A boiler heating surface oxide scale shedding monitoring device and prevention and control method

Abstract

The present invention provides a monitoring device and a prevention and control method for the exfoliation of oxide scale on the boiler heating surface. The prevention and control method includes the following steps: installing an oxide scale exfoliation monitoring device on each boiler steam outlet pipe; heating water into steam through the heating surface during boiler combustion; the steam carrying the exfoliated oxide scale enters the pipeline and impacts the oxide scale exfoliation monitoring device, generating a vibration signal, and the obtained vibration signal is converted and processed to obtain an oxide scale mass monitoring value; according to the relationship between the fuel operation condition, the air damper opening degree, the boiler efficiency, the NO x content at the furnace outlet, and the oxide scale mass target value, a neural network is established to obtain a calculation model; when the oxide scale mass monitoring value exceeds the limit value, the fuel operation condition and the air damper opening degree are adjusted according to the calculation model. This method effectively reduces the leakage of the four pipes of the boiler and the solid particle corrosion of the steam turbine by controlling the mass of the oxide scale particles, slows down the erosion of the steam turbine blade cascade, and the life of the blade cascade can be increased by 50%.

Description

Technical Field

[0001] The present invention belongs to the technical field of boiler monitoring, and in particular relates to a monitoring device and a prevention and control method for oxide scale shedding on a boiler heating surface. Background Art

[0002] Under the current power generation structure dominated by thermal power generation, it is inevitable for large coal-fired thermal power units to upgrade to ultra-supercritical technology with higher parameters. China has become the country with the largest number of advanced ultra-supercritical technology units in operation and newly built in the world. The development of ultra-supercritical technology has strongly supported the transformation and upgrading of my country's power production structure towards a more efficient and clean direction. However, in the process of promoting and applying the technology, it has also brought many problems to the reliable and safe operation of my country's power generation equipment.

[0003] Among them, the blockage and burst of boiler tubes caused by the shedding of oxide scale on the high-temperature heating surface and the erosion of solid particles in the turbine are the most prominent problems. The Super304H heat-resistant steel used as the reheater tube material of the latest domestic units is used in a harsh environment (operating temperature>623℃, steam pressure>6MPa). Under the action of high temperature and steam pressure, it will cause problems such as tissue aging and steam oxidation. At the same time, the high Cr content also increases its sensitivity to intergranular corrosion. This series of problems will bring safety hazards to boiler operation. In the actual operation of thermal power units, the two types of safety problems caused by the oxide scale on the steam side of the high-temperature heating surface are still very serious: on the one hand, the concentrated shedding of oxide scale can easily cause the local tube wall of the boiler to overheat and burst, and the reliability of the unit is reduced; on the other hand, the oxide scale can easily cause the main steam valve of the turbine to be stuck and the solid particle erosion (SPE) of the moving blades.

[0004] At present, the more advanced boiler heating surface scale prevention and control method is based on the mechanism model formed by metal characteristics. The production of scale is controlled by the operating state of the tube wall of the heating surface. This method has problems such as difficulty in determining the initial state of the tube, inaccurate judgment of scale shedding, and inability to monitor the shedding scale. There are many barriers to use on active units. There are also those who will correct the offline data model, but the amount of data that needs to be detected and collected is very large, which increases the workload of maintenance personnel.

[0005] CN111413243A discloses an intelligent quantitative detection method and detector for oxide scale accumulation in boiler tubes, wherein the method is: aligning the probe of the oxide scale detector with the bending point of the boiler tube, and moving along the center line of the outer arc surface of the bend to the end of the bend, during the movement, the oxide scale detector detects the unit mass equivalent of the oxide scale in the bend at the current moment after a preset time interval, and the oxide scale detector performs a summation operation on the unit mass equivalent to obtain the total mass equivalent of the oxide scale accumulated in the boiler. This method has strict requirements on operation, has a greater impact on the measurement results, and its accuracy needs to be verified.

[0006] In summary, how to provide a simple, efficient, and accurate method and device for detecting the quality of oxide scale on the heated surface has become a problem that needs to be solved urgently. Summary of the invention

[0007] In view of the problems existing in the prior art, the purpose of the present invention is to provide a boiler heating surface oxide scale shedding monitoring device and a prevention and control method. The prevention and control method utilizes the oxide scale shedding monitoring device and, by associating multiple parameters, finds a practical boiler heating surface oxide scale prevention and control operation strategy, which effectively extends the operating life of the device and has good economic benefits.

