System and method for predicting fuel consumption properties

Abstract

Disclosed is a system capable of artificial neural network-based virtual measurement for predicting real-time fuel consumption property information, which is difficult to identify with only a conventional steam power generation fuel management system. The system comprises: a fuel management device for generating predetermined interval fuel consumption property information by using an analysis result to be input therein; a driving management device for generating driving information about the predetermined interval by using driving operation information to be input therein; and a real-time fuel consumption property prediction device which generates an artificial neural network-based virtual measurement model by referring to trends in the fuel consumption property information and the driving information, and which generates a prediction result of real-time fuel consumption property information by using the virtual measurement model.

Description

System and method for predicting the properties of consumed fuel

[0001] The present invention relates to a technology for predicting real-time fuel consumption characteristics of thermal power generation, and more specifically, to a system and method for predicting real-time fuel consumption characteristics information, which is difficult to grasp with the current thermal power generation operating system, as operating information data in a virtual measurement model.

[0002] In addition, the present invention relates to a field linkage system and method that enables continuous improvement of predictive performance by linking a driving management system and a fuel management system.

[0003] Consumable fuel is a concept that mixes low-grade fuel and high-grade fuel. Since it is difficult to achieve normal power generation performance with only low-grade fuel, it is necessary to supplement the heat capacity by mixing high-grade fuel.

[0004] The design of the fuel to be consumed is to select the fuel type to be mixed and determine the mixing ratio by referring to the property information of the currently held incoming fuel to ensure compliance with the power generation operation standards.

[0005] However, it is very difficult to estimate the properties of consumed fuel using only the properties information of the incoming fuel managed on site, as the properties information of a sample group of only a few kilograms among the population of tens of thousands of tons of incoming fuel is used as the properties of the incoming fuel.

[0006] Furthermore, while analysis of the properties of consumed fuel utilizes samples from fuel mixing facilities, it takes several days (approximately five days) to conduct the analysis, while it also takes several hours (approximately ten hours) from sample acquisition to boiler supply and combustion. Therefore, the current thermal power generation fuel management system lacks a method for determining the properties of consumed fuel during boiler operation.

[0007] The properties of the fuel consumed are a very important variable that determines the stability of boiler combustion.

[0008] Most fuels generate heat at the bottom of the boiler furnace, where combustion begins. This creates a consistent temperature gradient within the boiler along the flow path of the combustion gases. Therefore, stable combustion refers to ideal combustion, where the ash contained in the fuel falls to the bottom of the furnace during combustion and is discharged, or is discharged outside the boiler along with the combustion gases without fusing.

[0009] On the other hand, combustion instability refers to a phenomenon in which sufficient heat generation is not achieved at the bottom of the furnace due to the ash contained in the consumed fuel, and heat generation continues to occur at the top of the furnace along the combustion gas flow path, thereby generating local high-temperature areas (nitrogen oxides (NOx) and ash deposition on the heat transfer surface) and low-temperature areas (carbon monoxide and unburned carbon).

[0010] This phenomenon can be resolved to some extent by accelerating combustion through manipulation of combustion-related operating variables (such as combustion air supply amount), but since real-time fuel consumption properties information is absent during boiler operation, optimization of operating operations or control adjustments is difficult.

[0011] Since thermal power boilers are large-scale facilities designed to handle base loads, even a slight manipulation of combustion-related operating variables can lead to economic losses in the hundreds of billions of won.

[0012] In particular, when calculating thermal efficiency, since the thermal efficiency information of incoming fuel is referenced instead of the real-time fuel consumption information, it is difficult to accurately calculate thermal efficiency during boiler operation, making it difficult to monitor thermal efficiency as well as identify and manage inhibitory factors.

[0013] Recently, with the social demand for improving the operational environment of thermal power generation, the need for reducing the concentration of pollutants in boiler exhaust gas and improving thermal efficiency while satisfying the target power generation output has been emphasized, and therefore, the need for obtaining real-time information on the properties of consumed fuel, such as not only the calorific value of consumed fuel but also ash content, ash melting temperature, and sulfur content, has increased.

[0014] In general, the fuel management system is a system that manages the history of incoming and consumed fuel, including the type of coal and the properties of each item. When consumed fuel is burned under similar mixed conditions to the past, the power generation operation manager refers to the history from the fuel management system and takes appropriate action.

[0015] The fuel management system (time series data with long acquisition intervals and irregular intervals) and the driving management system (time series data with regular acquisition intervals) have been operated independently of each other due to different management data characteristics.

