IES87197B2 - System and method for diagnosing state fault of desulfurization and denitration circulating pump - Google Patents

Abstract

Disclosed is a system and a method for diagnosing the state fault of a desulfurization and denitration circulating pump, including an input layer, a hidden layer, an output layer and a data acquisition system, wherein a data input end of the data acquisition system is connected with data output ends of a distributed pump unit monitoring system and a power plant-level real-time monitoring system, respectively; the hidden layer includes a fault diagnosis system, the fault diagnosis system includes a BP neural network server and a master controller, the BP neural network server electrically inputs to connect with the master controller, the master controller electrically outputs to connect with the output module, and the master controller electrically inputs to connect with the input module; the output of the BP neural network server trained from a large amount of operation data of the power plant is used as a fitness function of the follow-up intelligent optimization algorithm to incorporate two advanced algorithm means, so that the current problems of inaccurate fault diagnosis data, false results, poor universality of the optimization algorithm, insufficient real-time performance, and the like in the field of the desulfurization and denitration circulating pump are solved cooperatively.

Description

SYSTEM AND METHOD FOR DIAGNOSING STATE FAULT OF DESULFURIZATION AND DENITRATION CIRCULATING PUMP FIELD OF INVENTION 1. 1. 1. id="p-1" id="p-1"

[001] particular to a system and a method for diagnosing the state fault of a desulfurization and The invention relates to the technical field of thermal power generation unit, in denitration circulating pump.

BACKGROUND TO INVENTION 2. 2. 2. id="p-2" id="p-2"

[002] problem to address in the current environmental protection work. A large amount of sulfur and Sulfur and nitrate cause a high level of pollution to the environment, which is the major nitrate are generated in the production of a thermal power plant and may damage the environment greatly if not controlled, hence desulfurization and denitration devices are now generally used in the power industry. 3. 3. 3. id="p-3" id="p-3"

[003] desulphurization and denitration by repeatedly contacting the absorbent slurry in the A desulphurization and denitration circulating pump can achieve the effect of absorption tower with flue gas. Vlfith the continuous and rapid development of automation and intelligentization of power plants, more and more domestic power plants have conducted researches on the service life of the desulfurization and denitration circulating pump, which is subjected to many factors. Failure of the desulfurization and denitration circulating pump in the operation may lead to huge economic losses. The conventional failure analysis and diagnosis methods for the desulfurization and denitration circulating pump are time-consuming and labor-consuming, depending on the experience of field operators, featuring defects in the timeliness, speed and inefficiency of failure analysis and diagnosis. In most cases, an emergency repair is performed after the device is damaged and down, the normal operation of the desulfurization and denitration system has to hold up, and the environmental protection target of the power plant is hard to achieve. 4. 4. 4. id="p-4" id="p-4"

[004] overhaul the desulfurization and denitration circulating pump and reduce the economic loss, The fault diagnosis of the desulfurization and denitration circulating pump can timely therefore, the fault diagnosis has always played an active role in the field of scientific research and practical engineering related to quality and economic improvement. However, it is hard to find an optimal theoretical solution to the difficulties in the fault diagnosis of circulating pumps, for example, high dimensionality, non-convexity, discreteness, and non-linearity. At present, the conventional fault diagnosis methods for the desulfurization and denitration circulating pump include fault tree, vector machine, machine learning algorithm, and the like, and most of the conventional methods are based on the premise that the fault curve of the desulfurization and denitration circulating pump can be fitted by polynomials, showing a high threshold in use and limited optimization precision. . . . id="p-5" id="p-5"

[005] and foresight for the fault diagnosis of the desulfurization and denitration circulating pump is According to the invention, a BP neural network server which has certain predictability used as a fitness function of a follow-up intelligent optimization algorithm to incorporate two advanced algorithm means, so that the current problems of inaccurate fault diagnosis data, false results, poor universality of the optimization algorithm, insufficient real-time performance, and the like in the field of the desulfurization and denitration circulating pump are solved cooperatively.