[0008] To achieve this object, the present invention adopts the following technical solutions:

[0009] In a first aspect, the present invention provides a method for preventing and controlling the oxide scale shedding of a boiler heating surface, the method comprising the following steps:

[0010] (1) An oxide scale shedding monitoring device is installed on the steam outlet pipe of each boiler; when the boiler is burning, water is heated to steam through the heating surface; the steam carries the shed oxide scale into the pipeline and hits the oxide scale shedding monitoring device, generating a vibration signal, and the obtained vibration signal is converted and processed to obtain the oxide scale quality monitoring value;

[0011] (2) According to the fuel operation conditions, air door opening, boiler efficiency, furnace outlet NO x The relationship between the content and the target value of the oxide scale quality is established, and a neural network is established to obtain a calculation model;

[0012] When the oxide scale quality monitoring value exceeds the limit value, the fuel operation situation and the damper opening are adjusted according to the calculation model to achieve prevention and control.

[0013] The oxide scale shedding monitoring device of the present invention can be directly installed on existing equipment without replacing the entire equipment, and has a low cost of use. After installation, it can realize real-time online monitoring. The monitoring value referred to in step (1) specifically refers to the monitoring value of a period of time when more oxide scale is produced after the boiler has been running stably for a long time; and the target value in step (2) refers to the value under the optimal state when the boiler is running stably (that is, the period of time when less oxide scale is produced).

[0014] The prevention and control method of the present invention combines the measurement data with the neural network, and correlates the relevant parameters such as the boiler operation mode, which greatly improves the accuracy of the oxide scale mass measurement. The oxide scale shedding monitoring device is controlled according to the obtained calculation model to achieve the control of the oxide scale quality, effectively reduce the leakage of the four tubes of the boiler, and reduce the oxide scale in the steam from the boiler side to the steam turbine side. At the same time, it can also effectively reduce the corrosion of solid particles in the steam turbine, slow down the erosion of the turbine blades, and increase the blade life by 50%.

[0015] The following are preferred technical solutions of the present invention, but are not intended to be limitations of the technical solutions provided by the present invention. Through the following technical solutions, the technical objectives and beneficial effects of the present invention can be better achieved and realized.

[0016] As a preferred technical solution of the present invention, step (1) converts the vibration signal into a calculation formula for the oxide scale mass:

[0017]

[0018] Wherein, M is the mass of oxide scale in the steam pipe, mg / h; α is the signal correction coefficient, mg / (h·V); R is the radius of the steam pipe, m; L is the length of the connecting rod in the steam pipe, m; a is the steam flow rate correction coefficient s / m; b is the particle radial distribution correction coefficient; v is the steam flow rate, m / s; f is the voltage signal value received by the signal receiver, V.

[0019] As a preferred technical solution of the present invention, the neural network algorithm in step (2) includes a BP neural network algorithm.

[0020] As a preferred technical solution of the present invention, the types of neurons in the BP neural network input layer include 7 types, including unit load, main steam temperature, main steam pressure, superheater cooling water volume, coal mill current, coal mill outlet separator damper opening and air door damper opening.

[0021] Among them, the specific number of neurons in the input layer includes: 1 for unit load, 1 for main steam temperature, 1 for main steam pressure, 1 for superheater cooling water volume, the number of coal mill currents and the number of coal mill outlet separator damper openings depend on the specific number of coal mills, and the number of air door damper openings depends on the specific situation.

[0022] Preferably, the number of hidden layers of the BP neural network is 7.

[0023] Preferably, the number of neurons in the output layer of the BP neural network is 3, which are respectively boiler efficiency, furnace outlet NO x Content and quality of oxide scale.

[0024] As a preferred technical solution of the present invention, the neuron transfer function from the input layer to the hidden layer of the BP neural network includes: Here, t only refers to an unknown number and has no actual meaning.

[0025] Preferably, the neuron transfer function from the hidden layer to the output layer of the BP neural network includes: or Here, z only refers to an unknown number and has no actual meaning.