[0016] The present invention has been proposed to solve the problems according to the above background technology, and its purpose is to provide a system and method capable of virtual measurement based on an artificial neural network for predicting real-time information on the properties of consumed fuel, which is difficult to determine with only the current thermal power generation fuel management system.

[0017] In addition, another purpose of the present invention is to provide a real-time fuel consumption condition prediction system and method that enables continuous improvement (self-learning) of virtual measurement model performance by linking fuel consumption condition information and operating information that are managed independently from each other.

[0018] In order to achieve the task presented above, the present invention provides a system capable of virtual measurement based on an artificial neural network for predicting real-time fuel consumption properties that are difficult to grasp with only the current thermal power generation fuel management system.

[0019] The above system,

[0020] A fuel management device that generates fuel consumption characteristics information for a certain section using input analysis results;

[0021] A driving management device that generates driving information for a certain section using input driving operation information; and

[0022] It is characterized by including a real-time fuel consumption property prediction device that generates a virtual measurement model based on an artificial neural network by referring to the trend of the above fuel consumption property information and the above driving information, and generates a prediction result of real-time fuel consumption property information using the virtual measurement model.

[0023] In addition, the real-time fuel consumption property prediction device is characterized by including a data acquisition unit that acquires the fuel consumption property information and the driving information; a learning unit that generates a learning data set to be input into the artificial neural network by using past fuel consumption property information of a certain section among the fuel consumption property information and past driving information of a certain section among the driving information; a model generation unit that inputs the learning data set into the artificial neural network and generates the virtual measurement model through learning; and a prediction unit that applies real-time driving information among the driving information to the virtual measurement model to generate a prediction result of the real-time fuel consumption property information.

[0024] In addition, the fuel consumption characteristics information is characterized in that it is produced by extracting the original fuel consumption characteristics information, converting it into data having the same format as the past operation information, and calculating the fuel consumption characteristics supplied to each combustion burner at each point in the learning target time zone by utilizing the silo loading amount variable data and the fuel supply amount variable data.

[0025] In addition, the learning data set is characterized in that it is generated by merging the past driving information of the predetermined section and the consumption fuel characteristics supplied to each combustion burner at each point in time.

[0026] In addition, the above transformation is characterized in that it is performed in the same time series as the above past driving information.

[0027] In addition, the learning data set is structured in a table form, and the table is characterized by being composed of driving information data representing past driving information for a certain section and consumption fuel properties information data representing past fuel consumption properties information for a certain section.

[0028] Additionally, the table items are characterized by including date times, air flow, air pressure, Damper Throttle, HHV (Higher Heating Value) values, ash values, IDT values, and HT values. IDT is the ash melting point, which is the temperature at which a spherical ash weight begins to melt when heated, and HT is the hemispherical melting point, which is the temperature at which a spherical ash weight becomes half the diameter of the original weight when heated.

[0029] In addition, the above-mentioned fuel properties information is characterized by including the calorific value, ash content, ash melting temperature, and sulfur content of the fuel.

[0030] In addition, the prediction result of the real-time fuel consumption properties information is characterized in that each properties item of the real-time fuel consumption properties information is produced by inputting real-time driving information among the driving information into one virtual measurement model.

[0031] In addition, the prediction result of the real-time fuel consumption properties information is characterized in that each property item of the real-time fuel consumption properties information is produced by inputting the real-time driving information among the driving information into each virtual measurement model generated for each property item of the real-time fuel consumption properties information.

[0032] On the other hand, another embodiment of the present invention provides a method for predicting fuel consumption properties, comprising: (a) a step of generating fuel consumption properties information for a certain section using an analysis result input by a fuel management device; (b) a step of generating driving information for the certain section using driving operation information input by a driving management device; and (c) a step of generating a virtual measurement model based on an artificial neural network using a trend of the fuel consumption properties information and the driving information by a real-time fuel consumption properties prediction device, and generating a prediction result of real-time fuel consumption properties information using the virtual measurement model.

[0033] In addition, the step (c) is characterized by including: (c-1) a step in which a data acquisition unit acquires the fuel consumption characteristics information and the driving information; (c-2) a step in which a learning unit generates a learning data set to be input into the artificial neural network using past fuel consumption characteristics information of a certain section among the fuel consumption characteristics information and past driving information of a certain section among the driving information; (c-3) a step in which a model generation unit inputs the learning data set into the artificial neural network and generates the virtual measurement model through learning; and (c-4) a step in which a prediction unit applies real-time driving information among the driving information to the virtual measurement model to generate a prediction result of the real-time fuel consumption characteristics information.

