Modeling method of coal power unit fault early warning model based on multi-element asynchronous working condition

Abstract

The invention relates to a modeling method of a coal power unit fault early warning model based on a multivariate asynchronous working condition. The method comprises the following steps: selecting equipment characteristic parameters and constructing a historical database; dividing multiple asynchronous working conditions; constructing a multi-working-condition early warning model and calculating an alarm value; and real-time early warning of equipment operation is realized. The multi-element asynchronous working condition division method has the beneficial effects that the working conditions of related equipment of the coal power unit can be divided more finely, the accuracy of working condition division is improved through multi-element features, and the fineness of working condition division is improved through asynchronous operation.

Description

Technical Field

[0001] This invention belongs to the field of power generation information technology, and in particular relates to a modeling method for fault early warning models of coal-fired power units based on multiple asynchronous operating conditions. Background Technology

[0002] With the increasing proportion of new energy power generation in total social power generation, coal-fired power units play a crucial role in peak shaving for power supply. Therefore, the long-term safe and stable operation of coal-fired power units is of paramount importance. Consequently, accurate early warning technology for coal-fired power unit operational faults is becoming increasingly crucial. However, current early warning modeling techniques for coal-fired power unit operational faults suffer from two main problems: 1) using a single early warning model to model data across all operating conditions without further subdivision, resulting in poor generalization performance for many conditions; 2) modeling across all operating conditions leads to large model sizes, slow prediction speeds, and an inability to fully meet actual production needs. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a modeling method for fault early warning models of coal-fired power units based on multiple asynchronous operating conditions.

[0004] Firstly, a modeling method for fault early warning models of coal-fired power units based on multiple asynchronous operating conditions is provided, including:

[0005] Step 1: Select equipment characteristic parameters and build a historical database;

[0006] Step 2: Divide the work into multiple asynchronous working conditions;

[0007] Step 3: Construct a multi-condition early warning model and calculate the warning value;

[0008] Step 4: Real-time early warning of equipment operation.

[0009] Preferably, step 1 includes:

[0010] Step 1.1: Select the characteristics of the coal-fired power unit equipment;

[0011] Step 1.2: Utilize the characteristics of the coal-fired power unit equipment to select equipment operating condition characteristics;

[0012] Step 1.3: Collect historical values ​​of equipment characteristics using the aforementioned coal-fired power unit equipment characteristics;

[0013] Step 1.4: Use the historical values ​​of the device features to remove outliers.

[0014] Preferably, step 2 includes:

[0015] Step 2.1: Select the target operating condition feature and its corresponding historical value of the equipment feature in the coal-fired power unit equipment characteristics;

[0016] Step 2.2: Determine the number of divisions for the selected working conditions;

[0017] Step 2.3: Perform iterative data condition division to obtain preliminary working condition division results;

[0018] Step 2.4: Based on the target working condition characteristics and the preliminary working condition classification results, repeat steps 2.1 to 2.3 for each of the remaining working condition characteristics within the already classified working conditions to obtain the multi-variable asynchronous working condition classification results.

[0019] Preferably, step 3 includes:

[0020] Step 3.1: Select any working condition from the multi-asynchronous working condition division results, and perform dimensional processing on the historical values ​​of equipment characteristics corresponding to the selected working condition;

[0021] Step 3.2: Construct a single-condition early warning model based on the historical values ​​of equipment characteristics after dimensional processing;

[0022] Step 3.3: Calculate the single-condition warning value based on the single-condition early warning model.

[0023] Step 3.4: Repeat steps 3.1 to 3.3 until all operating conditions are traversed to obtain the early warning model and its warning value corresponding to each operating condition.

[0024] Preferably, step 4 includes:

[0025] Step 4.1: Obtain real-time operating characteristic data of the equipment;

[0026] Step 4.2: Based on the real-time operating characteristic data of the equipment and the multi-asynchronous operating condition classification results, determine the current operating condition of the equipment;

[0027] Step 4.3: Combine the multi-condition early warning model and its warning value to make fault early warning judgment.

[0028] As a preferred option, in step 2.3, the formula for dividing the working conditions is:

[0029]

[0030] Where E represents the operating condition C. m Below, x is the sum of the distances of all data from their physical centers. j For feature data X ai Under operating condition C m The following data, m j For operating condition C m The physical center of , m≤K.

[0031] Preferably, in step 3.1, the dimensionality processing formula is:

[0032]

[0033] Among them, Xs (j) i (m) represents the current operating condition characteristic C′ j Below, the m-th value of the i-th feature after dimensional processing; X (j) i (max) and X (j) i (min) represents the current operating condition characteristic C′ j The maximum and minimum values ​​of the i-th feature.