SUMMARY OF INVENTION 6. 6. 6. id="p-6" id="p-6"

[006] system and a method for diagnosing the state fault of a desulfurization and denitration It is an object of the invention to solve the problems in the prior art by providing a circulating pump. 7. 7. 7. id="p-7" id="p-7"

[007] system and a method for diagnosing the state fault of a desulfurization and denitration To achieve the above object, the invention provides the following technical solution: a circulating pump, comprising an input layer, a hidden layer, an output layer and a data acquisition system, wherein a data input end of the data acquisition system is connected with data output ends of a distributed pump unit monitoring system and a power plant-level real- time monitoring system, respectively; the input layer comprises an input module, the data acquisition system is in wireless communication connection with the input module through a wireless transmission module; the hidden layer comprises a fault diagnosis system, the fault diagnosis system comprises a BP neural network server and a master controller, the BP neural network server electrically inputs to connect with the master controller, the output layer comprises an output module, the master controller electrically outputs to connect with the output module, the master controller electrically inputs to connect with the input module, and the output module is in wireless communication connection with the distributed pump unit monitoring system through the wireless transmission module. 8. 8. 8. id="p-8" id="p-8"

[008] Preferably, the data acquisition system comprises a data acquisition module and a data sorting module. 9. 9. 9. id="p-9" id="p-9"

[009] Preferably, the input module and the output module are a wireless receiver and a wireless transmitter, respectively. . . . id="p-10" id="p-10"

[0010] Preferably, the wireless transmission module employs an OPC communication protocol. 11. 11. 11. id="p-11" id="p-11"

[0011] A method used for the system for diagnosing the state fault of a desulfurization and denitration circulating pump, comprising the steps of: 12. 12. 12. id="p-12" id="p-12"

[0012] S1: acquiring motion parameters acquired by a positron sensor of the desulfurization and denitration circulating pump in the distributed pump unit monitoring system for a period of time through the data acquisition module, at an interval of 30 minutes between adjacent groups of operation data, continuously acquiring each item of data in the period of time, identifying and removing transient operation data of a unit in the data through the data sorting module, and sorting out steady-state operation data of the unit; 13. 13. 13. id="p-13" id="p-13"

[0013] S2: acquiring operation parameters acquired by the desulfurization and denitration circulating pump recorded in the power plant-level real-time monitoring system, taking the same with the operation data of each unit obtained through sorting by the distributed pump unit monitoring system as input data for the input layer of the BP neural network server, and dividing the same into training data and test data of the network; 14. 14. 14. id="p-14" id="p-14"

[0014] S3: organizing the operation data acquired by the data acquisition system into a matrix form, taking the data as input data for the BP neural network server through the OPC communication protocol, executing an algorithm by the master controller to construct and test the BP neural network server, selecting a Sigmoid function as a transmission function for the hidden layer of the BP neural network server, selecting a linear function as the transmission function for the output layer, selecting an L-M optimization algorithm as a weight training algorithm to train for 5,000 times of iteration, and testing a trained neural network by using a test sample; . . . id="p-15" id="p-15"

[0015] 84: using a heuristic intelligent optimization algorithm as an optimization main program, using an output of the trained neural network in the main program as a fitness function of the heuristic intelligent optimization algorithm and as a basis for data evolution and iteration operated in the program, using a particle swarm optimization (PSO) algorithm with smaller calculation amount as a main algorithm for a load fault diagnosis algorithm by the heuristic intelligent optimization algorithm; and according to each value representing one fault diagnosis data of the unit, inputting the unit into the trained neural network corresponding to each unit by the algorithm, obtaining fault diagnosis data of the desulfurization and denitration circulating pump corresponding to each unit through neural network calculation, and adding up the fault diagnosis data to obtain fault diagnosis data of all units in the plant; and 16. 16. 16. id="p-16" id="p-16"

[0016] S5: continuously updating the operation parameters acquired by the desulfurization and denitration circulating pump and an updating speed according to the PS0 algorithm principle, continuously updating and calculating the fault diagnosis data of the unit until the fault diagnosis data do not change anymore (i.e., the algorithm is considered to be convergent), outputting the fault data of each unit corresponding to the fault diagnosis data as an actual instruction of each unit in the power plant meeting the fault diagnosis, terminating the algorithm, and transmitting the output back to the distributed pump unit monitoring system through the output module and the wireless transmission module. p. 17. 17. 17. id="p-17" id="p-17"

[0017] Preferably, the parameters are determined to have entered the steady state when the operation parameters acquired by the desulfurization and denitration circulating pump are smaller than a specified threshold value within 15 minutes, i.e., in a steady-state working condition; the parameter data are stored in a steady-motion working condition database, othen/vise, the next 5 more minutes are required to determine whether the parameters have entered the steady state again until the data meet the steady state-requirement. 18. 18. 18. id="p-18" id="p-18"

