Coal quality characteristic extraction method for coal source of coal-fired power plant

Abstract

The application discloses a coal quality characteristic extraction method of a coal source of a coal-fired power plant. The coal quality characteristic extraction method comprises the following steps: establishing a statistical database of original coal quality, including parameter groups of a plurality of coal types, and each parameter group of a same-named coal type including a plurality of batch sub-parameter groups; checking whether the quality consistency of the same-named coal type meets a threshold condition according to the plurality of batch sub-parameter groups of each same-named coal type; for the same-named coal type whose quality consistency meets the threshold condition, performing weighted average processing on the plurality of batch sub-parameter groups of the same-named coal type to obtain a parameter group of a characteristic coal; for the same-named coal type whose quality consistency does not meet the threshold condition, performing clustering subdivision processing on the plurality of batch sub-parameter groups of the same-named coal type to obtain a parameter group of a characteristic coal; and screening all the parameter groups of the characteristic coals according to a threshold parameter to obtain a candidate coal type set. The method can effectively extract characteristic coals with stable quality and safe reliability, and eliminates coal types with large fluctuations or dangers, thereby providing a high-quality data basis for precise coal blending.

Description

Technical Field

[0001] This invention relates to the field of fuel management and data mining technology for coal-fired power plants, specifically to a method for extracting coal quality characteristics from coal sources in coal-fired power plants. Background Technology

[0002] The quality of coal delivered to coal-fired power plants is a core factor determining the safe, stable operation and economical, efficient power generation of boiler units. In recent years, influenced by adjustments in the supply and demand structure of the domestic coal market, the coal sourcing channels for most power plants have become increasingly complex, with problems of mixed coal types and numerous batches becoming more prominent. Uncontrollable coal quality upon delivery can lead to problems such as coking and corrosion in power plant boilers during combustion. Especially under the truck transport model, the mine sources are complex and scattered, often with large and small mines mixed together. Key coal quality parameters such as volatile matter, calorific value, sulfur content, and ash fusion properties of a single batch of coal fluctuate frequently. Even when coal is from the same origin and of the same nominal type, there are often significant differences in coal quality indicators between different batches.

[0003] Current traditional coal yard management models in power plants mostly rely on simple classification based on coal type names, and process coal quality data from multiple batches and sources using only arithmetic averages or simplified weighted averages. This extensive management approach easily leads to distorted coal quality data, failing to accurately reflect the actual combustion characteristics of the coal entering the furnace. Applying this insufficiently precise coal quality data to coal blending calculations directly results in large prediction deviations in coal blending schemes, delayed boiler combustion optimization adjustments, and in severe cases, can cause safety hazards such as furnace coking, unstable combustion, or even flameout, hindering the safe and economical operation of the unit. Therefore, how to effectively extract stable and representative coal quality characteristics from discrete, fluctuating, and chaotic incoming coal quality data to provide accurate and reliable data support for intelligent coal blending in power plants has become a critical technical problem that coal-fired power plants urgently need to solve. Summary of the Invention

[0004] (a) The technical problem to be solved by the present invention (1) The coal quality parameters of different batches of the same coal type fluctuate greatly, and simple averaging cannot represent the true characteristics. (2) The massive historical coal data has not been effectively mined, and the data utilization rate is low. (3) The lack of a systematic feature extraction process leads to poor input data quality for subsequent coal blending optimization models, affecting the accuracy of decision-making.

[0005] (II) The technical solution adopted in this invention A method for extracting coal quality characteristics from coal sources used in coal-fired power plants, the method comprising: Establish a statistical database of raw coal quality, the statistical database including parameter groups for several coal types, and each parameter group for the same coal type includes multiple batch sub-parameter groups; The quality consistency of the same coal type is checked against the threshold condition based on the sub-parameter groups of multiple batches for each coal type with the same name. For the same type of coal whose quality consistency meets the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are weighted and averaged to obtain the parameter group of the characteristic coal; for the same type of coal whose quality consistency does not meet the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are clustered and subdivided to obtain the parameter group of the characteristic coal. The parameter sets of all characteristic coals are filtered according to the threshold parameters to obtain a set of candidate coal types.