[0026] As a preferred technical solution of the present invention, the maximum training cycle of the BP neural network is 100, and the training accuracy is 95%.

[0027] As a preferred technical solution of the present invention, the learning rate of the BP neural network is 0.1-0.4, such as 0.1, 0.2, 0.3 or 0.4, but is not limited to the listed values, and other unlisted values ​​within the numerical range are also applicable.

[0028] Exemplarily, the process of obtaining the calculation model using the BP neural network algorithm in the present invention includes:

[0029] S1: 38 input vectors and 3 output vectors are selected for control research. The input vectors are 1 unit load Pe (output load of the unit, unit: MW), 1 main steam temperature T (unit: °C), 1 main steam pressure (unit: MPa), 1 superheater cooling water volume (unit: t / h), 7 coal mill currents (unit: A), 7 coal mill outlet separator damper openings (unit: %), and 20 damper openings (unit: %), that is, X = [x1, x2, x3, x4, ..., x 38 ]; the output vectors are boiler efficiency (in %), furnace outlet NO x content (in ppm) and the target value of scale mass (mg / h), i.e., the expected output vector D = [d1, d2, d3] and the actual output vector O = [o1, o2, o3];

[0030] S2: Establish a 38:7:3 network. The ratio refers to the ratio of neurons in the BP neural network input layer, the BP neural network hidden layer, and the BP neural network:

[0031] Initialize the parameters of the BP neural network, including the connection weights v of the input layer and the hidden layer ij , the connection weights w between the hidden layer and the output layer jk , hidden layer threshold p j , and the output layer threshold q k ;

[0032] S3: According to the input vector X, the connection weight v between the input layer and the hidden layer ij , and the hidden layer threshold p j , calculate the hidden layer output Y, the formula is as follows:

[0033]

[0034] Among them, y i is the hidden layer output; n is the number of parameters in the input vector; h is the number of neurons in the hidden layer; f(·) is the neuron transfer function from the input layer to the hidden layer, f(t); v 0j =-1, x0 = p j ;

[0035] S3: According to the hidden layer output Y, the connection weight w between the hidden layer and the output layer jk , and the output layer threshold q k , calculate the actual output O of the output layer, the formula is as follows:

[0036]

[0037] Among them, O k is the actual output of the output layer; l is the number of neurons in the output layer; f(·) is the neuron transfer function from the hidden layer to the output layer, f(z);

[0038] S4: Calculate the overall network error E based on the actual output O and the expected output D. The formula is as follows:

[0039]

[0040] S5: Update each weight according to the overall network error E. The formula is as follows:

[0041] v ij =v ij +Δv ij ;

[0042] w jk =w jk +Δw jk ;

[0043]

[0044]

[0045]

[0046] Where η is the learning rate.

[0047] S6: Repeat the iteration until the overall error of the network meets the given accuracy and the oxide scale mass calculation model is obtained; when the oxide scale mass M value is too large, the operation is optimized according to the calculation model given by the neural network to control the wall temperature over-temperature amplitude and the variable load rate to guide the operation.

[0048] In a second aspect, the present invention provides an oxide scale shedding monitoring device as described in the first aspect, wherein the oxide scale shedding monitoring device comprises an oxide scale monitoring connecting rod, a signal converter and a signal receiver connected in sequence, and the oxide scale monitoring connecting rod is inserted into a steam pipe connected to the heating surface of the boiler.

[0049] As a preferred technical solution of the present invention, the material of the oxide scale monitoring connecting rod includes stainless steel.

[0050] As a preferred technical solution of the present invention, the signal converter includes a vibration pickup, an F / V signal conversion circuit, an amplification circuit and a standardization circuit.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] (1) The oxide scale shedding monitoring device of the present invention can be directly installed on existing equipment, and the concentration of oxide scale particles in the steam pipe can be determined by the impact strength of the oxide scale particles on the protrusions in the pipe. There is no need to replace the entire equipment, and the use cost is low. In addition, multiple monitoring devices can be aggregated into a collection center and then connected to the DCS, which has strong scalability and wide coverage.

