A method for predicting the operating status of a fan in real time and online based on data mining

Abstract

The present invention discloses a method for real-time online prediction of the operating status of a fan based on data mining. The method includes the steps of: obtaining historical operating status parameters of a unit and a fan; extracting data in a number of normal and stable operating modes to build a sample collection; obtaining a prediction model of the fluctuation relationship of the fan inlet flow rate with the load of the unit or the flow rate of the main steam; obtaining a prediction model of the fluctuation relationship of the flue gas volume with the resistance of the system from the furnace room to the inlet stage of the induced draft fan and the resistance of the system from the outlet of the induced draft fan to the chimney outlet stage; obtaining the status operating parameters of the fan based on the two prediction models; calculating the theoretical stall safety factor, pressure margin factor and flow margin factor of the fan; and evaluating whether the operating status of the fan is safe after the fluctuation of the status parameters of the flue gas system by comparing the deviation between the theoretical stall safety factor, pressure margin factor and flow margin factor and a threshold value. The present invention can improve the rationality of the fan stall prevention control strategy and improve the operating safety and economy of the fan.

Description

Technical Field

[0001] The present invention relates to an axial flow fan (including a static vane adjustable axial flow fan, a dynamic vane adjustable axial flow fan, etc.) used in an exhaust gas system of a coal-fired power plant, and particularly relates to a method for predicting the operating status of the fan in real time and online based on data mining.

Background Art

[0002] At present, as policies such as deep adjustment of flexibility are deeply implemented, the fans in power plants are facing the requirement of frequent regulation of wide load. However, after implementing the ultra-low emission transformation of the unit, the requirements of environmental protection indicators are becoming increasingly high, and there are more and more environmental protection facilities in the exhaust gas system. In actual operation, due to the improvement of environmental protection requirements, the ammonia emission amount in the exhaust gas system increases, and the abnormal blockage phenomenon of each equipment in the exhaust gas system formed by the presence of ammonium sulfate becomes increasingly serious. The operating parameters of the exhaust gas system deviate from its design parameters, and the stall of the induced draft fan in the high-load operation mode and the phenomenon of the reduction of the unit's load capacity occur frequently. For the high-load stall phenomenon of the induced draft fan, usually, the operators in power plants use the method of limiting the opening degree and current of the fan to prevent the stall of the fan. However, since the blockage situation of the exhaust gas system and the operating situation of the induced draft fan cannot be accurately predicted, the method of preventing the stall of the fan by limiting the opening degree and current of the fan is too conservative. After maintaining the exhaust gas system to remove the blockage, the resistance of the exhaust gas system has decreased significantly. In addition, due to the fan stall prevention operation strategy, the load capacity of the unit is limited. The conventional fan monitoring system in power plants can monitor the state parameters such as the pressure, flow rate, current and inlet temperature at the inlet and outlet of the fan in real time, and can evaluate the real-time performance state of the fan in real time. Therefore, based on the data mining of the past state parameters of the power plant fan, an online prediction method for the operating state of the power plant fan is established to accurately evaluate the operating situation of the fan after the blockage of the exhaust gas system, and to provide a basis for formulating the fan stall prevention operation strategy and adjusting the operation of the fan. In order to provide a basis for formulating technical transformation solutions related to the exhaust gas system for the change of the state parameters of the exhaust gas system after adding new equipment to the exhaust gas system, it is necessary to accurately predict the output state of the fan.

Summary of the Invention

Problems to be Solved by the Invention

[0003] The present invention proposes a method for online predicting the fan operation status in real time based on data mining in order to accurately predict the fan operation status after the operation status of the exhaust gas system fluctuates.

[0004] The present invention is realized by the following technical solutions.

