grade of coke oven heating flame control system performance afterwards according to maximum membership grade principle;

Wherein, operational symbol " o " expression multiplication;

α ₁, α ₂Be respectively that first estimates the factor and second weight of estimating the factor,

, i=1,2; λ _iBe i the degree of membership of estimating the factor, when being reference object with high-quality grade 1, λ _iComputing formula be

When being reference object with good level 2, λ _iComputing formula be

When being reference object with qualified grade 3, λ _iComputing formula be

When being reference object with defective class 4, λ _iComputing formula be λ _i=∨ μ _Ij, i=1,2; J=1, L, 4, wherein symbol ∨ represents to get the macrooperation symbol, in the Performance Evaluation of coke oven heating flame control system, selects good level 2 to calculate λ for reference object _i

Y _tSpan be 1200-1400, Y _SpSpan be 1260～1320, the span of N is 144～240; The numerical value of T is 40.

A kind of method for diagnosing faults of coke oven multi-loop control system, the method for diagnosing faults of this coke oven multi-loop control system are based on the assessment result of the performance estimating method of aforesaid coke oven multi-loop control system, and concrete diagnostic procedure is:

Step 1: set up the fault knowledge storehouse,, the fault of investigating out is associated together with improving coefficient of variation and process capability index, form a complete source of trouble information, this failure message is deposited in the fault knowledge storehouse when fire path temperature occurs when unusual;

Step 2: the failure message coupling, the output diagnostic result:

When the performance rate of heating flame control system when qualified and following, at three loops of coke oven heating flame control system: coke-oven gas flow control circuit, mixed gas pressure control loop and flue suction force control loop, calculate V respectively ₁, V ₂, V ₃, C _P1, C _P2, C _P3, V wherein ₁Be the improvement coefficient of variation of coke-oven gas flow control circuit, V _P1It is the process capability index of coke-oven gas flow control circuit; V ₂Be the improvement coefficient of variation of mixed gas pressure control loop, C _P2It is the process capability index of mixed gas pressure control loop; V ₃Be the improvement coefficient of variation of flue suction force control loop, C _P3It is the process capability index of flue suction force control loop;

Improvement coefficient of variation and process capability index with three control loops are that index is searched, if can find identical failure message the time, directly this failure message are exported as diagnostic result; In the time can't finding identical information, adopt failure message matching process based on Euclidean distance, calculate with the immediate failure message of index numerical value and export as equivalent information, the failure message that promptly satisfies following formula is an output information:

$\min =\left(\sqrt{\mathrm{\&lambda;}{\left(\frac{V_i-V_{i-j}}{V_{i\max }-V_{i\min }}\right)}^2+(1-\mathrm{\&lambda;}){\left(\frac{C_{\mathrm{pi}}-C_{\mathrm{pi}-j}}{C_{\mathrm{pi}\max }-C_{\mathrm{pi}\min }}\right)}^2}\right),i=\mathrm{1,2,3}j=\mathrm{1,2},\mathrm{<figure-callout\; id="1"\; label="LN"\; filenames="HDA0000021997700000013.png,HDA0000021997700000021.png"\; state="\{\{state\}\}">LN</figure-callout>}1,$

Wherein, V _iBe this improvement coefficient of variation that calculates, V _I-jBe the improvement coefficient of variation of the j bar failure message of existing i control loop in the fault information table, V _ImaxBe to comprise V _iMaximal value in interior improvement coefficient of variation, V _IminBe to comprise V _iMinimum value in interior improvement coefficient of variation, C _PiBe this process capability index that calculates, C _Pi-jBe the process capability index of the j bar failure message of existing i control loop in the fault information table, C _PimaxBe to comprise C _PiMaximal value in interior process capability index, C _PiminBe to comprise C _PiMinimum value in interior process capability index, N1 is the current quantity of failure message in the fault information table, and i represents three control loops, and λ is a weight coefficient, and the span of λ is 0.4-0.6.