[0053] (2) The prevention and control method of the present invention combines the measurement data with the neural network, and correlates the relevant parameters such as the boiler operation mode, which greatly improves the accuracy of the oxide scale mass measurement. The oxide scale shedding monitoring device is controlled according to the obtained calculation model to achieve the control of the oxide scale quality, effectively reduce the leakage of the four pipes of the boiler, and reduce the oxide scale in the steam from the boiler side to the steam turbine side. At the same time, it can also effectively reduce the corrosion of solid particles in the steam turbine, slow down the erosion of the turbine blades, and increase the blade life by 50%. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 Schematic diagram of the installation of the oxide scale shedding monitoring device provided in Example 1 of the present invention in a boiler system;

[0055] Figure 2 The present invention Figure 1 A partial enlarged schematic diagram of circle A.

[0056] Figure 3 It is a network structure diagram of the BP neural network provided in Example 1 of the present invention.

[0057] Among them, 1-scaling monitoring connecting rod, 2-signal converter, 3-signal receiver, 41-first steam turbine, 42-second steam turbine, 5-reheater, 6-economizer, 7-generator, 8-boiler.

[0058] The arrows indicate the direction of steam flow. DETAILED DESCRIPTION

[0059] In order to better illustrate the present invention and facilitate understanding of the technical solution of the present invention, the present invention is further described in detail below. However, the following embodiments are only simple examples of the present invention and do not represent or limit the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the claims.

[0060] The present invention provides an oxide scale shedding monitoring device, which comprises an oxide scale monitoring connecting rod 1, a signal converter 2 and a signal receiver 3 which are connected in sequence;

[0061] Furthermore, the material of the oxide scale monitoring connecting rod 1 includes stainless steel.

[0062] Furthermore, the signal converter 2 includes a vibration pickup, an F / V signal conversion circuit, an amplification circuit and a standardization circuit.

[0063] On the other hand, the schematic diagram of the installation structure of the oxide scale shedding monitoring device in the boiler system is as follows Figure 1 As shown, Figure 1 The enlarged view of the part of the middle circle A is as follows Figure 2 As shown;

[0064] The boiler system includes a boiler 8, a first steam turbine 41 and a second steam turbine 42 connected to the boiler 8 via steam pipes, a reheater 5 connected to the first steam turbine 41, and an economizer 6 connected to the second steam turbine 42;

[0065] The reheater 5 and the economizer 6 are arranged inside the boiler 8;

[0066] The first steam turbine 41 and the second steam turbine 42 are also connected to the generator 7;

[0067] The oxide scale monitoring connecting rod 1 of the oxide scale shedding monitoring device is inserted into the steam pipe.

[0068] The following are typical but non-limiting embodiments of the present invention:

[0069] Embodiment 1:

[0070] This embodiment provides a method for preventing and controlling the oxide scale shedding on the heating surface of a boiler, which is performed using the boiler system and the oxide scale shedding monitoring device described in the specific implementation manner. The prevention and control method includes the following steps:

[0071] (1) An oxide scale shedding monitoring device is installed on the steam outlet pipe of the boiler; when the boiler is burning, water is heated to steam through the heating surface; the steam carries the shed oxide scale into the pipeline and hits the oxide scale shedding monitoring device, generating a vibration signal. The obtained vibration signal is converted and processed to obtain the oxide scale quality monitoring value, which specifically includes:

[0072]

[0073] Among them, α=0.253mg / (h·V); R=0.291m; L=0.021m; a=0.13s / m; b=1.1; v=30m / s; f=10V. Substituting each parameter, the calculated oxide scale mass M=13.45mg / h;

[0074] (2) According to the fuel operation conditions, air door opening, boiler efficiency, furnace outlet NO x The relationship between the content and the target value of the oxide scale quality is established, and a neural network is established to obtain a calculation model, which specifically includes:

[0075] S1: 38 input vectors and 3 output vectors are selected for control research. The input vectors are unit load Pe (output load of the unit, unit MW), main steam temperature T (unit ℃), main steam pressure (unit MPa), superheater desuperheating water volume (unit t / h), 7 coal mill currents (unit A), 7 coal mill outlet separator damper openings (unit %), 20 damper openings (unit %), that is, X=[500,605,28,20,52,52,52,0,52,0,60,60,60,60,60,60,50,100,50,80,50,50,50,30,50,50,50,70,50,80,50,100,50,100,100,100]; the output vector is the boiler efficiency (in %), the furnace outlet NO x content (in ppm) and the target value of scale mass (mg / h), including the expected output vector D = [94.2, 280, 10] and the actual output vector O = [o1, o2, o3];