[0005] The method for online predicting the fan operation status in real time based on data mining includes: 1) obtaining the past operation status parameters of the unit and the fan based on the fan online monitoring system and the DCS system; 2) using data analysis technology to process the past operation status parameters of the fan, extracting data in a plurality of normal and stable operation modes, and constructing a sample set; 3) training the extracted sample set based on an artificial neural network to obtain a prediction model of the variation relationship of the fan inlet flow rate with respect to the unit load or the main steam flow rate, and a prediction model of the resistance of the system from the furnace chamber to the induced fan inlet section and the resistance of the system from the induced fan outlet to the chimney outlet section according to the variation relationship of the exhaust gas volume; 4) giving a predicted value of the variation of the status parameters of the exhaust gas system, and obtaining the status operation parameters of the fan based on the two prediction models; 5) calculating the theoretical stall safety factor, pressure margin coefficient, and flow margin coefficient of the fan based on the fan performance curve and the predicted values of the fan operation parameters; 6) setting thresholds for each stall margin coefficient based on past stall operation mode analysis and a large number of stall test statistical analyses, and comparing the deviations between the theoretical stall safety factor, pressure margin coefficient, and flow margin coefficient and the thresholds to evaluate whether the fan operation status after the variation of the status parameters of the exhaust gas system is safe.

[0006] A further improvement of the present invention is that in step 1), the past operation status parameters of the unit and the fan are Unit load L, boiler evaporation D b , fan inlet temperature T in , fan inlet volumetric flow rate Q v , fan inlet total pressure P t、in , fan outlet total pressure P t、out , fan opening β, fan inlet static pressure P e、in and fan outlet static pressure P e、out are included, the time period t is 10 to 30 days, and the time interval is such that △t = 1 to 5 minutes.

[0007] JPEG0007705458000001.jpg68159

[0008] JPEG0007705458000002.jpg21159

[0009] A further improvement of the present invention is that in step 3), based on the sample points, training is performed respectively using an artificial neural network, and the variation relationship models Q v of the fan inlet volumetric flow rate Q b and the main steam flow rate D v = f(D b ), the variation relationship model P e、in of the fan inlet static pressure P v and the fan inlet volumetric flow rate Q e、in = f(Q v ), the variation relationship model P e、out of the fan outlet static pressure P v and the fan inlet volumetric flow rate Q e、out = f(Q v ), the variation relationship model P t、in of the fan inlet total pressure P v and the fan inlet volumetric flow rate Q t、in = f(Q v ) and the variation relationship model P t、out of the fan outlet total pressure P v and the fan inlet volumetric flow rate Q t、out = f(Q v ) are to be obtained.

[0010] A further improvement of the present invention is that in step 3), the situation variation of the exhaust gas system is predicted, and the variation value corresponding to the resistance before and after the induced draft fan in the operation mode, that is, (D' b , ΔP in ), (D' b , ΔP out ) is obtained.

[0011] A further improvement of the present invention is that in step 4), the fan's situation operation parameters are Based on the basic design parameters of the boiler and the actual operation mode of the unit, the boiler evaporation amount interval [D b、min , D b、BMCR within the width load adjustment range of the unit is determined, and in the said interval, m typical operation modes are selected, and the set of boiler evaporation amounts {D b、i} is obtained. Based on each parameter relationship model, the set of situation parameter points in each operation mode, that is, {(D b , T in , Q v , P e、in , P e、out , P t、in , P t、out ,...) i} is calculated and obtained, where i = 1, 2, 3,..m and m >= 3.

[0012] JPEG0007705458000003.jpg59165

[0013] Based on the variation of the fan's inlet flue gas temperature and static pressure, the fan's inlet volume flow rate is corrected. Based on the predicted values of the fan's situation parameters in each typical operation mode, the fan's total pressure Pt and specific pressure energy Y are calculated and obtained. The fan's operation prediction parameters are {(Q' v , P' t , Y') i}.

[0014] A further improvement of the present invention is that in step 5), the fan's operation prediction point parameters {(Q' v , P' t , Y')i Based on (i = 1, 2, 3,... m), it is displayed on the performance curve of the fan, and the opening β corresponding to the fan operating point of each operating mode i is obtained, and the intersection point {(Q v、s , P t、s ) i} of the constant opening line and the theoretical stall line is identified and obtained.