[0018] Preferably, a process for dividing the input data into the training data and the test data of the network includes conducting continuous differentiation on the operation parameters acquired by the desulfurization and denitration circulating pump, employing supervised- learning generalized radial basis function for the network to use both a center position and weights, selecting 75% of the total data randomly as training samples for the training of the BP neural network server, and taking the rest 25% samples as the network test samples. 19. 19. 19. id="p-19" id="p-19"

[0019] Compared with the prior art, the invention has the following advantages: . . . id="p-20" id="p-20"

[0020] (1) in the invention, a large amount of power plant operation data are used for training the BP neural network server providing certain predictability and foresight on the coal consumption characteristics, so that accurate and rapid calculation of the fault diagnosis of the desulfurization and denitration circulating pump can be realized, and accurate and effective fault diagnosis data have important values for guiding the safe and economic operation of the power plant; 21. 21. 21. id="p-21" id="p-21"

[0021] (2) in the invention, the heuristic intelligent optimization algorithm is used in the fault diagnosis calculation of the desulfurization and denitration circulating pump; compared with the prior art or system, the heuristic intelligent optimization algorithm has better adaptability to data sources, better universality and greater potentials to popularize; moreover, the heuristic intelligent optimization algorithm optimizes for the fault diagnosis of the desulfurization and denitration circulating pump, featuring faster in the calculation and higher in calculation accuracy, satisfying the requirements of real-time performance and accuracy in the production of a power plant; and 22. 22. 22. id="p-22" id="p-22"

[0022] (3) in the invention, the output of the BP neural network server trained from a large amount of operation data of the power plant is used as a fitness function of the follow-up intelligent optimization algorithm to incorporate two advanced algorithm means, so that the current problems of inaccurate fault diagnosis data, false results, poor universality of the optimization algorithm, insufficient real-time performance, and the like in the field of the desulfurization and denitration circulating pump are solved cooperatively.

BRIEF DESCRIPTION OF DRAWINGS 23. 23. 23. id="p-23" id="p-23"

[0023] To more clearly illustrate the technical solution of the embodiments of the present invention, the drawings used for describing the embodiments will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained from those drawings by those skilled in the art without involving any inventive effort. 24. 24. 24. id="p-24" id="p-24"

[0024] FIG. 1 is a schematic view showing a structure of the present invention. . . . id="p-25" id="p-25"

[0025] In the drawings, the parts and components indicated by the reference numerals are listed as follows: 1-input layer, 2-hidden layer, 3-output layer, 4-data acquisition system, 5- distributed pump unit monitoring system, 6-power plant-level real-time monitoring system, 7- fault diagnosis system, 8-BP neural network server, 9-master controller, 10-input module, 11- output module, 12-wireless transmission module.

DETAILED DESCRIPTION OF DRAWINGS 26. 26. 26. id="p-26" id="p-26"

[0026] The technical solutions in the embodiments of the present invention will be clearly and fully described below in conjunction with the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only some rather than all of the embodiments of the present invention. On such a basis, all other embodiments obtained by those of ordinary skill in the art without inventive effort shall fall within the protection scope of the present invention. 27. 27. 27. id="p-27" id="p-27"

[0027] Referring to FIG. 1, the invention provides the following technical solution: a system and a method for diagnosing the state fault of a desulfurization and denitration circulating pump, including an input layer 1, a hidden layer 2, an output layer 3 and a data acquisition system 4, wherein a data input end of the data acquisition system 4 is connected with data output ends of a distributed pump unit monitoring system 5 and a power plant-level real-time monitoring system 6, respectively; the input layer 1 includes an input module 10, the data acquisition system 4 is in wireless communication connection with the input module 10 through a wireless transmission module 12; the hidden layer 2 includes a fault diagnosis system 7, the fault diagnosis system 7 includes a BP neural network server 8 and a master controller 9, the BP neural network server 8 electrically inputs to connect with the master controller 9, the output layer 3 includes an output module 11, the master controller 9 electrically outputs to connect with the output module 11, the master controller 9 electrically inputs to connect with the input module 10, and the output module 11 is in wireless communication connection with the distributed pump unit monitoring system 5 through the wireless transmission module 12; wherein the data acquisition system 4 includes a data acquisition module and a data sorting module; the input module 10 and the output module 11 are a wireless receiver and a wireless the wireless transmission module 12 employs an OPC transmitter, respectively; communication protocol. 28. 28. 28. id="p-28" id="p-28"

[0028] A specific application of the embodiment is a method used for the system for diagnosing the state fault of a desulfurization and denitration circulating pump, including the steps of: 29. 29. 29. id="p-29" id="p-29"