[0034] Secondly, a fault early warning modeling system for coal-fired power units based on multiple asynchronous operating conditions is provided, for executing any of the methods described in the first aspect, including:

[0035] The selection module is used to select device characteristic parameters and build a historical database;

[0036] The partitioning module is used to partition multiple asynchronous operating conditions;

[0037] The building module is used to construct a multi-condition early warning model and calculate the warning value;

[0038] The early warning module is used for real-time early warning of equipment operation.

[0039] Thirdly, a computer storage medium is provided, wherein a computer program is stored therein; when the computer program is run on a computer, the computer causes the computer to perform any of the methods described in the first aspect.

[0040] Fourthly, an electronic device is provided, comprising:

[0041] Memory, used to store computer programs;

[0042] A processor for executing the computer program to implement the method as described in any of the first aspects.

[0043] The beneficial effects of this invention are:

[0044] 1. This invention proposes a multi-asynchronous operating condition classification method, which can more precisely classify the operating conditions of relevant equipment in coal-fired power units. It improves the accuracy of operating condition classification through multi-features and enhances the precision of operating condition classification through asynchronous operation.

[0045] 2. This invention performs fault early warning modeling for different operating conditions. Unlike traditional fault modeling methods, the early warning model of this invention only needs to focus on the fault under the current operating condition, which greatly improves the accuracy of model prediction.

[0046] 3. Due to the detailed division of working conditions in this invention, different models only learn from a small portion of the operating data, resulting in smaller model sizes, improved prediction speed, and the ability to quickly provide equipment early warning information. Attached Figure Description

[0047] Figure 1 This is a flowchart of the modeling method for fault early warning model of coal-fired power units based on multiple asynchronous operating conditions provided by the present invention;

[0048] Figure 2 This is a schematic diagram of the multi-asynchronous working condition division method provided by the present invention;

[0049] Figure 3 This is a schematic diagram of the multi-asynchronous working condition division principle provided by the present invention;

[0050] Figure 4 This is a schematic diagram comparing the old and new models provided by this invention for predicting vacuum in a high-pressure condenser. Figure 4 The red curve represents the prediction of the new model, the blue curve represents the prediction of the old model, and the black curve represents the actual value.

[0051] Figure 5 This is a schematic diagram comparing the prediction of the outlet temperature of the feature high-pressure heater using the new and old models provided by this invention; Figure 5 The red curve represents the prediction of the new model, the blue curve represents the prediction of the old model, and the black curve represents the actual value. Detailed Implementation

[0052] The present invention will be further described below with reference to embodiments. The description of the embodiments below is only for the purpose of helping to understand the present invention. It should be noted that those skilled in the art can make several modifications to the present invention without departing from the principle of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0053] Example 1:

[0054] To address the problems of existing technologies, Embodiment 1 of this application summarizes the experience of coal-fired power unit equipment operation, identifies the characteristics that can reflect equipment operation and the operating conditions of the equipment, and proposes a multi-dimensional asynchronous operating condition classification method based on the operating condition characteristics. Taking a data standardization project of a certain group as an example, early warning modeling and warning value extraction are performed for different operating conditions under the subdivided operating conditions, and the method is applied to the real-time fault early warning of coal-fired power unit equipment operation.

[0055] Specifically, such as Figure 1 As shown, the modeling method for fault early warning model of coal-fired power units based on multiple asynchronous operating conditions provided by the present invention includes:

[0056] Step 1: Select equipment characteristic parameters and build a historical database.

[0057] Step 1 includes:

[0058] Step 1.1: Select the characteristics of coal-fired power unit equipment.

[0059] The coal-fired power unit equipment characteristics mentioned in step 1.1 should be clearly defined as those that can acquire data over a long period of time, and should take into account as many related equipment characteristics as possible, including the equipment's own sensor characteristics, external sensor characteristics, and related characteristics of associated equipment.

[0060] For example, based on the operating characteristics of the relevant equipment in the coal-fired power unit and the actual installation status of the relevant measurement sensors, n features T1, T2, T3, ..., T of the coal-fired power unit equipment to be used for diagnostic modeling are selected. n .

[0061] For example, taking the high-pressure heater equipment of the steam extraction and condensate drainage system of a coal-fired power unit as an example, the relevant characteristics selected are shown in Table 1 below:

[0062] Table 1

[0063]

[0064]

[0065] Step 1.2: Select equipment operating condition characteristics using the characteristics of the coal-fired power unit.

[0066] The equipment operating condition characteristics mentioned in step 1.2 should be the macroscopic characteristics of the environment in which the equipment is located and the direct characteristics of the equipment's own state, such as atmospheric pressure, ambient temperature, and the equipment's current.