[0015] JPEG0007705458000004.jpg66156

[0016] JPEG0007705458000005.jpg39159

[0017] For each fan, when it is predicted that the operating point parameters do not meet the above conditions, it is explained that the fan cannot meet the full load range conditions even if the resistance of the exhaust gas system increases, and it is necessary to perform stall prevention control of the fan. By reducing the increase values ΔP in and ΔP out , re-evaluate until the operating situation parameters that meet the conditions are obtained. Furthermore, by monitoring the actual static pressure difference ΔP e between the inlet and outlet of the fan and the opening β0 of the fan, a stall prevention control strategy for the fan is formulated, that is, ΔP e < (P e , out + ΔP out ) BMCR - (P e、in - ΔP in ) BMCR、 β0 < 0.8β max is satisfied. Effect

[0018] The present invention has at least the following beneficial technical effects. The present invention provides a method for online predicting the operating status of a fan in real time based on data mining. By mining and utilizing the past monitoring data of the fan, a prediction model for each situation parameter can be constructed, and the main operating status parameters of the fan in typical operating modes can be obtained. Further, by predicting the fluctuation value of the resistance of the exhaust gas system, the operating status parameters of the fan after the resistance of the exhaust gas system fluctuates can be accurately predicted. Based on the design performance curve of the fan, a stall safety system, a stall pressure margin coefficient, and a stall flow margin coefficient corresponding to the operating status parameters of the fan in multiple operating modes are calculated. Based on the stall margin coefficient and the opening degree of the fan, the operating safety of the fan is evaluated, whether the fan can operate safely and stably is determined, a basis for the operation adjustment of the fan is provided, the rationality of the stall prevention control strategy of the fan is improved, and the operating safety and economy of the fan can be improved.

Brief Description of the Drawings

[0019]

Figure 1

Figure 2

Figure 3

Embodiments for Carrying Out the Invention

[0020] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure should not be limited by the embodiments described herein and can be implemented in various forms. In contrast, these embodiments are provided to enable a more complete understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other when they do not conflict. Hereinafter, the present invention will be described in detail with reference to the accompanying drawings in combination with the embodiments.

[0021] The present invention provides a method for online predicting the fan operation status in real time based on data mining. The method includes: 1) obtaining the past operation status parameters of the unit and the fan based on the online monitoring system and DCS system of the fan; 2) using data analysis technology to process the past operation status parameters of the fan, extracting data in a plurality of normal and stable operation modes, and constructing a sample set; 3) training the extracted sample set based on an artificial neural network to obtain a prediction model for the variation relationship between the fan inlet flow rate and the unit load or the main steam flow rate, and a prediction model for the resistance of the system from the furnace chamber to the induced draft fan inlet stage and the resistance of the system from the induced draft fan outlet to the chimney outlet stage based on the variation relationship of the exhaust gas volume; 4) giving the predicted value of the variation of the status parameters of the exhaust gas system, and obtaining the status operation parameters of the fan based on the two prediction models; 5) calculating the theoretical stall safety factor, pressure margin coefficient, and flow margin coefficient of the fan based on the performance curve of the fan and the predicted values of the operation parameters of the fan; 6) setting the threshold values of each stall margin coefficient based on the analysis of past stall operation modes and the statistical analysis of a large number of stall tests, and comparing the deviations between the theoretical stall safety factor, pressure margin coefficient, and flow margin coefficient and the threshold values to evaluate whether the operation status of the fan after the variation of the status parameters of the exhaust gas system is safe. The specific implementation method of the present invention is as follows.

[0022] 1. Extract the most recent boiler and fan main monitoring parameter data sets, including parameters such as the unit load L, boiler evaporation D b , fan inlet temperature T in , fan inlet volumetric flow rate Q v , fan inlet total pressure P t、in , fan outlet total pressure P t、out , fan opening β, fan inlet static pressure P e、in and fan outlet static pressure P e、out from the power plant fan online monitoring system and DCS system. The time period t is 10 - 30 days, the time interval is Δt = 1 - 5 minutes, and the data set needs to cover most of the normal operating load range.

[0023] 2. Organize the data for the past, submit the main monitoring parameters in multiple stable load modes, and construct a sample set.

