A generator set early warning diagnosis method based on association rule mining

Abstract

The present invention relates to a generator set early warning diagnosis method based on association rule mining, comprising: constructing a fault knowledge base based on unstructured data; constructing a parameter early warning model based on equipment characteristic parameters, issuing early warnings on characteristic parameters, and determining abnormal parameters; mining association rules between characteristic parameters and fault modes, and performing fault diagnosis on abnormal parameters; mining association rules between fault modes and diagnostic decisions, matching fault handling methods for abnormal parameters, and providing guidance for decision-making. The beneficial effect of the present invention is that the present invention realizes the automation of the entire process from monitoring and early warning to diagnosis to decision-making guidance, without the need for manual analysis in the middle, and can directly provide guidance opinions and push them to professionals for decision-making, which has important application value in assisting production.

Description

Technical Field

[0001] The present invention relates to the field of early warning diagnosis of generator sets, and more specifically, to an early warning diagnosis method for generator sets based on association rule mining. Background Art

[0002] With the continuous development of computer technology, it is now common to build early warning systems for power plant equipment to provide early warnings of equipment anomalies. Various data-driven methods are used to construct early warning models, such as similarity principle algorithms and neural network algorithms. Essentially, these methods construct a normal operating condition matrix to generate warnings for abnormal conditions. Furthermore, fault diagnosis techniques for power plant equipment are becoming increasingly diverse, with both mechanistic and mathematical models offering their own advantages. However, in the current technological landscape, early warning and diagnostic technologies are relatively isolated, with few integrated systems. Warning results require manual analysis, which is labor-intensive and inefficient. Furthermore, after fault diagnosis, the decision on how to proceed still requires manual analysis and judgment, a process that relies on human experience and is prone to incorrect decisions and misoperations. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a generator set early warning diagnosis method based on association rule mining.

[0004] In a first aspect, a generator set early warning diagnosis method based on association rule mining is provided, comprising:

[0005] Step 1: Build a fault knowledge base based on unstructured data; the unstructured data includes operating procedures, defect logs, alarm logs, and processing records;

[0006] Step 2: Build a parameter warning model based on the equipment characteristic parameters, issue warnings on the characteristic parameters, and identify abnormal parameters;

[0007] Step 3: Mining association rules between characteristic parameters and fault modes, and performing fault diagnosis on abnormal parameters;

[0008] Step 4: Mining the association rules between fault modes and diagnostic decisions, matching fault handling methods based on abnormal parameters, and providing guidance for decision making.

[0009] Preferably, in step 1, the fault knowledge base includes characteristic parameters, fault modes, and fault handling methods of each device.

[0010] For devices , and equipment The relevant failure mode set is denoted as , and equipment The relevant set of fault handling methods is recorded as , , For equipment the number of relevant failure modes; , For equipment The number of related fault handling methods.

[0011] Preferably, step 2 specifically includes the following steps:

[0012] Step 2.1: Remember the device The characteristic parameter time series of , For devices The number of characteristic parameters, is the sequence length, , is the measured value of the characteristic parameter;

[0013] Step 2.2: For the time series of characteristic parameters , based on a certain period of historical data, build and train parameter warning models;

[0014] Step 2.3: Use the parameter warning model constructed in step 2.2 to monitor the equipment Conduct early warning of characteristic parameters and identify abnormal parameters.

[0015] Preferably, step 3 specifically includes the following steps:

[0016] Step 3.1: Remember the device The abnormal characteristic parameters determined by step 2 are , , , For this device The number of abnormal characteristic parameters determined;

[0017] Step 3.2: Mining abnormal feature parameters through association rule mining algorithm and failure mode collection The association rules between the failure modes in the , .

[0018] Preferably, step 4 specifically includes the following steps:

[0019] Step 4.1: For the Corresponding failure mode , mining fault patterns through association rule mining algorithm Combined with fault handling methods The association rules of fault handling methods in the , get the association rule with the highest support , ;

[0020] Step 4.2: Combine the association rules in step 3 to obtain the complete association rules of early warning - diagnosis - decision making .