[0029] S1: acquiring motion parameters acquired by a positron sensor of the desulfurization and denitration circulating pump in the distributed pump unit monitoring system 5 for a period of time through the data acquisition module, at an interval of 30 minutes between adjacent groups of operation data, continuously acquiring each item of data in the period of time, identifying and removing transient operation data of a unit in the data through the data sorting module, and sorting out steady-state operation data of the unit; . . . id="p-30" id="p-30"

[0030] wherein the system is determined to have entered the steady state when the operation parameters acquired by the desulfurization and denitration circulating pump are smaller than a specified threshold value within 15 minutes, i.e., in a steady-state working condition; the parameter data are stored in a steady-motion working condition database, othen/vise, the next more minutes are required to determine whether the system has entered the steady state again until the data meet the steady state-requirement; 31. 31. 31. id="p-31" id="p-31"

[0031] S2: acquiring operation parameters acquired by the desulfurization and denitration circulating pump recorded in the power plant-level real-time monitoring system 6, taking the same with the operation data of each unit obtained through sorting by the distributed pump unit monitoring system 5 as input data for the input layer 1 of the BP neural network server 8, and dividing the same into training data and test data of the network; 32. 32. 32. id="p-32" id="p-32"

[0032] wherein it’s necessary to conduct continuous differentiation on the operation parameters acquired by the desulfurization and denitration circulating pump, employ supervised-learning generalized radial basis function for the network to use both a center position and weights, select 75% of the total data randomly as training samples for the training of the BP neural network server 8, and take the rest 25% samples as the network test samples, to correspond to the operation data; 33. 33. 33. id="p-33" id="p-33"

[0033] S3: organizing the operation data acquired by the data acquisition system 4 into a matrix form, taking the data as input data for the BP neural network server 8 through the OPC communication protocol, executing an algorithm by the master controller 9 to construct and test the BP neural network server 8, selecting a Sigmoid function as a transmission function for the hidden layer 2 of the BP neural network server 8, selecting a linear function as the transmission function for the output layer, selecting an L-M optimization algorithm as a weight training algorithm to train for 5,000 times of iteration, and testing a trained neural network by using a test sample; 34. 34. 34. id="p-34" id="p-34"

[0034] 84: using a heuristic intelligent optimization algorithm as an optimization main program, using an output of the trained neural network in the main program as a fitness function of the heuristic intelligent optimization algorithm and as a basis for data evolution and iteration operated in the program, using a particle swarm optimization (PSO) algorithm with smaller calculation amount as a main algorithm for a load fault diagnosis algorithm by the heuristic intelligent optimization algorithm; and according to each value representing one fault diagnosis data of the unit, inputting the unit into the trained neural network corresponding to each unit by the algorithm, obtaining fault diagnosis data of the desulfurization and denitration circulating pump corresponding to each unit through neural network calculation, and adding up the fault diagnosis data to obtain fault diagnosis data of all units in the plant; and . . . id="p-35" id="p-35"

[0035] S5: continuously updating the operation parameters acquired by the desulfurization and denitration circulating pump and an updating speed according to the PS0 algorithm principle, continuously updating and calculating the fault diagnosis data of the unit until the fault diagnosis data do not change anymore (i.e., the algorithm is considered to be convergent), outputting the fault data of each unit corresponding to the fault diagnosis data as an actual instruction of each unit in the power plant meeting the fault diagnosis, terminating the algorithm, and transmitting the output back to the distributed pump unit monitoring system through the output module 11 and the wireless transmission module 12. 36. 36. 36. id="p-36" id="p-36"

[0036] In the description, the terms “one embodiment”, “an example”, “a specific example”, and the like indicate that a specific feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In the description, general expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific feature, structure, material, or characteristic described may be combined in any suitable manner in any one or more embodiments or examples. 37. 37. 37. id="p-37" id="p-37"

[0037] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiment does not describe all the details in detail, nor does it limit the invention to only the described specific embodiments. Obviously, many modifications and changes can be made according to the disclosure of the description.

This description selects and specifically describes these embodiments to better explain the principles and practical applications of the present invention, so that those skilled in the art can better understand and use the present invention. The present invention is only defined by the claims in their full scope and equivalents.