[0024] JPEG0007705458000006.jpg34159

[0025] JPEG0007705458000007.jpg21159

[0026] 3. Based on the sample point set, use artificial neural networks to train respectively, and obtain the variation relationship models of the fan inlet volumetric flow rate Q v and the main steam flow rate D b : Q v = f(D b ), the variation relationship model of the fan inlet static pressure P e、in and the fan inlet volumetric flow rate Q v : P e、in = f(Q v ), the variation relationship model of the fan outlet static pressure P e、out and the fan inlet volumetric flow rate Q v : P e、out = f(Q v ), the variation relationship model of the fan inlet total pressure P t、in and the fan inlet volumetric flow rate Q v : P t、in = f(Q vand the total outlet pressure P of the fan t、out and the inlet volume flow rate Q of the fan v to obtain the variation relationship model P t、out = f(Q v ).

[0027] 4. Combine the actual operating conditions of the unit and the planned retrofit solutions to predict the variation of the exhaust gas system situation, and obtain the variation values of the resistance before and after the induced draft fan in the operating mode, that is, (D' b , ΔP in ), (D' b , ΔP out ).

[0028] JPEG0007705458000008.jpg155161

[0029] JPEG0007705458000009.jpg108159

[0030] JPEG0007705458000010.jpg46159

[0031] If it is predicted that the operating point parameters of each fan do not meet the above conditions, it means that even if the resistance of the exhaust gas system increases, the fan cannot meet the full load range conditions, and it is necessary to perform stall prevention control on the fan. By reducing the increase values ΔP in and ΔP out , re-evaluate until the operating condition parameters that meet the conditions are obtained. Furthermore, by monitoring the actual static pressure difference ΔP e between the inlet and outlet of the fan and the opening degree β0 of the fan, formulate a stall prevention control strategy for the fan, that is, ΔP e < (P e , out + ΔP out ) BMCR - (P e、in - ΔP in ) BMCR、 β0 < 0.8β max is satisfied.

[0032] 5. This method is characterized by high efficiency, high reliability, and strong robustness, and is applicable to the stall prevention monitoring of the induced draft fan of a large coal-fired unit and the prediction of the fan's operating state.

[0033] Embodiment The induced draft fan of a 300MW unit in China is a dynamic vane adjustable axial flow fan. Each environmental protection facility in the exhaust gas system is clogged to varying degrees due to the presence of ammonium bisulfate, which causes stall in the high-load operation mode. Therefore, in order to ensure the safe and stable operation of the unit, the power plant adopts an adjustment method to limit the opening and current, restricting the load capacity of the unit. Also, after the major overhaul and clearance of the unit's congestion, since there is no reliable basis for evaluation, the unit's output is restricted by operating the unit with the established fan adjustment method. The method of the present invention is realized using an object-oriented programming language. By incorporating this functional module into the online monitoring and fault warning system of the induced draft fan, the operating state of the fan is estimated in real-time according to the resistance fluctuation situation of the exhaust gas system, the safety of the operation is evaluated, and early warnings are given. Based on this method, a more reliable fan adjustment method is formulated. By comprehensively monitoring the changes in the resistance of the mainly congestion-prone facilities in the exhaust gas system, the static pressure difference between the inlet and outlet of the fan, the opening and current of the fan, etc., a reliable basis for fan adjustment is provided, improving the output of the fan and the load capacity of the unit. The algorithm of the present invention is efficient and highly reliable. Under typical operating modes, the distribution of the actual fan operating points and the predicted operating points on its performance curve is shown in Figure 3, and the state evaluation calculation results of each operating point are shown in Table 1.

[0034] JPEG0007705458000011.jpg243165JPEG0007705458000012.jpg246165JPEG0007705458000013.jpg21165

[0035] As described above, the present invention has been described in detail using general descriptions and specific embodiments. However, it is obvious to those skilled in the art that some modifications or improvements can be made based on the present invention. Therefore, any of these modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Explanation of Signs

[0036] Power generation load of the L unit, unit: kW D b Evaporation capacity of the boiler, unit: t / h T in Inlet flue gas temperature of the fan, unit: °C Q v Inlet volume flow rate of the fan, unit: m 3 / s P t、in Inlet total pressure of the fan, unit: Pa P t、out Outlet total pressure of the fan, unit: Pa P e、in Inlet static pressure of the fan, unit: Pa P e、out Outlet total pressure of the fan, unit: Pa P t Total pressure of the fan, unit: Pa Y Fan specific pressure energy, unit: kJ / kg β Feedback value of the DCS fan opening degree, unit: ° l Data point order (l = 1, 2,..., n) j Time interval (j = 1, 2,..., n) I Load mode number (i = 1, 2,..., m) S Standard deviation k s Stall safety factor Kp Stall pressure margin coefficient kq Stall flow margin coefficient β min Minimum opening degree value of the fan, unit: ° β max Maximum opening degree value of the fan, unit: ° β BMCRThe opening value corresponding to the operating point of the fan in the BMCR operating mode, unit: °