[0006] Optionally, the parameters of the sub-parameter group include batch unit price, ash content, ash fusion point, lower heating value, and coal quantity delivered to the plant.

[0007] Optionally, the step of checking whether the quality consistency of the same type of coal meets the threshold condition based on multiple batches of sub-parameter groups for each type of coal includes: Calculate the range of key coal quality parameters between different batches; If the range of each key coal quality parameter does not exceed the upper limit of the threshold, then the quality consistency of the same type of coal meets the threshold condition. If the range of at least one key coal quality parameter exceeds the upper limit of the threshold, then the quality consistency of the same type of coal does not meet the threshold condition.

[0008] Optionally, the key coal quality parameters include ash content, ash melting point, and lower heating value.

[0009] Optionally, for the same type of coal whose quality consistency does not meet the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are clustered and subdivided into parameter groups of characteristic coals, including: Clustering algorithms are used to divide the sub-parameter groups of multiple batches of coal of the same name into multiple categories; If the category contains at least two batches of sub-parameter groups, then the sub-parameter groups of all batches in the category are weighted averaged to obtain the parameter group of the characteristic coal. If the category contains a batch of sub-parameter groups, then the batch of sub-parameter groups is used as the parameter group of the characteristic coal.

[0010] Optionally, the threshold parameters include a statistical threshold for the amount of coal entering the plant and a minimum permissible threshold for ash fusion point.

[0011] (III) Beneficial Effects This method can effectively extract characteristic coals with stable quality and reliable safety, while eliminating coal types with large fluctuations or dangers, providing a high-quality data foundation for precise coal blending. Attached Figure Description

[0012] Figure 1 This is a flowchart of a method for extracting coal quality characteristics from coal sources of a coal-fired power plant according to one or more embodiments. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0014] Before describing the various embodiments of this application in detail, the technical concept of this application is first briefly described: Existing coal quality parameters of different batches of the same coal type fluctuate greatly, and simple averaging cannot represent the true characteristics. Massive historical coal data has not been effectively mined, resulting in low data utilization and a lack of a systematic feature extraction process. This leads to poor input data quality for subsequent coal blending optimization models, affecting the accuracy of decision-making. Therefore, the key improvement of the coal quality feature extraction method for coal sources in coal-fired power plants provided by this application lies in first verifying the quality consistency of the same coal type. Based on the verification results, different methods are used to calculate the characteristic coal parameters, making the characteristic parameters more accurately reflect the coal quality characteristics and avoiding errors caused by simple averaging. The calculation process fully utilizes historical coal data, and the overall feature extraction process has good scalability, with the characteristic coal database being dynamically updated. The specific principles of the coal quality feature extraction method for coal sources in coal-fired power plants of this application are described below with reference to more embodiments.

[0015] Specifically, such as Figure 1 As shown, the coal quality characteristic extraction method for coal sources in this embodiment of the coal-fired power plant includes the following steps: Step S10: Establish a statistical database of raw coal quality. The statistical database includes parameter groups for several coal types. Each parameter group for a coal type with the same name includes multiple batches of sub-parameter groups.

[0016] Step S20: Check whether the quality consistency of the same coal type meets the threshold condition based on the sub-parameter groups of multiple batches of each coal type.

[0017] Step S30: For the same type of coal whose quality consistency meets the threshold condition, perform weighted average processing on the sub-parameter groups of multiple batches of the same type of coal to obtain the parameter group of the characteristic coal; for the same type of coal whose quality consistency does not meet the threshold condition, perform clustering and subdivision processing on the sub-parameter groups of multiple batches of the same type of coal to obtain the parameter group of the characteristic coal.

[0018] Step S40: Filter the parameter groups of all characteristic coals according to the threshold parameters to obtain candidate coal types.

[0019] In one or more embodiments, a statistical database of raw coal quality is established. For each batch of coal, regardless of size or name, statistics are performed. The coal quality statistical database must include the coal name, unit price of the batch, ash content of the batch, total quantity of the batch, ash fusion point of the batch, and calorific value of the batch. The longer the database statistics and data collection period and the larger the data volume, the more beneficial it is for the implementation of this embodiment. For each batch of coal entering the plant, the following parameters are recorded: (1) Coal type name and code N i (2) Batch unit price P ij (3) Ash content A ij (Unit: %); (4) Ash melting point T ST , ij (Unit: °C); (5) Lower heating value Q net , ij (Unit: MJ / kg); (6) Coal input quantity (m) ij (Unit: tons). The database stores all historical coal arrival data in tabular form.