Claims (5)

1. A system for diagnosing the state fault of a desulfurization and denitration circulating pump, comprising an input layer (1), a hidden layer (2), an output layer (3) and a data acquisition system (4), characterized in that a data input end of the data acquisition system (4) is connected with data output ends of a distributed pump unit monitoring system (5) and a power plant-level real-time monitoring system (6), respectively; the input layer (1) comprises an input module (10), the data acquisition system (4) is in wireless communication connection with the input module (10) through a wireless transmission module (12); the hidden layer (2) comprises a fault diagnosis system (7), the fault diagnosis system (7) comprises a BP neural network server (8) and a master controller (9), the BP neural network server (8) electrically inputs to connect with the master controller (9), the output layer (3) comprises an output module (11), the master controller (9) electrically outputs to connect with the output module (11), the master controller (9) electrically inputs to connect with the input module (10), and the output module (11) is in wireless communication connection with the distributed pump unit monitoring system (5) through the wireless transmission module (12); wherein the data acquisition system (4) comprises a data acquisition module and a data sorting module; and the input module (10) and the output module (11) are a wireless receiver and a wireless transmitter, respectively.

2. The system and method for diagnosing the state fault of a desulfurization and denitration circulating pump according to claim 1, characterized in that the wireless transmission module (12) employs an OPC communication protocol.

3. A method used for the system for diagnosing the state fault of a desulfurization and denitration circulating pump, characterized by comprising the steps of: S1: acquiring motion parameters acquired by a positron sensor of the desulfurization and denitration circulating pump in the distributed pump unit monitoring system (5) for a period of time through the data acquisition module, at an interval of 30 minutes between adjacent groups of operation data, continuously acquiring each item of data in the period of time, identifying and removing transient operation data of a unit in the data through the data sorting module, and sorting out steady-state operation data of the unit; 82: acquiring operation parameters acquired by the desulfurization and denitration circulating pump recorded in the power plant-level real-time monitoring system (6), taking the same with the operation data of each unit obtained through sorting by the distributed pump unit p.8 monitoring system (5) as input data for the input layer (1) of the BP neural network server (8), and dividing the same into training data and test data of the network; S3: organizing the operation data acquired by the data acquisition system (4) into a matrix form, taking the data as input data for the BP neural network server (8) through the OPC communication protocol, executing an algorithm by the master controller (9) to construct and test the BP neural network server (8), selecting a Sigmoid function as a transmission function for the hidden layer (2) of the BP neural network server (8), selecting a linearfunction as the transmission function for the output layer, selecting an L-M optimization algorithm as a weight training algorithm to train for 5,000 times of iteration, and testing a trained neural network by using a test sample; S4: using a heuristic intelligent optimization algorithm as an optimization main program, using an output ofthe trained neural network in the main program as a fitness function of the heuristic intelligent optimization algorithm and as a basis for data evolution and iteration operated in the program, using a particle swarm optimization (PSO) algorithm with smaller calculation amount as a main algorithm for a load fault diagnosis algorithm by the heuristic intelligent optimization algorithm; and according to each value representing one fault diagnosis data of the unit, inputting the unit into the trained neural network corresponding to each unit by the algorithm, obtaining fault diagnosis data of the desulfurization and denitration circulating pump corresponding to each unit through neural network calculation, and adding up the fault diagnosis data to obtain fault diagnosis data of all units in the plant; and S5: continuously updating the operation parameters acquired by the desulfurization and denitration circulating pump and an updating speed according to the PS0 algorithm principle, continuously updating and calculating the fault diagnosis data of the unit until the fault diagnosis data do not change anymore (i.e., the algorithm is considered to be convergent), outputting the fault data of each unit corresponding to the fault diagnosis data as an actual instruction of each unit in the power plant meeting the fault diagnosis, terminating the algorithm, and transmitting the output back to the distributed pump unit monitoring system (5) through the output module (11) and the wireless transmission module (12).

4. The method used for the system for diagnosing the state fault of a desulfurization and denitration circulating pump according to claim 3, characterized in that the system is determined to have entered the steady state when the operation parameters acquired by the desulfurization and denitration circulating pump are smaller than a specified threshold value within 15 minutes, i.e., in a steady-state working condition; the parameter data are stored in a steady-motion working condition database, othen/vise, the next 5 more minutes are required p.9 10 to determine whether the system has entered the steady state again until the data meet the steady state-requirement.

5. The method used for the system for diagnosing the state fault of a desulfurization and denitration circulating pump according to claim 3, characterized in that a process for dividing the input data into the training data and the test data of the network comprises conducting continuous differentiation on the operation parameters acquired by the desulfurization and denitration circulating pump, employing supervised-learning generalized radial basis function for the network to use both a center position and weights, selecting 75% of the total data randomly as training samples for the training of the BP neural network server (8), and taking the rest 25% samples as the network test samples. p.10