[0067] For example, using the coal-fired power unit equipment characteristics selected in step 1.1, select a features T that can reflect different operating conditions of the equipment. a1 ,T a2 ,T a3 ,…,T aa a ≤ n. For example, a = 3.

[0068] Step 1.3: Collect historical values ​​of equipment characteristics using the aforementioned coal-fired power unit equipment characteristics.

[0069] For example, using the equipment characteristics of the coal-fired power unit selected in step 1.1, and taking a time interval of 10 minutes, a historical operating condition dataset of the unit of size m*n (m=52562, n=32) is formed, X={x1,x2,x3,…,x n}, x i Description feature T i Historical time series data.

[0070] Step 1.4: Use the historical values ​​of the device features to remove outliers.

[0071] For example, outlier removal is performed using the historical device feature values ​​obtained in step 1.3, resulting in the outlier-removed data {X1,X2,X3,…,X...}. n}

[0072] The outlier removal method adopted is a density clustering-based outlier removal method.

[0073] Step 2: Divide into multiple asynchronous working conditions.

[0074] Step 2 includes:

[0075] Step 2.1: Select the target operating condition feature and its corresponding historical value of equipment feature from the equipment features of the coal-fired power unit.

[0076] For example, the a equipment operating condition characteristics T selected in step 1.1 a1 ,T a2 ,T a3 ,…,T aa The data {X1,X2,X3,…,X} obtained after outlier removal in step 1.4 n} Select the feature T for this cycle to divide the working conditions. ai and its corresponding data X ai .

[0077] Step 2.2: Determine the number of divisions for the selected working conditions.

[0078] For example, the feature T of the working condition division obtained from step 2.1 ai and its corresponding data X ai The process involves determining the number of divisions, K, for the current work condition (e.g., K = 10). To ensure accuracy in work condition division, the number K should be determined based on the characteristics T of the currently selected work condition divisions in the loop. ai and its corresponding data X ai The curve characteristics and classification based on professional experience.

[0079] Step 2.3: Perform iterative division of data working conditions and obtain preliminary working condition division results.

[0080] In step 2.3, the formula for dividing the working conditions is:

[0081]

[0082] Where, x j For feature data X ai Under operating condition C m The following data, m jFor operating condition C m The physical center of , m≤K.

[0083] For example, using the feature data X for the working condition division obtained in step 2.1 ai Following step 2.2, determine the number K of selected working conditions for this iteration, perform iterative data working condition partitioning, and ensure that the partitioning result satisfies the minimum E, thus obtaining the feature T of the working condition partition. ai The result of the K working condition partitioning {C1,C2,C3,…,C k}

[0084] Step 2.4: Based on the target working condition characteristics and the preliminary working condition classification results, repeat steps 2.1 to 2.3 for each of the remaining working condition characteristics within the already classified working conditions to obtain the multi-variable asynchronous working condition classification results.

[0085] For example, the feature T selected in step 2.1 ai The working condition division results obtained from step 2.3 are {C1,C2,C3,…,C…} k Within the already defined operating conditions, repeat steps 2.1 to 2.3 for each of the remaining operating condition characteristics (including the operating condition defined by the inlet flow rate of pump A and the inlet flow rate of pump B) to obtain the multi-element asynchronous operating condition division result {C′1,C′2,C′3,…,C′}. k* The value of k* is the product of the number of working conditions divided by all working condition characteristics. For example, k* = 1000.

[0086] like Figure 3 As shown, each additional working condition segmentation feature requires one more asynchronous working condition segmentation operation. Each asynchronous working condition segmentation is performed based on the previous segmentation result. As the number of segmentations increases, the number of segmented working conditions will also continue to increase.

[0087] Step 3: Construct a multi-condition early warning model and calculate the warning value.

[0088] Step 4: Real-time early warning of equipment operation.

[0089] Example 2:

[0090] Based on Example 1, Example 2 of this application provides a more specific modeling method for fault early warning models of coal-fired power units based on multiple asynchronous operating conditions, including:

[0091] Step 1: Select equipment characteristic parameters and build a historical database.

[0092] Step 2: Divide into multiple asynchronous working conditions.

[0093] Step 3: Construct a multi-condition early warning model and calculate the warning value.

[0094] Step 3 includes:

[0095] Step 3.1: Select any working condition from the multi-asynchronous working condition division results, and perform dimensional processing on the historical values ​​of equipment characteristics corresponding to the selected working condition.