Technical thought of the present invention is: at first, propose one based on the Performance Evaluation index of improving coefficient of variation; Then, the process capability index that calculates coke oven fire path temperature is as another Performance Evaluation index, with above two Performance Evaluation indexs as estimating the factor, grade be will blur and high-quality, good, qualified, defective four grades will be divided into, estimate the actual conditions of the factor according to each, adopt trapezoidal subordinate function to calculate the degree of membership of respectively estimating the factor, use the performance rate that maximum membership grade principle is judged coke oven heating flame control system; At last, many control loops are set up based on the fault knowledge storehouse of improving coefficient of variation and process capability index, deposit fault and corresponding improvement coefficient of variation and the process capability index investigated out in the fault knowledge storehouse according to sequence number, and in the process of control system operation real-time update fault knowledge storehouse, the performance rate of coke oven heating flame control system qualified and below in, to improve coefficient of variation and process capability index as index, principle according to Optimum Matching, search out all possible source of trouble, and these sources of trouble are embodied intuitively by textual form.

Beneficial effect:

Adopt coke oven heating flame process control Performance Evaluation of the present invention and method for diagnosing faults, make the performance evaluation and the fault diagnosis of control system become more directly perceived, automatic computing by computing machine, improved the automaticity of performance evaluation and fault diagnosis, need attention location system operation conditions constantly than the past staff, and at any time at the not good situation of operation, infer the fault that may exist according to self experience, this method has reduced staff's labour intensity greatly, has improved work efficiency.

Description of drawings

Fig. 1 is a control system Performance Evaluation structured flowchart;

Fig. 2 is Performance Evaluation and fault diagnosis structured flowchart;

Fig. 3 is for improving coefficient of variation membership function figure;

Fig. 4 is process capability index membership function figure.

Embodiment

Below with reference to figure and specific implementation process the present invention is described in further details.

Embodiment 1:

Coke oven heating flame process multi-loop control system is carried out Performance Evaluation and fault diagnosis, at first need to set up the index of assessment needs, the index of the source of trouble search when these indexs also are simultaneously the fault diagnosis, so the first step is set up improvement coefficient of variation and two indexs of process capability index exactly.

1, in order to embody the fitting degree between fire path temperature detected value and the setting value, definition improves coefficient of variation as performance index, and its formula is as follows:

$V=\frac{\sqrt{\frac{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{(Y_t-Y_{\mathrm{sp}})}^2}{N}}}{Y_{\mathrm{sp}}}---\left(1\right)$

Wherein, Y _tBe the fire path temperature detected value, Y _SpBe the fire path temperature setting value, V is based on Y _SpThe improvement coefficient of variation, N is the number of fire path temperature detected value.Y _tSpan be 1200-1400, Y _SpSpan be 1260-1320, the span of N is 144-240.Table 1 is a grading standard of improving coefficient of variation.

Table 1 improves coefficient of variation grading standard

Improve coefficient of variation	Less than 0.02	??0.02-0.08	??0.08-0.15	Greater than 0.15
					The grading standard	Ability is outstanding, should	Ability is good, shape	State is general	State difference, needs
	Keep (outstanding)	Attitude is stablized (well)	(qualified)	Rectify (defective)

Simultaneously, for the state of a control that embodies the heating flame process intuitively satisfies the ability of technical standard, calculate the process capability index of heating flame control system, its expression formula is as follows:

$C_p=\frac{T}{6\mathrm{\&sigma;}}---\left(2\right)$

Wherein, C _pBe the process capability index of heating flame control system, σ is the population standard deviation of one group of fire path temperature detected value, and its computation process is as follows: at first, calculate the mean value of one group of fire path temperature detected value; Then, calculate the quadratic sum of the difference of this group fire path temperature detected value and their mean value; At last, with the number of the quadratic sum that calculates divided by this group fire path temperature detected value, it promptly is σ that the result who obtains is asked arithmetic square root, and T is the technical tolerance amplitude of fire path temperature technological requirement, because the fluctuation range that coke oven fire path temperature requires is positive and negative 20 ℃, so the numerical value of T is 40.In coke oven fire path process capability index, T reflection be technical requirement to controlled variable, what σ reflected is the consistance of parametric procedure control, both are compared, just reflected that the parameter control procedure satisfies the degree of controlled variable technical requirement, this value is big more, and process capability is strong more, and table 2 is grading standards of process capability index.