Claims

1. A method for predicting the operating status of a fan in real time and online based on data mining, comprising: 1) obtaining the past operating status parameters of the unit and the fan based on the online monitoring system and DCS system of the fan; 2) using data analysis technology to process the past operating status parameters of the fan, extracting data in a plurality of normal and stable operating modes, and constructing a sample set; 3) training the extracted sample set based on an artificial neural network to obtain a prediction model of the variation relationship between the fan inlet flow rate and the unit load or the main steam flow rate, and a prediction model of the resistance of the system from the furnace chamber to the induced fan inlet section and the resistance of the system from the induced fan outlet to the chimney outlet section based on the variation relationship of the exhaust gas volume; 4) giving a predicted value of the variation of the status parameters of the exhaust gas system, and obtaining the status operating parameters of the fan based on the two prediction models; 5) calculating the theoretical stall safety factor, pressure margin factor and flow margin factor of the fan based on the performance curve of the fan and the predicted values of the operating parameters of the fan; 6) setting the threshold values of the theoretical stall safety factor, pressure margin factor and flow margin factor based on the analysis of past stall operating modes and a large number of stall test statistical analyses, and comparing the deviations between the theoretical stall safety factor, pressure margin factor and flow margin factor and the threshold values to evaluate whether the operating status of the fan after the variation of the status parameters of the exhaust gas system is safe. A method for predicting the operating status of a fan in real time and online based on data mining, characterized by including the above steps.

2. In step 1), the past operating status parameters of the unit and the fan are Unit load L, boiler evaporation capacity D ｂ , fan inlet temperature T ｉｎ , fan inlet volumetric flow rate Q ｖ , fan inlet total pressure P ｔ、ｉｎ , fan outlet total pressure P ｔ、ｏｕｔ , fan opening β, fan inlet static pressure P ｅ、ｉｎ and fan outlet static pressure P ｅ、ｏｕｔ A method for predicting the operating status of a fan in real time and online based on data mining according to claim 1, comprising the above, wherein the time period t is 10 to 30 days and the time interval is Δt = 1 to 5 minutes.

3. In step 2), when constructing the sample set, specifically, Evaporation rate D of the boiler ｂ Based on the set of data points of time t and D, i.e., {(t ｊ , D ｂ、ｊ ),} the data is screened at a time interval of Δt = 2 h, and the set of intervals {(t ｉ )} in which all data points {D ｊ , t ｊ + 2 h)} satisfy the following conditions are selected. where j = 1, 2,..., m and i = 1, 2,..., k, In the selected data set, all data points in the time range { (t ｊ + 0.5 h, t ｊ + 1.5 h)} are submitted, the average value is calculated, and the set of sample points: A method for predicting the operating status of a fan in real time and online based on data mining according to claim 2, characterized by obtaining the above.

4. In step 3), based on the sample points, each is trained using an artificial neural network, and the inlet volume flow rate Q of the fan ｖ and the flow rate D of the main steam ｂ variation relationship model Q ｖ = f(D ｂ ), the inlet static pressure P of the fan ｅ、ｉｎ and the inlet volume flow rate Q of the fan ｖ variation relationship model P ｅ、ｉｎ = f(Q ｖ ), the outlet static pressure P of the fan ｅ、ｏｕｔ and the inlet volume flow rate Q of the fan ｖ variation relationship model P ｅ、ｏｕｔ = f(Q ｖ ), the inlet total pressure P of the fan ｔ、ｉｎ and the inlet volume flow rate Q of the fan ｖ variation relationship model P ｔ、ｉｎ = f(Q ｖ ) and the outlet total pressure P of the fan ｔ、ｏｕｔ and the inlet volume flow rate Q of the fan ｖ variation relationship model P ｔ、ｏｕｔ = f(Q ｖ ) are obtained. A method for predicting the operating status of a fan in real time and online based on the data mining according to claim 3.

