Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application more apparent, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that, in the embodiments of the present application, features in the embodiments and the examples may be arbitrarily combined with each other without conflict.
Fig. 1 is a flowchart of a method for monitoring a feedwater pump in a thermal power plant according to an embodiment of the present disclosure. As shown in fig. 1, the method shown in fig. 1 includes:
step 101, acquiring real-time flow information flowing through a water supply pump;
in an exemplary embodiment, real-time flow information of the feedwater pump may be obtained by interfacing with a Distributed Control System (DCS) for controlling the thermal power plant.
In an exemplary embodiment, flow information through the feedwater pump may be determined to determine performance information affecting the feedwater pump, according to the equipment configuration in the thermal power plant.
Step 102, determining the prediction information of the running state of the water feeding pump corresponding to the real-time flow information by using a preset neural network model;
in one exemplary embodiment, a neural network model is utilized to determine predictive information of the operating condition of the feedwater pump, providing a data reference value for online real-time diagnosis of leaks in the feedwater pump recirculation valve.
103, detecting internal leakage of a recirculation valve of the water feed pump according to the prediction information of the running state of the water feed pump and a real-time value of the running state of the water feed pump;
in one exemplary embodiment, the leakage in the recirculation valve of the feed water pump is detected according to the prediction information of the running state of the feed water pump and the real-time value of the running state of the feed water pump, so that online real-time diagnosis of the leakage in the recirculation valve of the feed water pump is realized.
The method provided by the embodiment of the application determines the prediction information of the running state of the water feeding pump corresponding to the real-time flow information by acquiring the real-time flow information of the water feeding pump and utilizing the preset neural network model, and detects the internal leakage of the recirculation valve of the water feeding pump according to the prediction information of the running state of the water feeding pump and the real-time value of the running state of the water feeding pump, so that the online real-time diagnosis of the internal leakage of the recirculation valve of the water feeding pump is realized, the diagnosis timeliness is improved, and the stability and the safety of the running of a water feeding system and a unit are effectively ensured.
The method provided by the embodiments of the present application is explained as follows:
and when parameters representing the leakage fault in the recirculation valve of the feed pump deviate from the output of a neural network model and trigger the logic fault tree in the feed water system, directly sending a leakage alarm signal in the recirculation valve of the feed pump to an operator through the DCS, and realizing the online real-time monitoring and diagnosis of the leakage fault in the recirculation valve of the feed pump.
In one exemplary embodiment, the real-time flow information includes at least one of economizer inlet flow, desuperheater flow, and desuperheater flow;
the operation state of the feed water pump includes at least one of a feed water pump rotational speed, a feed water pump inlet flow rate, and a feed water pump current.
And the real-time flow information and the running state of the water feeding pump are used as input parameters and output parameters of a neural network model, so that the monitoring operation is completed by the neural network model.
In one exemplary embodiment, the neural network model is built using historical data of the feedwater pump under all operating conditions.
Historical data of the water supply system can be derived from the DCS, the derived historical data can realize the full-working-condition coverage of the unit, and the data traversal performance is guaranteed.
In an exemplary embodiment, the detecting of the water supply pump recirculation valve internal leakage based on the predicted information of the operation state of the water supply pump and the real-time value of the operation state of the water supply pump includes:
taking the prediction information of the running state of the water-feeding pump and the real-time value of the running state of the water-feeding pump as input parameters of a preset fault tree to obtain an output result of the fault tree, wherein an alarm condition of the running state is recorded in the fault tree;
and when the output result of the fault tree is that the water feeding pump recirculation valve has internal leakage, outputting alarm information.
The internal leakage logic fault tree of the recirculation valve of the feed pump is constructed based on the deep neural network and expert knowledge, the internal leakage diagnosis of the recirculation valve of the feed pump is realized, and the diagnosis accuracy is improved.
In an exemplary embodiment, the alarm condition of the operation state is obtained by:
recording the prediction result of the neural network model within a preset time length;
and determining threshold information used by the alarm condition of the running state according to the mean value and/or mean square error information of the prediction result.
Threshold information is determined based on a prediction result of the neural network model, so that the accuracy of threshold setting can be effectively improved, the current working environment can be matched, and the accuracy of alarming is improved.
In an exemplary embodiment, the determination condition recorded in the fault tree is determined by determining that:
determining the fault phenomenon of internal leakage of a recirculation valve of a feed pump from a preset knowledge base;
and determining a judgment condition for describing the fault phenomenon according to the fault phenomenon.
Based on the fault phenomenon occurring when the fault occurs in the knowledge base, the change information of the operation state when the fault occurs is further deduced, so that the judgment condition for describing the fault phenomenon is determined according to the change information, and the accuracy of the judgment condition used by the fault tree is improved.
In an exemplary embodiment, the fault tree further comprises at least one of the following conditions:
the water feeding pump is in a running state;
the feed pump has no high pressure leakage fault;
wherein, when all conditions of the fault tree are yes, it is determined that there is an internal leak in the feedwater pump recirculation valve.
The inner leakage detection can be more accurately executed through the judgment condition, and the monitoring accuracy is improved.
In one exemplary embodiment, when the operation state of the feed water pump includes a feed water pump rotation speed, a feed water pump inlet flow rate, and a feed water pump current, and the determination conditions of the operation state are yes when the determination results of two of the feed water pump rotation speed, the feed water pump inlet flow rate, and the feed water pump current are yes, the determination conditions of the operation state are determined to be yes.
And when two judgment results in the 3 judgment results are yes, determining that the judgment condition of the running state is yes, so as to realize the purpose of finding the potential risk as soon as possible.