Table 2 process capability index grading standard

The process capability index	??1.67-2	??1.33-1.67	??1-1.33	Less than 1
					The grading standard	Ability is outstanding, should keep (outstanding)	Ability good, in stable condition (well)	State general (qualified)	State difference, need to rectify (defective)

2, calculate after the improvement coefficient of variation and process capability index of fire path temperature, utilize the two, adopt fuzzy synthetic appraisement method, calculate the performance rate of coke oven heating flame control system, as shown in Figure 1.Fuzzy synthetic appraisement method is to determine its weight according to the sensitivity of the different evaluation factor, to realize the accurate Performance Evaluation to coke oven heating flame control system.With the improvement coefficient of variation of fire path temperature and process capability index as estimating the factor.Fuzzy grade is divided into high-quality, good, qualified, defective four grades, and grade 1 is the high-quality grade, and it is a highest ranking.According to each actual conditions of estimating the factor, adopt trapezoidal subordinate function to calculate the degree of membership of respectively estimating the factor.By calculating the degree of membership of each index, can obtain fuzzy overall evaluation matrix R.

$R=\left[\begin{array}{c}{R}_1\\ R_2\end{array}\right]=\left[\begin{array}{c}{\mathrm{\&mu;}}_{11}{\mathrm{\&mu;}}_{12}{\mathrm{\&mu;}}_{13}{\mathrm{\&mu;}}_{14}\\ {\mathrm{\&mu;}}_{21}{\mathrm{\&mu;}}_{22}{\mathrm{\&mu;}}_{23}{\mathrm{\&mu;}}_{24}\end{array}\right]---\left(3\right)$

Wherein, R ₁For improving the degree of membership set of coefficient of variation, R ₂Be the degree of membership set of process capability index, μ _IjBe i and estimate the degree that the factor is under the jurisdiction of j grade.

Estimate the threshold value of factor modulus gelatinization and estimate the grading standard and the expertise decision of the factor by each, table 3 is the threshold value tables that improve coefficient of variation, table 4 is threshold value tables of process capability index, the membership function that improves coefficient of variation as shown in Figure 3, the membership function of process capability index is as shown in Figure 4.

Table 3 improves the coefficient of variation threshold value table

Threshold value	??B a	??B b1	??B b2	??B b3	??B b4	??B c
							Improve coefficient of variation	??0.01	??0.02	??0.05	??0.08	??0.15	??0.2

Table 4 process capability index threshold table

Threshold value	??C a	??C b1	??C b2	??C b3	??C b4	??C c
							The process capability index	??0.43	??0.67	??1.0	??1.33	??1.67	??2.0

Utilize μ _IjCalculate λ _i, be reference object with good level 2, λ _iBe that i the evaluation factor belongs to good level and above degree of membership, i.e. λ ₁, λ ₂Be respectively that first is estimated factor and second and estimates the degree of membership that the factor is in good level and high-quality grade, its computing formula is as follows:

λ _iIt is poor more that the factor is estimated in i of more little expression, and its corresponding weights also should be big more, therefore defines the weight of respectively estimating the factor to be:

${\mathrm{\&alpha;}}_i=\frac{1}{{\mathrm{\&lambda;}}_i}/\underset{k=1}{\overset{2}{\mathrm{\&Sigma;}}}\frac{1}{{\mathrm{\&lambda;}}_k},(i=\mathrm{1,2})---\left(5\right)$

α ₁, α ₂Be respectively that first estimates the factor and second weight of estimating the factor.