[0076] S2: Establish a 38:7:3 network, such as Figure 3 As shown, the ratio refers to the ratio of the BP neural network input layer, the BP neural network hidden layer and the neurons of the BP neural network:

[0077] Initialize the parameters of the BP neural network, including the connection weights v of the input layer and the hidden layer ij , the connection weights w between the hidden layer and the output layer jk , hidden layer threshold p j , and the output layer threshold q k ;

[0078] S3: According to the input vector X, the connection weight v between the input layer and the hidden layer ij , and the hidden layer threshold p j , calculate the hidden layer output Y, the formula is as follows:

[0079]

[0080] Among them, y i is the hidden layer output; n is the number of parameters in the input vector; h is the number of neurons in the hidden layer; f(·) is the neuron transfer function from the input layer to the hidden layer, v 0j =-1, x0 = p j ;

[0081] S3: According to the hidden layer output Y, the connection weight w between the hidden layer and the output layer jk , and the output layer threshold q k , calculate the actual output O of the output layer, the formula is as follows:

[0082]

[0083] Among them, O k is the actual output of the output layer; l is the number of neurons in the output layer; f(·) is the neuron transfer function from the hidden layer to the output layer,

[0084] S4: Calculate the overall network error E based on the actual output O and the expected output D. The formula is as follows:

[0085]

[0086] S5: Update each weight according to the overall network error E. The formula is as follows:

[0087] v ij =v ij +Δv ij ;

[0088] w jk =w jk +Δw jk ;

[0089]

[0090]

[0091]

[0092] Among them, η is the learning rate, which is 0.2.

[0093] S6: Repeat the iteration until the overall network error meets the given accuracy (the maximum training cycle is set to 100 and the training accuracy is set to 95%), and the scale quality calculation model is obtained;

[0094] When the scale mass M value is too large (i.e. the monitoring value exceeds 25 mg / h), the operation of each coal mill input mode, the damper door opening of each coal mill outlet separator and the opening of each air damper are optimized according to the calculation model given by the neural network. In this embodiment, the input vector X is adjusted to [500, 605, 28, 20, 52, 52, 52, 52, 52, 0, 0, 50, 50, 50, 50, 50, 50, 50, 15, 100, 15, 80, 15, 50, 15, 25, 50, 35, 50, 70, 50, 80, 50, 100, 50, 75, 80, 100] to achieve prevention and control.

[0095] The present invention illustrates the device and detailed method of the present invention through the above embodiments, but the present invention is not limited to the above devices and detailed methods, that is, it does not mean that the present invention must rely on the above devices and detailed methods to be implemented. Those skilled in the art should understand that any improvement of the present invention, equivalent replacement of the operation of the present invention, addition of auxiliary operations, selection of specific methods, etc., all fall within the protection scope and disclosure scope of the present invention.