5. In step 3), predict the situation change of the exhaust gas system, and obtain the variation value corresponding to the resistance before and after the induced draft fan in the operation mode, that is, (D' ｂ , ΔP ｉｎ ), (D' ｂ , ΔP ｏｕｔ ). A method for predicting the operating condition of a fan in real time and online based on the data mining according to claim 4, characterized by the above.

6. In step 4), the status operating parameters of the fan are Based on the basic design parameters of the boiler and the actual operating mode of the unit, within the wide load adjustment range of the unit, the boiler evaporation amount interval [D ｂ、ｍｉｎ , D ｂ、ＢＭＣＲ is determined. In this interval, m typical operating modes are selected, and the evaporation amount set {D ｂ、ｉ} of the boiler is obtained. Based on each parameter relationship model, the set of situation parameter points in each operating mode, that is, {(D ｂ , T ｉｎ , Q ｖ , P ｅ、ｉｎ , P ｅ、ｏｕｔ , P ｔ、ｉｎ , P ｔ、ｏｕｔ ,...) ｉ} is calculated and obtained, where i = 1, 2, 3,..., m, and m >= 3. Resistance fluctuation value of the exhaust gas system (D' ｂ , ΔP ｉｎ ), based on (D' ｂ , ΔP ｏｕｔ ), the predicted values of the fan status parameters in each typical operating mode, that is, {(Q ｖ , P ｅ、ｉｎ - ΔP ｉｎ , P ｅ、ｏｕｔ + ΔP ｏｕｔ , P ｔ、ｉｎ + ΔP ｉｎ , P ｔ、ｉｎ + ΔP ｉｎ ) ｉ} (i = 1, 2, 3,... m) are calculated and obtained. Here, where Based on the fluctuations of the inlet smoke temperature and static pressure of the fan, correct the inlet volume flow rate of the fan, and calculate and obtain the total pressure Pt and specific pressure energy Y of the fan based on the predicted values of the fan's condition parameters in each typical operating mode. The predicted operating parameters of the fan are {(Q' ｖ , P' ｔ , Y') ｉ}. A method for predicting the operating status of a fan in real time and online based on data mining according to claim 5, characterized in that.

7. In step 5), based on the fan operating estimated point parameters {(Q' ｖ , P' ｔ , Y') ｉ} (i = 1, 2, 3,... m), display on the fan performance curve and obtain the opening β ｉ corresponding to the fan operating point of each operating mode, and identify and obtain the intersection points {(Q ｖ、ｓ , P ｔ、ｓ ) ｉ} of the constant opening line and the theoretical stall line, Based on the estimated point and stall point of the fan operation, each stall margin coefficient {(k ｐ , k ｑ , k ｓ ) ｉ} is calculated, The stall pressure margin factor of the fan is: where The stall flow margin factor of the fan is: where The stall safety factor of the fan is: A method for predicting the operating status of a fan in real time and online based on data mining according to claim 6, characterized in that it is as follows.

8. In step 6), the operating parameters of the fan at each operating mode point for ensuring safe and stable operation of the fan are k ｓ ≧ 1.35 and k ｐ ≧ 1.15, k ｑ ≧ 1.08; and When it is predicted by each fan that the operating point parameters do not satisfy the above conditions, even if the resistance of the exhaust gas system increases, it is explained that the fan cannot satisfy the full load range conditions, and it is necessary to perform stall prevention control of the fan. The increase value ΔP of the resistance of the exhaust gas system ｉｎ and ΔP ｏｕｔ are reduced, and re-evaluated until the operating condition parameters that satisfy the conditions are obtained. Furthermore, by monitoring the actual static pressure difference ΔP ｅ between the inlet and outlet of the fan and the opening degree β ０ of the fan, a stall prevention control strategy for the fan is formulated, that is ΔP ｅ <(P ｅ 、 ｏｕｔ +ΔP ｏｕｔ ) ＢＭＣＲ -(P ｅ、ｉｎ -ΔP ｉｎ ) ＢＭＣＲ、 β ０ <0.8β ｍａｘ A method for predicting the operating status of a fan in real-time online based on data mining according to claim 7, characterized by satisfying

Images