[0034] According to the present invention, by quantifying real-time fuel consumption characteristics information, it is possible to analyze the causal relationship with combustion-related operating variables, thereby enabling systematic response to negative phenomena such as the generation of pollutants during combustion or slag adhesion in the heat transfer section.

[0035] In addition, another effect of the present invention is that it can contribute to improving the safety of power generation sites by significantly reducing the number of times field workers access the facility to obtain consumed fuel samples.

[0036] FIG. 1 is a block diagram of a system for predicting fuel consumption properties according to an embodiment of the present invention.

[0037] Figure 2 is a conceptual diagram showing the combustion instability phenomenon of a typical steam power boiler.

[0038] Figure 3 is a detailed block diagram of the real-time fuel consumption status prediction device illustrated in Figure 1.

[0039] FIG. 4 is a conceptual diagram for calculating combustion results using process situation variables, combustion-related measurement variables, and properties information of consumed fuel according to one embodiment of the present invention.

[0040] Figure 5 is a conceptual diagram for calculating thermal efficiency using power generation, properties information of consumed fuel, properties information of incoming fuel, etc. according to one embodiment of the present invention.

[0041] FIG. 6 is an example of an artificial neural network-based virtual measurement model according to one embodiment of the present invention.

[0042] Figure 7 is a conceptual diagram showing a procedure for constructing a learning data set according to one embodiment of the present invention.

[0043] FIG. 8 is an example showing an algorithm for calculating the start of sampling (ts) of consumed fuel supplied at a specific time (tb) according to one embodiment of the present invention.

[0044] Figure 9 is an example of the configuration of a learning data set according to one embodiment of the present invention.

[0045] Figure 10 is a conceptual diagram showing a case in which one virtual prediction model is utilized according to one embodiment of the present invention.

[0046] Figure 11 is a conceptual diagram showing a case where a virtual measurement model for each property item is utilized according to another embodiment of the present invention.

[0047] Figure 12 is a conceptual diagram for processing input / output data of a virtual measurement model according to one embodiment of the present invention.

[0048] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention.

[0049] When describing each drawing, similar reference numerals are used for similar components.

[0050] The terms "first," "second," etc. may be used to describe various components, but these components should not be limited by these terms. These terms are used solely to distinguish one component from another.

[0051] For example, without departing from the scope of the present invention, a first component could be referred to as a second component, and similarly, a second component could also be referred to as a first component. The term "and / or" includes any combination of a plurality of related listed items or any one of a plurality of related listed items.

[0052] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0053] Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense unless expressly defined in this application.

[0054] Hereinafter, with reference to the attached drawings, a fuel consumption property prediction system and method according to an embodiment of the present invention will be described in detail.

[0055]

[0056] The principle of the present invention is to emulate the ability of a power generation operation manager to qualitatively estimate (learn and predict by experience) the characteristics of fuel consumption in a certain section by referring to the operating information trend of the section, using a virtual measurement model based on artificial intelligence (artificial neural network, etc.).

[0057] FIG. 1 is a block diagram of a fuel consumption property prediction system (100) according to an embodiment of the present invention. Referring to FIG. 1, the fuel consumption property prediction system (100) may be configured to include a fuel management terminal (110) for inputting analysis results, a fuel management device (120) for generating fuel consumption property information for a certain section using the input analysis results, a driving management terminal (130) for inputting driving operation information, a driving management device (140) for generating driving information for the certain section using the driving operation information, and a real-time fuel consumption property prediction device (150) for qualitatively estimating the fuel consumption property for a certain section by imitating it with an artificial neural network-based virtual measurement model with reference to the fuel consumption property information and the trend of the driving information, and generating a prediction result of real-time fuel consumption property information.

[0058] The fuel management terminal (110) performs the function of inputting analysis results into the fuel management device (120). The analysis results can be generated using a computer program. Of course, the administrator can input the already analyzed results using a memory stick, or the administrator can input them manually.

[0059] The fuel management device (120) performs the function of managing the history of incoming and consumed fuel, including the origin, coal type, and individual properties. Therefore, when consumed fuel is combusted under similar mixed conditions to the past, the power generation operation manager can refer to the history from the fuel management device (120) and take appropriate action. To this end, the fuel management device (120) is configured with a microprocessor, a communication circuit, memory, software, data, etc.