The evaluation result of then coking production run performance is:

E＝AoR＝(e ₁，e ₂，e ₃，e ₄)????(6)

Wherein, A is a weight vector, A=(α ₁, α ₂), E is the fuzzy overall evaluation result vector, shows the subjection degree of the general status of this evaluation point coke oven heating flame control system to each grade fuzzy subset, e ₁Expression coke oven heating flame control system is to the subjection degree of high-quality grade, e ₂Expression coke oven heating flame control system is to the subjection degree of good level, e ₃Expression coke oven heating flame control system is to the subjection degree of qualified grade, e ₄Expression coke oven heating flame control system is judged the grade of coke oven heating flame control system performance afterwards to the subjection degree of defective grade according to maximum membership grade principle.

For example, calculate the improvement coefficient of variation V=0.064 of one group of fire path temperature detected value, C _p=1.14, can calculate according to the membership function figure and the process capability index membership function figure that improve coefficient of variation

Calculate λ according to formula (4) _i=0.53, λ ₂=0.42, with λ _i, λ ₂Bring formula (5) into and calculate α ₁=0.44, α ₂=0.56, weight vector A=(0.44,0.56) calculates according to formula (6)

According to the principle that maximum is subordinate to, the performance rate of coke oven heating flame control system is qualified.

3, calculate the performance rate of coke oven heating flame control system after, when the performance rate of heating flame control system qualified and below in, need diagnose causing the not good failure cause of performance, as shown in Figure 2.Coke oven heating flame control system is made up of three control loops, and they are: coke-oven gas flow control circuit, mixed gas pressure control loop, flue suction force control loop.The ruuning situation of three control loops is the immediate causes that cause fire path temperature unusual, therefore when fire path temperature is unusual, at these three control loops, calculates V respectively ₁, V ₂, V ₃, C _P1, C _P2, C _P3, V wherein ₁Be the improvement coefficient of variation of coke-oven gas flow control circuit, C _P1It is the process capability index of coke-oven gas flow control circuit; V ₂Be the improvement coefficient of variation of mixed gas pressure control loop, C _P2It is the process capability index of mixed gas pressure control loop; V ₃Be the improvement coefficient of variation of flue suction force control loop, C _P3It is the process capability index of flue suction force control loop.

4, set up the fault knowledge storehouse,, the fault of investigating out is associated together with improving coefficient of variation and process capability index, form a complete source of trouble information, this failure message is deposited in the fault knowledge storehouse when fire path temperature occurs when unusual.Shown in the following tabulation lattice of the form that deposits in:

Table 5 coke-oven gas flow control circuit failure message table

Table 6 mixed gas pressure control loop failure message table

Table 7 flue suction force control loop failure message table

Table 5, table 6, table 7 have been enumerated the failure message form that three control loops deposit the fault knowledge storehouse in.In the process of control system operation, at various operation exception situations, bring in constant renewal in knowledge base with new failure message, the failure message of knowledge base is enriched constantly.

When searching failure message, improvement coefficient of variation and process capability index with three control loops are that index is searched, in the time can't finding identical information, employing is based on the failure message matching process of Euclidean distance, calculate with the immediate failure message of index numerical value and export as equivalent information, consider the situation that the order of magnitude of two entry index numerical value there are differences, adopt method for normalizing that the numerical value of two entry indexs is transformed into [0,1] between, as follows through the failure message matching process computing formula based on Euclidean distance of normalized:

$\min =\left(\sqrt{\mathrm{\&lambda;}{\left(\frac{V_i-V_{i-j}}{V_{i\max }-V_{i\min }}\right)}^2+(1-\mathrm{\&lambda;}){\left(\frac{C_{\mathrm{pi}}-C_{\mathrm{pi}-j}}{C_{\mathrm{pi}\max }-C_{\mathrm{pi}\min }}\right)}^2}\right),i=\mathrm{1,2,3}j=\mathrm{1,2},\mathrm{LN}---\left(7\right)$