[0096] For example, the working condition division result obtained from step 2.3 is {C′1,C′2,C′3,…,C′ k* Combining step 1.4, we obtain the data {X1, X2, X3, ..., X} after outlier removal. n}, select a working condition C′ j For the data {X} corresponding to the current working condition (j) 1,X (j) 2,X (j) 3,…,X (j) n}∈{X1,X2,X3,…,X n} Perform dimensional processing to obtain the corresponding working condition data {Xs} after dimensional processing. (j) 1,Xs (j) 2,Xs (j) 3,…,Xs (j) n}

[0097] In step 3.1, the dimensionality processing formula is:

[0098]

[0099] Among them, Xs (j) i (m) represents the current operating condition characteristic C′ j Below, the m-th value of the i-th feature after dimensional processing; X (j) i (max) and X (j) i (min) represents the current operating condition characteristic C′ j The maximum and minimum values ​​of the i-th feature.

[0100] Step 3.2: Construct a single-condition early warning model based on the historical values ​​of equipment characteristics after dimensional processing.

[0101] For example, the operating condition C′ obtained from step 3.1 j The corresponding working condition data after dimensional processing {Xs (j) 1,Xs (j) 2,Xs (j) 3,…,Xs (j) n}, for the feature T that requires early warning n Data Xs (j) nModel building:

[0102]

[0103] in, λ is a correction factor, which can be configured within a range greater than 0 depending on the actual situation.

[0104] Step 3.3: Calculate the single-condition warning value based on the single-condition early warning model.

[0105] For example, the single-condition early warning model y constructed in step 3.2 k*＝j The corresponding working condition data {Xs} obtained in step 3.1 after dimensional processing (j) 1,Xs (j) 2,Xs (j) 3,…,Xs (j) n Substitute the values ​​into the early warning model and calculate the residuals:

[0106]

[0107] Single-condition warning value:

[0108]

[0109] μ is the weight value, which can be adjusted according to the actual use case.

[0110] Step 3.4: Repeat steps 3.1 to 3.3 until all operating conditions are traversed to obtain the early warning model and its warning value corresponding to each operating condition.

[0111] For example, iterate through all working conditions {C′1,C′2,C′3,…,C′ k For each operating condition, steps 3.1 to 3.3 are executed to obtain the early warning model and its warning value corresponding to each operating condition, which are then combined to obtain the early warning feature T. n Multi-condition combined early warning model and combinations of warning values ​​for multi-condition early warning models

[0112] Step 4: Real-time early warning of equipment operation.

[0113] Step 4 includes:

[0114] Step 4.1: Obtain real-time operating characteristic data of the equipment.

[0115] For example, based on the equipment characteristics of the coal-fired power unit selected in step 1.1, real-time characteristic data of the equipment are obtained, forming a real-time characteristic dataset Z = {Z1, Z2, Z3, ..., Zn} of size 1*n. n}

[0116] Step 4.2: Based on the real-time operating characteristic data of the equipment and the multi-asynchronous operating condition classification results, determine the current operating condition of the equipment.

[0117] For example, the real-time feature dataset Z = {Z1, Z2, Z3, ..., Zn} of size 1*n obtained from step 4.1 is... n The result of the multi-factor asynchronous working condition partitioning obtained in step 2.4 is {C′1,C′2,C′3,…,C′}. k*} Determine the current operating condition of the equipment, which is condition C′. j .

[0118] Step 4.3: Combine the multi-condition early warning model and its warning value to make fault early warning judgment.

[0119] For example, the current equipment operating condition C′ obtained from step 4.2 belongs to the operating condition C′. j For the real-time feature dataset Z = {Z1, Z2, Z3, ..., Zn} of size 1*n obtained in step 4.1 n The dimensions are processed to obtain the real-time feature data Z after dimension processing. norm ={Z′1,Z′2,Z′3,…,Z′ n}, combining the multi-condition early warning model and warning value combination obtained in step 3.4, select the early warning model f corresponding to the condition. j Real-time data prediction of each device and its characteristics is performed to obtain the predicted value Z. est ={Z*1,Z*2,Z*3,…,Z* n}. Using the formula Δ=|Z norm -Z est | and the corresponding operating condition threshold T obtained in step 3.4 j When Δ > T j When this occurs, it is determined to be a device malfunction, and an early warning is issued.

[0120] The dimensional treatment formula in step 4.3 under operating condition C′j is as follows:

[0121] It should be noted that the parts in this embodiment that are the same as or similar to those in Embodiment 1 can be referred to each other, and will not be repeated in this application.

[0122] Example 3:

[0123] Based on Example 2, Example 3 of this application provides a modeling system for fault early warning of coal-fired power units based on multiple asynchronous operating conditions, including:

[0124] The selection module is used to select device characteristic parameters and build a historical database.