[0020] Here, the subscript i indicates the total number of coal types with different names, and the subscript j indicates the batch numbered j for a given coal type with name i. For example, for a coal type with name number 7, the main parameters of the 15th shipment delivered to the plant since the start of data collection can be represented as P. 7-15 A 7-15 T ST，7-15 Q net,7-15 m 7-15 .

[0021] In one or more embodiments, verifying whether the quality consistency of a coal type meets a threshold condition based on sub-parameter groups of multiple batches for each coal type includes: calculating the range of key coal quality parameters between different batches; if the range of each key coal quality parameter does not exceed the upper limit of the threshold, then the quality consistency of the coal type meets the threshold condition; if the range of at least one key coal quality parameter exceeds the upper limit of the threshold, then the quality consistency of the coal type does not meet the threshold condition. For example, key coal quality parameters include ash content, ash fusion point, and lower heating value.

[0022] For example, if the range values ​​of each key coal quality parameter meet the following conditions, then the quality consistency meets the threshold condition: , In the formula, α A α T α Q These represent the upper threshold values ​​for the range of ash content, the upper threshold values ​​for the range of ash melting point, and the upper threshold values ​​for the range of lower heating value, respectively.

[0023] In one or more embodiments, for coal types with the same name whose quality consistency meets the threshold condition, the sub-parameter groups of multiple batches of the same coal type are weighted and averaged to obtain the parameter group of the characteristic coal, as follows: , In the formula, These represent the characteristic ash content, characteristic ash fusion point, characteristic calorific value, and average price of coal type i, respectively. The characteristic coal quantity is directly obtained by summing the quantities of each coal type, which is... .

[0024] In one or more embodiments, for the same type of coal whose quality consistency does not meet the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are clustered and subdivided into the parameter group of the characteristic coal. This includes: using a clustering algorithm to divide the sub-parameter groups of multiple batches of the same type of coal into multiple categories; if the category contains at least two batches of sub-parameter groups, then the sub-parameter groups of all batches in the category are weighted and averaged to obtain the parameter group of the characteristic coal; if the category contains one batch of sub-parameter groups, then the sub-parameter groups of that batch are used as the parameter group of the characteristic coal.

[0025] For example, for all batches of this coal type, the parameter vector for each batch [A] is used. ij ,T ST , ij Q net , ij ] are used as sample points. Clustering algorithms (such as K-means, DBSCAN, or hierarchical clustering) are used to divide the sample points into a categories. The number of clusters a can be automatically determined by the elbow rule or silhouette coefficient.

[0026] For example, for each category c (c=1,2,…,a) containing multiple batch sub-parameter groups, the parameters of all batches within that category are weighted by the amount of coal delivered to the plant to obtain the characteristic parameters of that category: , At this point, the raw coal type name i is subdivided into a "characteristic coals", each of which has independent characteristic parameters and inventory.

[0027] After performing steps 2 and 3 above on all coal type names i, a set of parameters for characteristic coals is obtained. Each characteristic coal is represented by an index k (k = 1, 2, ..., K, where K is the total number of characteristic coals). The parameters for each characteristic coal k include: characteristic ash content A. k Characteristic ash melting point T ST,k Characteristic calorific value Q net,k Average price P k Total inventory m k .

[0028] To further improve data quality and computational efficiency, the following threshold parameters are set for filtering: (1) Threshold for statistical measurement of coal entering the plant m L (e.g. m) L =50 tons): If the inventory of a certain characteristic coal is m k <m L If the coal is not found in the characteristic coal bin, it will be removed from the bin to reduce the computational burden in subsequent calculations.

[0029] (2) Minimum allowable threshold T for ash melting point L (e.g., T) L =1150℃): If the ash fusion point T of a certain characteristic coal is... ST,k <T L If the coal type is not suitable for blending, it will be excluded because it poses a serious risk of coking.