[0021] Preferably, the parameter warning model in step 2 is a model based on a similarity principle algorithm.

[0022] Preferably, the association rule mining algorithm in step 3 and step 4 is an apriori association rule mining algorithm.

[0023] In a second aspect, a generator set early warning diagnosis device based on association rule mining is provided, which performs any of the generator set early warning diagnosis methods based on association rule mining described in the first aspect, including:

[0024] A first building module is used to build a fault knowledge base based on unstructured data; the unstructured data includes operating procedures, defect logs, alarm logs and processing records;

[0025] The second building module is used to build a parameter early warning model based on the device characteristic parameters, issue early warnings on the characteristic parameters, and determine abnormal parameters;

[0026] The first mining module is used to mine the association rules between characteristic parameters and fault modes and perform fault diagnosis on abnormal parameters;

[0027] The second mining module is used to mine the association rules between fault modes and diagnostic decisions, match fault handling methods based on abnormal parameters, and provide guidance for decision-making.

[0028] In a third aspect, a computer storage medium is provided, in which a computer program is stored; when the computer program is run on a computer, the computer executes any of the generator set early warning diagnosis methods based on association rule mining described in the first aspect.

[0029] In a fourth aspect, a computer program product is provided. When the computer program product is run on a computer, the computer is enabled to execute any of the generator set early warning diagnosis methods based on association rule mining described in the first aspect.

[0030] The beneficial effects of the present invention are: the present invention realizes the automation of the entire process from monitoring and early warning to diagnosis to decision-making guidance, without the need for manual analysis in the middle, and can directly give guidance opinions and push them to professionals for decision-making, which has important application value for auxiliary production. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1This is a flow chart of a generator set early warning diagnosis method based on association rule mining;

[0032] Figure 2 This is a structural diagram of a generator set early warning and diagnosis device based on association rule mining;

[0033] Figure 3 This is a diagram of parameter warning results in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The present invention will be further described below with reference to the following examples. The following examples are provided only to facilitate understanding of the present invention. It should be noted that, without departing from the principles of the present invention, it is possible for a person skilled in the art to make various modifications to the present invention, and such improvements and modifications fall within the scope of the claims of the present invention.

[0035] Example 1:

[0036] A generator set early warning diagnosis method based on association rule mining, comprising:

[0037] Step 1: Build a fault knowledge base based on unstructured data; unstructured data includes operating procedures, defect logs, alarm logs, and processing records;

[0038] Step 2: Build a parameter warning model based on the equipment characteristic parameters, issue warnings on the characteristic parameters, and identify abnormal parameters;

[0039] Step 3: Mining association rules between characteristic parameters and fault modes, and performing fault diagnosis on abnormal parameters;

[0040] Step 4: Mining the association rules between fault modes and diagnostic decisions, matching fault handling methods based on abnormal parameters, and providing guidance for decision making.

[0041] In step 1, the fault knowledge base includes the characteristic parameters, fault modes and fault handling methods of each device.

[0042] For devices , and equipment The relevant failure mode set is denoted as , and equipment The relevant set of fault handling methods is recorded as , , For equipment the number of relevant failure modes; , for devices The number of related fault handling methods.

[0043] Step 2 specifically includes the following steps:

[0044] Step 2.1: Remember the device The characteristic parameter time series of , For devices The number of characteristic parameters, is the sequence length, , is the measured value of the characteristic parameter;

[0045] Step 2.2: For the time series of characteristic parameters , based on a certain period of historical data, build and train parameter warning models;

[0046] Step 2.3: Use the parameter warning model constructed in step 2.2 to monitor the equipment Conduct early warning of characteristic parameters and identify abnormal parameters.

[0047] Step 3 specifically includes the following steps:

[0048] Step 3.1: Remember the device The abnormal characteristic parameters determined by step 2 are , , , For this device The number of abnormal characteristic parameters determined;

[0049] Step 3.2: Mining abnormal feature parameters through association rule mining algorithm and failure mode collection The association rules between the failure modes in the , .