Wherein, V _iBe this improvement coefficient of variation that calculates, V _I-jBe existing improvement coefficient of variation in the fault information table, V _ImaxBe to comprise V _iMaximal value in interior improvement coefficient of variation, V _IminBe to comprise V _iMinimum value in interior improvement coefficient of variation, C _PiBe this process capability index that calculates, C _Pi-jBe existing process capability index in the fault information table, C _PimaxBe to comprise C _PiMaximal value in interior process capability index, C _PiminBe to comprise C _PiMinimum value in interior process capability index, N is the current quantity of failure message in the fault information table, and i represents three control loops, and λ is a weight coefficient, and the span of λ is 0.4-0.6.

Be example with coke-oven gas flow control circuit, mixed gas pressure control loop, flue suction force control loop respectively, carry out the failure message coupling according to the method described above.

For the coke-oven gas flow control circuit, calculate V ₁=0.042, C _P1=1.250, wherein, N=100, i=1, λ=0.5, V _1max=0.714, V _1min=0.012, C _P1max=1.86, C _P1min=0.44 substitution formula (7) is calculated as follows:

$\min \left(\sqrt{0.5{\left(\frac{0.042-V_{1-j}}{0.714-0.012}\right)}^2+0.5{\left(\frac{1.350-C_{p1-j}}{1.86-0.44}\right)}^2}\right)=0.054$

In the formula, V _1-jBe improvement coefficient of variation, the C of j bar failure message in the coke-oven gas flow control circuit failure message table _P1-jIt is the process capability index of j bar failure message in the coke-oven gas flow control circuit failure message table.The failure message corresponding with 0.054 is the 2nd failure message, and its index is V _1-2=0.063, C _P1-2=1.1, failure message is " flow-control module existing problems ".

For the mixed gas pressure control loop, calculate V ₂=0.501, C _P2=0.7, wherein, N=50, i=2, λ=0.4, V _2max=0.205, V _2min=0.014, C _P2max=1.92, C _P2min=0.86, bring formula (7) into and be calculated as follows:

$\min \left(\sqrt{0.4{\left(\frac{0.501-V_{2-\mathrm{<figure-callout\; id="0"\; label="j"\; filenames="HDA0000021997700000013.png,HDA0000021997700000021.png"\; state="\{\{state\}\}">j</figure-callout>}}}{0.205-0.014}\right)}^2+0.6{\left(\frac{0.7-C_{p2-\mathrm{<figure-callout\; id="1"\; label="j"\; filenames="HDA0000021997700000013.png,HDA0000021997700000021.png"\; state="\{\{state\}\}">j</figure-callout>}}}{1.92-0.86}\right)}^2}\right)=0.093$

Wherein, V _2-jBe improvement coefficient of variation, the C of j bar failure message in the mixed gas pressure control loop failure message table _P2-jIt is the process capability index of j bar failure message in the mixed gas pressure control loop failure message table.The failure message corresponding with 0.093 is the 9th failure message, and its index is V _2-9=0.473, C _P2-9=0.72, failure message is " there is fault in pressure transducer ".

For the flue suction force control loop, calculate V ₃=0.247, C _P3=1.28, wherein, N=60, i=3, λ=0.45, V _3max=0.82, V _3min=0.02, C _P3max=1.72, C _P3min=0.37, bring formula (7) into and be calculated as follows:

$\min \left(\sqrt{0.45{\left(\frac{0.247-V_{3-\mathrm{<figure-callout\; id="0"\; label="j"\; filenames="HDA0000021997700000013.png,HDA0000021997700000021.png"\; state="\{\{state\}\}">j</figure-callout>}}}{0.82-0.02}\right)}^2+0.55{\left(\frac{1.28-C_{p3-\mathrm{<figure-callout\; id="1"\; label="j"\; filenames="HDA0000021997700000013.png,HDA0000021997700000021.png"\; state="\{\{state\}\}">j</figure-callout>}}}{1.72-0.37}\right)}^2}\right)=0.015$