[0030] After screening, the final characteristic coal pool is the set of candidate coal types that can be used for coal blending and combustion optimization.

[0031] This embodiment uses a coal-fired power plant as an example to illustrate the application process of the above-mentioned coal quality characteristic extraction method. The coal source of this power plant is complex, mainly from multiple small coal mines. Coal is transported by truck, with 10 to 20 truckloads entering the plant every day. The coal types are named "Mine 1", "Mine 2", "Mine 3", etc., but the quality of coal types with the same name fluctuates greatly.

[0032] I. Establish a statistical database of raw coal quality. Data on coal delivered to the plant over the past year is collected, and some records are shown in the table below (example): .

[0033] II. Quality Consistency Inspection of Coal Types with the Same Name Set the upper limit of the threshold α A =5%, α T =50℃, α Q =2MJ / kg.

[0034] (1) For “Ore 1”: Ash content range 10.2-15.8%, range 5.6%, greater than the set threshold of 5%; Ash melting point range 1180-1280℃, range 100℃, greater than the set threshold of 50℃; Calorific value range 22.1~24.2MJ / kg, range 2.1MJ / kg, greater than the set threshold of 2MJ / kg. Clustering subdivision is required.

[0035] (2) For "Mine 2": Ash content 18.5-19.1%, range 0.6% < 5%; Ash fusion point 1330-1350℃, range 20℃ < 50℃; Calorific value 20.1~20.5 MJ / kg, range 0.4 < 2. Meeting consistency, the weighted average of each characteristic coal parameter is obtained directly: A矿2 =(18.5×200+19.1×180) / (200+180)=18.79%; T ST,矿2 =(1350×200+1330×180) / 380=1340.5℃; Q net,矿2 =(20.5×200+20.1×180) / 380=20.31MJ / kg; P 矿2 =(620×200+630×180) / 380=624.7 yuan / ton; m 矿2 =380 tons.

[0036] (3) For "Ore 3": Ash melting point 1120~1130℃, range 10℃ < 50℃, but the ranges of ash content and calorific value are both small, so they can be considered consistent. The calculated characteristic parameters are: A 矿3 =25.05%, T ST,矿3 =1125℃, Q net,矿3 =17.65MJ / kg, P 矿3 =555 yuan / ton, m 矿3 =550 tons.

[0037] (4) Cluster subdivision of “mine 1” Using three batches from Mine 1 as sample points, this example is only for illustrating the calculation process. In reality, because this method is applicable to complex coal blending situations, the actual number of batches often reaches dozens or even hundreds. After K-means clustering (k=2), two categories are obtained. Since clustering is a commonly used method in the field of data statistics, the clustering process will not be elaborated here. The results of the example are shown below: Category 1: Includes batches 1 and 3 (lower ash content, higher ash melting point, higher calorific value), weighted average: A 矿1,1 =(12.5×120+10.2×150) / (120+150)=11.20%; T ST,矿1,1 =(1250×120+1280×150) / 270=1266.7℃; Q net,矿1,1 =(23.5×120+24.2×150) / 270=23.89MJ / kg; P 矿1,1 =(680×120+675×150) / 270=677.2 yuan / ton; m 矿1,1 =270 tons.

[0038] Category 2: Includes batch 2 (high ash content, low ash melting point, low calorific value), the characteristic parameters of which are the data of this batch: A 矿1,2 =15.8%, T ST,矿1,2 =1180℃, Q net,矿1,2 =22.1 MJ / kg, P 矿1,2 =690 yuan / ton, m 矿1,2 =80 tons, At this point, "mine 1" was subdivided into two characteristic coals: mine 1-1 (high quality 1) and mine 1-2 (low quality).

[0039] IV. Constructing a Characteristic Coal Storage The parameter sets for all characteristic coals are summarized below: .

[0040] V. Screening of Characteristic Coal Set m L =100 tons, T L =1150℃. K2 (mine 1-2) inventory 80 tons < 100 tons, eliminated. K4 (mine 3) ash fusion point 1125℃ < 1150℃, eliminated. The final retained characteristic coals are K1 and K3, namely mine 1-1 and mine 2, with inventories of 270 tons and 380 tons respectively. These two characteristic coals are the candidate coal types for subsequent coal blending optimization.