[0125] The partitioning module is used to partition multiple asynchronous operating conditions.

[0126] The module is used to build a multi-condition early warning model and calculate the warning value.

[0127] The early warning module is used for real-time early warning of equipment operation.

[0128] It should be noted that the system provided in this embodiment is the corresponding system of the method provided in embodiment 2. Therefore, the parts that are the same as or similar to those in embodiment 2 in this embodiment can be referred to each other, and will not be described again in this application.

Claims

1. A modeling method for a coal-fired unit fault early warning model based on multiple asynchronous working conditions, characterized in that, The method comprises the following steps: Step 1, selecting equipment characteristic parameters and constructing a historical database; Step 2, dividing multiple asynchronous working conditions; Step 3, constructing a multi-working condition early warning model and calculating an alarm value; Step 4, real-time early warning of equipment operation.

2. The multi-element asynchronous operating condition-based coal-fired generating unit fault early warning model modeling method according to claim 1, characterized in that, Step 1 comprises: Step 1.1, selecting coal-fired unit equipment characteristics; Step 1.2, selecting equipment working condition characteristics by using the coal-fired unit equipment characteristics; Step 1.3, collecting equipment characteristic historical values by using the coal-fired unit equipment characteristics; Step 1.4, removing outliers by using the equipment characteristic historical values.

3. The multi-element asynchronous operating condition-based coal-fired generating unit fault early warning model modeling method according to claim 2, characterized in that, Step 2 comprises: Step 2.1, selecting target working condition characteristics and corresponding equipment characteristic historical values from the coal-fired unit equipment characteristics; Step 2.2, determining the number of divided working conditions; Step 2.3, performing data working condition iterative division to obtain a preliminary working condition division result; Step 2.4, from the target working condition characteristics and the preliminary working condition division result, repeatedly performing steps 2.1 to 2.3 on the remaining working condition characteristics in the divided working conditions to obtain a multiple asynchronous working condition division result.

4. The multi-element asynchronous operating condition-based coal-fired generating unit fault early warning model modeling method according to claim 3, characterized in that, Step 3 comprises: Step 3.1, selecting any working condition in the multiple asynchronous working condition division result and performing dimension processing on the equipment characteristic historical values corresponding to the selected working condition; Step 3.2, constructing a single-working condition early warning model according to the dimension-processed equipment characteristic historical values; Step 3.3, calculating a single-working condition alarm value according to the single-working condition early warning model; Step 3.4, repeating steps 3.1 to 3.3 until all working conditions are traversed to obtain the early warning model and the alarm value corresponding to each working condition.

5. The multi-element asynchronous operating condition-based coal-fired generating unit fault early warning model modeling method according to claim 4, characterized in that, Step 4 comprises: Step 4.1, obtaining equipment real-time operation characteristic data; Step 4.2, determining the working condition to which the current equipment working condition belongs according to the equipment real-time operation characteristic data and the multiple asynchronous working condition division result; Step 4.3, combining the multi-working condition early warning model and the alarm value to perform fault early warning determination.

6. The multi-element asynchronous operating condition-based coal-fired generating unit fault early warning model modeling method according to claim 5, characterized in that, In step 2.3, the working condition division formula is: where E denotes the data under the condition C m The sum of distances of all data from their physical center, x j is the feature data X ai The data under the condition C m is the physical center of the condition C j is the physical center of the condition C m , m ≤ K.

7. The multi-element asynchronous operating condition-based coal-fired generating unit fault warning model modeling method of claim 5, wherein, In step 3.1, the dimension processing formula is: Among them, Xs (j) i (m) represents the current operating condition characteristic C′ j Below, the m-th value of the i-th feature after dimensional processing; X (j) i (max) and X (j) i (min) represents the current operating condition characteristic C′ j The maximum and minimum values ​​of the i-th feature.

8. A multi-asynchronous working condition based coal-fired generating unit fault early warning model modeling system, characterized in that, The computer storage medium stores a computer program; when the computer program runs on a computer, the computer executes the method of any one of claims 1 to 7. The computer storage medium stores a computer program; when the computer program runs on a computer, the computer executes the method of any one of claims 1 to 7. The computer storage medium stores a computer program; when the computer program runs on a computer, the computer executes the method of any one of claims 1 to 7. The computer storage medium stores a computer program; when the computer program runs on a computer, the computer executes the method of any one of claims 1 to 7. The computer storage medium stores a computer program; when the computer program runs on a computer, the computer executes the method of any one of claims 1 to 7.

9. A computer storage medium, characterized in that ​ 10. An electronic device, comprising: ​ ​ ​

Images