Claims

1. A method for preventing and controlling the oxide scale shedding of a boiler heating surface, characterized in that: The prevention and control method comprises the following steps: (1) An oxide scale shedding monitoring device is installed on the steam outlet pipe of each boiler; when the boiler is burning, water is heated to steam through the heating surface; the steam carries the shed oxide scale into the pipeline and hits the oxide scale shedding monitoring device, generating a vibration signal, and the obtained vibration signal is converted and processed to obtain the oxide scale quality monitoring value; (2) According to the fuel operation conditions, air door opening, boiler efficiency, furnace outlet NO x The relationship between the content and the target value of the oxide scale quality is established, and a neural network is established to obtain a calculation model; The neural network algorithm includes a BP neural network algorithm; The process of obtaining the calculation model using the BP neural network algorithm includes: S1: 38 input vectors and 3 output vectors are selected for control research. The input vectors are 1 unit load Pe, which is the output load of the unit in MW, 1 main steam temperature T in °C, 1 main steam pressure in MPa, 1 superheater cooling water volume in t / h, 7 coal mill currents in A, 7 coal mill outlet separator damper openings in %, and 20 damper openings in %, that is, X = [x1, x2, x3, x4, ..., x 38 ]; the output vector is boiler efficiency, in %, furnace outlet NO x content, in ppm, and the target value of scale mass, in mg / h, including the expected output vector D = [d1, d2, d3] and the actual output vector O = [o1, o2, o3]; S2: Establish a 38:7:3 network. The ratio refers to the ratio of neurons in the BP neural network input layer, the BP neural network hidden layer, and the BP neural network: Initialize the parameters of the BP neural network, including the connection weights v of the input layer and the hidden layer ij , the connection weights w between the hidden layer and the output layer jk , hidden layer threshold p j , and the output layer threshold q k ; The number of neurons in each hidden layer of the BP neural network is 7; S3: According to the input vector X, the connection weight v between the input layer and the hidden layer ij , and the hidden layer threshold p j , calculate the hidden layer output Y, the formula is as follows: Among them, y i is the hidden layer output; n is the number of parameters in the input vector; h is the number of neurons in the hidden layer; f(·) is the neuron transfer function from the input layer to the hidden layer, f(t); v 0j =-1, x0 = p j ; S3: According to the hidden layer output Y, the connection weight w between the hidden layer and the output layer jk , and the output layer threshold q k , calculate the actual output O of the output layer, the formula is as follows: Among them, O k is the actual output of the output layer; l is the number of neurons in the output layer; f(·) is the neuron transfer function from the hidden layer to the output layer, f(z); S4: Calculate the overall network error E based on the actual output O and the expected output D. The formula is as follows: S5: Update each weight according to the overall network error E. The formula is as follows: v ij =v ij +Δv ij ; In jk =in jk +Δw jk ; Where η is the learning rate; S6: Repeat the iteration until the overall error of the network meets the given accuracy and the oxide scale mass calculation model is obtained; when the oxide scale mass M value is too large, the operation is optimized according to the calculation model given by the neural network, and the wall temperature over-temperature amplitude and the variable load rate are controlled to guide the operation; When the oxide scale quality monitoring value exceeds the limit value, the fuel operation situation and the damper opening are adjusted according to the calculation model to achieve prevention and control.

2. The control method according to claim 1, characterized in that: Step (1) converts the vibration signal into the calculation formula of the oxide scale mass: Wherein, M is the mass of oxide scale in the steam pipe, mg / h; α is the signal correction coefficient, mg / (h·V); R is the radius of the steam pipe, m; L is the length of the connecting rod in the steam pipe, m; a is the steam flow rate correction coefficient s / m; b is the particle radial distribution correction coefficient; v is the steam flow rate, m / s; f is the voltage signal value received by the signal receiver, V.

3. The prevention and control method according to claim 1, characterized in that: The neuron types of the BP neural network input layer include 7 types, including unit load, main steam temperature, main steam pressure, superheater desuperheating water volume, coal mill current, coal mill outlet separator damper opening and air door damper opening.

4. The control method according to claim 1, characterized in that: The number of neurons in the output layer of the BP neural network is 3, which are boiler efficiency, furnace outlet NO x Content and quality of oxide scale.

5. The prevention and control method according to claim 1, characterized in that: The neuron transfer function from the input layer to the hidden layer of the BP neural network includes:

6. The control method according to claim 5, characterized in that: The neuron transfer function from the hidden layer to the output layer of the BP neural network includes: or 7. The control method according to claim 1, characterized in that: The maximum training cycle of the BP neural network is 100, and the training accuracy is 95%.

8. The control method according to claim 1, characterized in that: The learning rate η of the BP neural network is 0.1-0.

4.

9. An oxide scale shedding monitoring device used in the prevention and control method according to any one of claims 1 to 8, characterized in that: The oxide scale shedding monitoring device comprises an oxide scale monitoring connecting rod, a signal converter and a signal receiver which are connected in sequence, and the oxide scale monitoring connecting rod is inserted into a steam pipe connected to the heating surface of the boiler.

10. The oxide scale shedding monitoring device according to claim 9, characterized in that: The material of the oxide scale monitoring connecting rod includes stainless steel.

11. The oxide scale shedding monitoring device according to claim 9, characterized in that: The signal converter includes a vibration pickup, an F / V signal conversion circuit, an amplification circuit and a standardization circuit.

Images