The method provided by the embodiments of the present application is explained as follows:
aiming at the defects existing in the prior art in the internal leakage detection of the water feeding pump recirculation valve of the thermal generator set, the technical problem to be solved by the embodiment of the application is to provide an online monitoring and diagnosis method capable of realizing the internal leakage of the water feeding pump recirculation valve on a DCS (distributed control system) of the thermal generator set.
FIG. 2 is a flow chart of a method for monitoring internal leakage of a recirculation valve of a feedwater pump of a thermal power plant according to an embodiment of the present disclosure. As shown in fig. 2, the method shown in fig. 2 adds a computing platform server to a DCS system for controlling a generator set in the related art; the computing platform server is connected with a real-time control network of the DCS, and training of a neural network and building of a model are achieved by providing a deep neural network algorithm. The method shown in fig. 2 comprises the following steps:
1. collecting DCS historical data;
historical data of a water supply system in the DCS are exported on a computing platform, the exported historical data can realize the full-working-condition coverage of the unit, and the data traversal is ensured.
2. Preprocessing the acquired data;
and performing preprocessing operations such as outlier rejection and missing value filling in the acquired data.
Detecting outliers by adopting a normality test analysis method, and then performing linear regression processing on the outliers, namely replacing samples with serious data deviation from the total through an approximate function fitted by normal data; and filling missing data in a cubic spline interpolation mode.
3. Determining input parameters and output parameters of the model;
the output parameters are determined according to the result of fault analysis, wherein the fault analysis means that the fault phenomenon of the internal leakage of the recirculation valve of the feed water pump is determined, and the specific steps are as follows:
the inventor finds that when the recirculation valve of the feed water pump leaks, the following fault phenomena exist, including:
(1) under the same unit load, the rotating speed of the water feeding pump is higher than that under the normal working condition;
(2) under the same unit load, the inlet flow of the feed pump is larger than that under the normal working condition.
(3) Under the same unit load, the current of the water supply pump is larger than that under the normal working condition.
And establishing a water feeding pump fault knowledge base according to the water feeding pump internal leakage fault phenomenon, and performing neural network training on the water feeding pump inlet flow, the water feeding pump rotating speed and the water feeding pump current to establish a neural network model.
Before the neural network model is established, input and output variables of the model are determined according to actual conditions. When determining the input parameters and the output parameters, the rotating speed of the water feeding pump, the inlet flow rate of the water feeding pump and the current of the water feeding pump are determined as judgment basis of the water feeding pump in the analysis, so the three variables are used as output variables of the neural network training. The input variables select the economizer inlet flow, the superheating and temperature-reducing water flow and the reheating and temperature-reducing water flow.
4. Training a parameter model;
and training the deep neural network model by using the preprocessed data, adjusting parameters such as the number of hidden layers, the number of nodes, weight, learning rate, activation function and the like of the deep neural network model according to the number of input/output variables of the system and the complexity of the system, and repeatedly training until the precision of the deep neural network prediction model meets the requirement.
5. Establishing a water supply system model;
and (3) building a neural network prediction model on a computing platform by using the weight parameters of the neural network obtained after training, taking the real-time values of the input variables of the system as the input data of the neural network prediction model, predicting the output values of the system in real time by using the neural network prediction model, and setting the threshold value of parameter alarm based on the mean value and mean square deviation of the monitored parameters in a certain time period.
FIG. 3 is a schematic diagram of a water supply system model provided in an embodiment of the present application. As shown in fig. 3, the economizer inlet flow, the superheating and temperature-reducing water flow, and the reheating and temperature-reducing water flow collected in real time are used as inputs, and the water pump rotation speed, the water feed pump inlet flow, and the water feed pump current are predicted by using an audit network model.
6. Establishing a fault knowledge base;
and (3) establishing a fault logic knowledge base in the DPU according to the parameters of the neural network early warning and the system operation condition, and monitoring the internal leakage of the recirculation valve of the feed pump through the fault logic knowledge base.
FIG. 4 is a schematic diagram of a leakage logic fault tree in a recirculation valve of a feedwater pump according to an embodiment of the present disclosure. As shown in fig. 4, the judgment basis of the internal leakage of the recirculation valve of the feed pump a is 3, and the internal leakage of the recirculation valve of the feed pump a is determined on the premise that all the 3 conditions are yes; wherein the 3 conditions include:
condition 1: a, the feed pump is in a running state;
condition 2: the feed pump has no high pressure leakage fault;
condition 3: the judgment result is determined to be yes when 2 judgment factors in the 3 judgment factors are satisfied, wherein the 3 judgment factors comprise:
a. a, the current of the water supply pump is larger than a current predicted value;
b. the inlet flow of the water supply pump is larger than the inlet flow predicted value;
b. a, the rotating speed of the water feeding pump is greater than a rotating speed predicted value;
7. and executing the alarm operation of the leakage in the recirculation valve of the feed water pump.
According to the method provided by the embodiment of the application, online real-time diagnosis of the inner leakage of the recirculation valve of the feed pump is realized in the DCS, the diagnosis timeliness is improved, the stability and the safety of the operation of the feed system and the unit are effectively guaranteed, the logical fault tree of the inner leakage of the recirculation valve of the feed pump is constructed based on the deep neural network and expert knowledge, the inner leakage diagnosis of the recirculation valve of the feed pump is realized, and the diagnosis accuracy is improved.
An embodiment of the present application provides a storage medium, in which a computer program is stored, wherein the computer program is configured to perform the method described in any one of the above when the computer program runs.
An embodiment of the application provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the computer program to perform the method described in any one of the above.
It will be understood by those of ordinary skill in the art that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.