[0050] Step 4 specifically includes the following steps:

[0051] Step 4.1: For the Corresponding failure mode , mining fault patterns through association rule mining algorithm Combined with fault handling methods The association rules of fault handling methods in the , get the association rule with the highest support , ;

[0052] Step 4.2: Combine the association rules in step 3 to obtain the complete association rules of early warning - diagnosis - decision making .

[0053] The parameter warning model in step 2 is a model based on the similarity principle algorithm.

[0054] The association rule mining algorithm in steps 3 and 4 is the Apriori association rule mining algorithm.

[0055] Example 2:

[0056] As an example, the specific operation steps and the effectiveness of the verification method are illustrated using a real-world case study of a 1A coal mill at a thermal power plant. In this example, unstructured data was collected from January 2019 to December 2021. Feature parameter values ​​were collected from 00:00 on December 1, 2020, to 24:00 on December 31, 2021, with a sampling interval of 10 minutes.

[0057] The specific execution process of the generator set early warning diagnosis method based on association rule mining is as follows:

[0058] Step 1: Based on unstructured data such as operating procedures, defect logs, alarm logs, and processing records, a fault knowledge base for the 1A coal mill is constructed. The fault knowledge base contains the characteristic parameters, fault modes, and fault handling methods of the equipment for the 1A coal mill, including 69 characteristic parameters, 19 fault modes, and 168 handling methods.

[0059] For 1A coal mill, the set of fault modes related to 1A coal mill is recorded as, and the set of fault handling methods related to 1A coal mill is recorded as , , For equipment The number of relevant failure modes, ; , is the number of fault handling methods related to 1A coal mill, ;

[0060] Step 2: Build an early warning model based on the equipment characteristic parameters and issue early warnings on the parameters;

[0061] Step 2.1: Let the time series of characteristic parameters of 1A coal mill be , For devices The number of characteristic parameters, , is the sequence length, , is the measured value of the characteristic parameter;

[0062] Step 2.2: For the time series of characteristic parameters , based on one year of historical data (from 0:00 on December 1, 2020 to 24:00 on November 30, 2021, with a sampling interval of 10 minutes), a parameter early warning model was constructed and trained;

[0063] Step 2.3: Using the early warning model constructed in step 2.2, the data of 1A coal mill from 0:00 on December 1, 2021 to 24:00 on December 31, 2021 is used as test data to perform parameter early warning on 1A coal mill, and the early warning results are as follows: Figure 3 shown.

[0064] Step 3: Mining association rules between characteristic parameters and fault modes, and performing fault diagnosis on abnormal parameters;

[0065] Step 3.1: The abnormal characteristic parameters of the 1A coal mill that are warned in step 2 are lubricating oil temperature, lubricating oil pressure, and lubricating oil return temperature;

[0066] Step 3.2: By using the association rule mining algorithm, the association rules between the abnormal characteristic parameters lubricating oil temperature, lubricating oil pressure, and lubricating oil return temperature and 19 fault modes are mined. The association rule mining results are shown in Table 1. The association rule with the highest support is [lubricating oil temperature abnormality, lubricating oil pressure abnormality, lubricating oil return temperature abnormality] → [lubricating oil system abnormality].

[0067] Table 1

[0068]

[0069] Step 4: Mining association rules between fault modes and diagnostic decisions, matching fault handling methods based on abnormal parameters, and providing guidance for decision-making;

[0070] Step 4.1: For the fault mode [lubricating oil system abnormality] corresponding to [lubricating oil temperature abnormality, lubricating oil pressure abnormality, lubricating oil return temperature abnormality] obtained in step 3, the association rules between the fault mode [lubricating oil system abnormality] and 168 fault handling methods are mined using the association rule mining algorithm. As shown in Table 2, the association rule with the highest support is [lubricating oil system abnormality] → [cooler cleaning or replacement];

[0071] Table 2

[0072]

[0073] Step 4.2: Combined with the association rules in step 3, the complete association rules of early warning - diagnosis - decision-making are obtained [abnormal lubricating oil temperature, abnormal lubricating oil pressure, abnormal lubricating oil return temperature] → [abnormal lubricating oil system] → [cooler cleaning or replacement].