[0041] As can be seen from this embodiment, this method effectively extracts characteristic coals with stable quality and reliable safety, and eliminates coal types with large fluctuations or dangers, providing a high-quality data foundation for precise coal blending.

[0042] Compared with existing technologies, this method has the following advantages: (1) Improved data accuracy: By using clustering algorithms, coal types with the same name but different quality are subdivided into multiple stable subclasses, so that the characteristic parameters can more realistically reflect the coal quality characteristics and avoid the errors caused by simple averaging.

[0043] (2) High data utilization: Make full use of historical coal arrival data to explore the hidden quality distribution patterns and provide rich information for subsequent coal blending optimization.

[0044] (3) High engineering practicality: setting coal quantity threshold and ash melting point safety threshold, eliminating invalid or dangerous coal types, ensuring both calculation efficiency and boiler safety.

[0045] (4) Good scalability: The characteristic coal warehouse can be dynamically updated. Each new batch can be integrated into the warehouse according to the same process, realizing data self-evolution.

[0046] The specific embodiments of the present invention have been described in detail above. Although some embodiments have been shown and described, those skilled in the art should understand that modifications and improvements can be made to these embodiments without departing from the principles and spirit of the present invention as defined by the claims and their equivalents, and such modifications and improvements should also be within the protection scope of the present invention.

Claims

1. A method for extracting coal quality characteristics from coal sources used in coal-fired power plants, characterized in that, The coal quality feature extraction method includes: Establish a statistical database of raw coal quality, the statistical database including parameter groups for several coal types, and each parameter group for the same coal type includes multiple batch sub-parameter groups; The quality consistency of the same coal type is checked against the threshold condition based on the sub-parameter groups of multiple batches for each coal type with the same name. For the same type of coal whose quality consistency meets the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are weighted and averaged to obtain the parameter group of the characteristic coal; for the same type of coal whose quality consistency does not meet the threshold condition, the sub-parameter groups of multiple batches of the same type of coal are clustered and subdivided to obtain the parameter group of the characteristic coal. The parameter sets of all characteristic coals are filtered according to the threshold parameters to obtain a set of candidate coal types.

2. The method for extracting coal quality characteristics from coal sources for coal-fired power plants according to claim 1, characterized in that, The parameters of the sub-parameter group include batch unit price, ash content, ash melting point, lower heating value, and coal quantity delivered to the plant.

3. The method for extracting coal quality characteristics from coal sources for coal-fired power plants according to claim 2, characterized in that, The step of verifying whether the quality consistency of the same coal type meets the threshold condition based on multiple batches of sub-parameter groups for each coal type includes: Calculate the range of key coal quality parameters between different batches; If the range of each key coal quality parameter does not exceed the upper limit of the threshold, then the quality consistency of the same type of coal meets the threshold condition. If the range of at least one key coal quality parameter exceeds the upper limit of the threshold, then the quality consistency of the same type of coal does not meet the threshold condition.

4. The method for extracting coal quality characteristics from coal sources for coal-fired power plants according to claim 3, characterized in that, The key coal quality parameters include ash content, ash melting point, and lower heating value.

5. The method for extracting coal quality characteristics from coal sources for coal-fired power plants according to claim 3, characterized in that, For coal types with the same name whose quality consistency does not meet the threshold condition, the sub-parameter groups of multiple batches of the same coal type are clustered and subdivided into parameter groups for characteristic coals, including: Clustering algorithms are used to divide the sub-parameter groups of multiple batches of coal of the same name into multiple categories; If the category contains at least two batches of sub-parameter groups, then the sub-parameter groups of all batches in the category are weighted averaged to obtain the parameter group of the characteristic coal. If the category contains a batch of sub-parameter groups, then the batch of sub-parameter groups is used as the parameter group of the characteristic coal.

6. The method for extracting coal quality characteristics from coal sources for coal-fired power plants according to claim 1, characterized in that, The threshold parameters include the statistical threshold for coal input and the minimum allowable threshold for ash melting point.

Images