Wherein, V _3-jBe improvement coefficient of variation, the C of j bar failure message in the flue suction force control loop failure message table _P3-jIt is the process capability index of j bar failure message in the flue suction force control loop failure message table.The failure message corresponding with 0.015 is the 26th failure message, and its index is V _3-26=0.23, C _P3-26=1.29, failure message is " valve linearization block existing problems ".

This method all is suitable for for various forms of multiloop control systems, can demonstrate current possible failure message intuitively, accurately, easily, for the reference of technologist as malfunction elimination.

Claims (3)

1. the performance estimating method of a coke oven multi-loop control system is characterized in that, adopts fuzzy synthetic appraisement method to calculate the performance rate of coke oven heating flame control system; Described fuzzy synthetic appraisement method has 2 and estimates the factor: improve coefficient of variation and process capability index;

Wherein, improving coefficient of variation is defined as:

Wherein, Y _tBe the fire path temperature detected value, Y _SpBe the fire path temperature setting value, V improves coefficient of variation, and N is the number of fire path temperature detected value;

The process capability index definition is:

Wherein, C _pBe the process capability index of heating flame control system, σ is the population standard deviation of one group of fire path temperature detected value, and T is the technical tolerance amplitude of fire path temperature technological requirement;

Fuzzy grade is divided into high-quality, good, qualified, defective four grades, counts grade 1,2,3 and 4 respectively, adopts trapezoidal subordinate function to determine respectively to estimate the degree of membership of the factor, and fuzzy overall evaluation matrix R is;

$R=\left[\begin{array}{c}{R}_1\\ R_2\end{array}\right]=\left[\begin{array}{cccc}{\mathrm{\&mu;}}_{11}& {\mathrm{\&mu;}}_{12}& {\mathrm{\&mu;}}_{13}& {\mathrm{\&mu;}}_{14}\\ {\mathrm{\&mu;}}_{21}& {\mathrm{\&mu;}}_{22}& {\mathrm{\&mu;}}_{23}& {\mathrm{\&mu;}}_{24}\end{array}\right];$

Wherein, R ₁For improving the degree of membership set of coefficient of variation, R ₂Be the degree of membership set of process capability index, μ _IjBe i and estimate the degree that the factor is under the jurisdiction of j grade, the i value is 1 and 2, and the j value is 1～4;

The threshold value table that improves coefficient of variation and process capability index is as follows:

Improve the coefficient of variation threshold value table

Threshold value ??B _a ??B _b1 ??B _b2 ??B _b3 ??B _b4 ??B _c Improve coefficient of variation ??0.01 ??0.02 ??0.05 ??0.08 ??0.15 ??0.2

Process capability index threshold table

Threshold value ??C _a ??C _b1 ??C _b2 ??C _b3 ??C _b4 ??C _c The process capability index ??0.43 ??0.67 ??1.0 ??1.33 ??1.67 ??2.0

The evaluation result of coking production run performance is:

E=AoR=(e ₁, e ₂, e ₃, e ₄); Wherein A is a weight vector, A=(α ₁, α ₂), E is the fuzzy overall evaluation result vector, shows the subjection degree of the general status of this evaluation point coke oven heating flame control system to each grade, e ₁Expression coke oven heating flame control system is to the subjection degree of high-quality grade, e ₂Expression coke oven heating flame control system is to the subjection degree of good level, e ₃Expression coke oven heating flame control system is to the subjection degree of qualified grade, e ₄Expression coke oven heating flame control system is judged the grade of coke oven heating flame control system performance afterwards to the subjection degree of defective grade according to maximum membership grade principle;

Wherein, operational symbol " o " expression multiplication;