[0074] In summary, the present invention first constructs a knowledge base of failure modes and diagnostic decisions by organizing unstructured data such as power plant operating procedures, defect logs, alarm logs, and processing records. It then builds a parameter early warning model for equipment, providing early warning of abnormal equipment parameters. Finally, it constructs an association rule mining model based on characteristic parameters, failure modes, and diagnostic decisions, performing diagnostic reasoning on abnormal parameters and providing decision guidance. This invention automates the entire process, from monitoring and early warning to diagnosis and decision guidance, and has important application value in assisting production.

Claims

1. A generator set early warning diagnosis method based on association rule mining, characterized in that: include: Step 1: Build a fault knowledge base based on unstructured data; the unstructured data includes operating procedures, defect logs, alarm logs, and processing records; Step 2: Build a parameter warning model based on the equipment characteristic parameters, issue warnings on the characteristic parameters, and identify abnormal parameters; Step 3: Mining association rules between characteristic parameters and fault modes, and performing fault diagnosis on abnormal parameters; Step 3 specifically includes the following steps: Step 3.1: Remember the device The abnormal characteristic parameters determined by step 2 are , , , For devices The number of abnormal characteristic parameters determined; Step 3.2: Mining abnormal feature parameters through association rule mining algorithm and failure mode collection The association rules between the failure modes in the , ; Step 4: Mining association rules between fault modes and diagnostic decisions, matching fault handling methods with abnormal parameters, and providing guidance for decision making. Step 4 specifically includes the following steps: Step 4.1: For the Corresponding failure mode , mining fault patterns through association rule mining algorithm Combined with fault handling methods The association rules of fault handling methods in the , get the association rule with the highest support , ; Step 4.2: Combine the association rules in step 3 to obtain the complete association rules of early warning - diagnosis - decision making .

2. The generator set early warning diagnosis method based on association rule mining according to claim 1 is characterized in that: In step 1, the fault knowledge base includes the characteristic parameters, fault modes and fault handling methods of each device. For devices , and equipment The relevant failure mode set is denoted as , and equipment The relevant set of fault handling methods is recorded as , , For equipment the number of relevant failure modes; , For equipment The number of related fault handling methods.

3. The generator set early warning diagnosis method based on association rule mining according to claim 2 is characterized in that: Step 2 specifically includes the following steps: Step 2.1: Remember the device The characteristic parameter time series of , For devices The number of characteristic parameters, is the sequence length, , is the measured value of the characteristic parameter; Step 2.2: For the time series of characteristic parameters , based on a certain period of historical data, build and train parameter warning models; Step 2.3: Use the parameter warning model constructed in step 2.2 to monitor the equipment Conduct early warning of characteristic parameters and identify abnormal parameters.

4. The generator set early warning diagnosis method based on association rule mining according to claim 1 is characterized in that: The parameter warning model in step 2 is a model based on the similarity principle algorithm.

5. The generator set early warning diagnosis method based on association rule mining according to claim 1 is characterized in that: The association rule mining algorithm in step 3 and step 4 is a priori association rule mining algorithm.

6. A generator set early warning diagnosis device based on association rule mining, characterized in that: The method for early warning diagnosis of a generator set based on association rule mining according to any one of claims 1 to 5 comprises: A first building module is used to build a fault knowledge base based on unstructured data; the unstructured data includes operating procedures, defect logs, alarm logs and processing records; The second building module is used to build a parameter early warning model based on the device characteristic parameters, issue early warnings on the characteristic parameters, and determine abnormal parameters; The first mining module is used to mine the association rules between characteristic parameters and fault modes and perform fault diagnosis on abnormal parameters; The second mining module is used to mine the association rules between fault modes and diagnostic decisions, match fault handling methods based on abnormal parameters, and provide guidance for decision-making.

7. A computer storage medium, characterized in that The computer storage medium stores a computer program; when the computer program is run on a computer, the computer executes the generator set early warning diagnosis method based on association rule mining according to any one of claims 1 to 5.

8. A computer program product, characterized in that When the computer program product is run on a computer, the computer is enabled to execute the generator set early warning diagnosis method based on association rule mining according to any one of claims 1 to 5.

Images