α ₁, α ₂Be respectively that first estimates the factor and second weight of estimating the factor,

${\mathrm{\&alpha;}}_i=\frac{1}{{\mathrm{\&lambda;}}_i}/\underset{k=1}{\overset{2}{\mathrm{\&Sigma;}}}\frac{1}{{\mathrm{\&lambda;}}_k},i=\mathrm{1,2};$

λ _iBe i the degree of membership of estimating the factor, when being reference object with high-quality grade 1, λ _iComputing formula be

When being reference object with good level 2, λ _iComputing formula be

When being reference object with qualified grade 3, λ _iComputing formula be

When being reference object with defective class 4,

λ _iComputing formula be λ _i=∨ μ _Ij, i=1,2, j=1, L, 4, wherein symbol ∨ represents to get the macrooperation symbol.

2. the performance estimating method of coke oven multi-loop control system according to claim 1 is characterized in that, Y _tSpan be 1200-1400, Y _SpSpan be 1260～1320, the span of N is 144～240; The numerical value of T is 40.

3. the method for diagnosing faults of a coke oven multi-loop control system, it is characterized in that, the method for diagnosing faults of this coke oven multi-loop control system is based on the assessment result of the performance estimating method of the described coke oven multi-loop control system of claim 2, and concrete diagnostic procedure is:

Step 1: set up the fault knowledge storehouse,, the fault of investigating out is associated together with improving coefficient of variation and process capability index, form a complete source of trouble information, this failure message is deposited in the fault knowledge storehouse when fire path temperature occurs when unusual;

Step 2: the failure message coupling, the output diagnostic result:

When the performance rate of heating flame control system when qualified and following, at three loops of coke oven heating flame control system: coke-oven gas flow control circuit, mixed gas pressure control loop and flue suction force control loop, calculate V respectively ₁, V ₂, V ₃, C _P1, C _P2, C _P3, V wherein ₁Be the improvement coefficient of variation of coke-oven gas flow control circuit, C _P1It is the process capability index of coke-oven gas flow control circuit; V ₂Be the improvement coefficient of variation of mixed gas pressure control loop, C _P2It is the process capability index of mixed gas pressure control loop; V ₃Be the improvement coefficient of variation of flue suction force control loop, C _P3It is the process capability index of flue suction force control loop;

Improvement coefficient of variation and process capability index with three control loops are that index is searched, if can find identical failure message the time, directly this failure message are exported as diagnostic result; In the time can't finding identical information, adopt failure message matching process based on Euclidean distance, calculate with the immediate failure message of index numerical value and export as equivalent information, the failure message that promptly satisfies following formula is an output information:

$\min \left(\sqrt{\mathrm{\&lambda;}{\left(\frac{V_i-V_{i-j}}{V_{i\max }-V_{i\min }}\right)}^2+(1-\mathrm{\&lambda;}){\left(\frac{C_{\mathrm{pi}}-C_{\mathrm{pi}-j}}{C_{\mathrm{pi}\max }-C_{\mathrm{pi}\min }}\right)}^2}\right),i=\mathrm{1,2,3}j=\mathrm{1,2},\mathrm{L\; N}1,$

Wherein, V _iBe this improvement coefficient of variation that calculates, V _I-jBe the improvement coefficient of variation of the j bar failure message of existing i control loop in the fault information table, V _ImaxBe to comprise V _iMaximal value in interior improvement coefficient of variation, V _IminBe to comprise V _iMinimum value in interior improvement coefficient of variation, C _PiBe this process capability index that calculates, C _Pi-jBe the process capability index of the j bar failure message of existing i control loop in the fault information table, C _PimaxBe to comprise C _PiMaximal value in interior process capability index, C _PiminBe to comprise C _PiMinimum value in interior process capability index, N1 is the current quantity of failure message in the fault information table, and i represents three control loops, and λ is a weight coefficient, and the span of λ is 0.4-0.6.

Images