[0060] The memory may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., Secure Digital (SD) or eXtreme Digital (XD) memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read only memory (ROM), an electrically erasable programmable read only memory (EEPROM), a programmable read only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory may operate in relation to a web storage or a cloud server that performs a storage function on the internet.

[0061] Referring to FIG. 1, the driving management terminal (130) performs the function of inputting driving operation information to the driving management device (140). Of course, in this case, the driving management terminal (130) generates driving operation information by referencing past fuel consumption growth information from the real-time fuel consumption characteristics prediction device (150). Of course, the manager inputs driving operation information to the driving management device (140) by referencing past fuel consumption characteristics information through the driving management terminal (130).

[0062] The driving management device (140) manages the operation of a steam power boiler (not shown) based on driving operation information, performs the function of generating driving information, and transmits this driving information to a real-time fuel consumption characteristics prediction device (150). For this purpose, the driving management device (140) is also configured with a microprocessor, a communication circuit, a memory, software, data, etc.

[0063] A real-time fuel consumption property prediction device (150) performs the ability to qualitatively estimate (learn and predict by experience) the fuel consumption property of a certain section by imitating it as a virtual measurement model based on an artificial neural network, with reference to past fuel consumption property information and trends of past driving information of a certain section. In other words, it performs the function of qualitatively estimating the fuel consumption property of a specific section by imitating it as a virtual measurement model based on an artificial neural network, with reference to the past fuel consumption property information and trends of the past driving information.

[0064] The fuel management terminal and driving management terminal (130) illustrated in FIG. 1 may be a mobile phone, a smart phone, a laptop computer, a digital broadcasting terminal, a PDA (Personal Digital Assistant), a PMP (Portable Multimedia Player), a navigation device, a note pad, etc.

[0065] Figure 2 is a conceptual diagram illustrating the combustion instability phenomenon of a typical steam power boiler (120). Referring to Figure 2, since the steam power boiler (120) is a large-scale facility intended for base load, even a slight manipulation of combustion-related operating variables can lead to economic losses in the hundreds of billions of won. In other words, due to the low calorific value and high ash content of low-grade fuel, the ash fuses to the heat transfer area at the bottom of the furnace, preventing sufficient heat generation and resulting in local overheating.

[0066] That is, since sufficient heat generation is not achieved at the bottom of the furnace and heat generation continues along the combustion gas flow path to the top of the furnace, a phenomenon occurs in which local high-temperature areas (nitrogen oxides and ash deposits on the heat transfer surface) and low-temperature areas (carbon monoxide and unburned carbon) are created. This phenomenon is called combustion instability.

[0067] Fig. 3 is a detailed block diagram of the real-time fuel consumption property prediction device (150) illustrated in Fig. 1. Referring to Fig. 3, the real-time fuel consumption property prediction device (150) may be configured to include a data acquisition unit (310), a data storage unit (320), a learning unit (330), a model generation unit (340), a prediction unit (350), etc.

[0068] The data acquisition unit (310) performs a function of acquiring data such as fuel consumption characteristics information and driving information. The data acquisition unit (310) is connected to a fuel management device (120) and a driving management device (140) via a communication network (not shown), and the data acquisition unit (310) may be configured to include a communication modem, memory, etc.

[0069] A communications network refers to a connection structure that enables information exchange between each node, such as multiple terminals and servers, and can be a public switched telephone network (PSTN), a public switched data network (PSDN), an integrated services digital network (ISDN), a broadband ISDN (BISDN), a local area network (LAN), a metropolitan area network (MAN), or a wide area network (WLAN).

[0070] However, the present invention is not limited thereto, and may be a wireless communication network such as CDMA (Code Division Multiple Access), WCDMA (Wideband Code Division Multiple Access), Wibro (Wireless Broadband), WiFi (Wireless Fidelity), HSDPA (High Speed ​​Downlink Packet Access) network, Bluetooth, NFC (Near Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc. Or, it may be a combination of these wired communication networks and wireless communication networks.

[0071] The data storage (320) performs a function of storing data acquired through the data acquisition unit (310). To this end, the data storage (320) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., SD (Secure Digital) or XD (eXtreme Digital) memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read only memory (ROM), an electrically erasable programmable read only memory (EEPROM), a programmable read only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk.

[0072] Additionally, it may operate in connection with web storage and cloud servers that perform storage functions on the Internet.

[0073] Of course, the data storage (320) can also store data generated by the learning unit (330), model generation unit (340), etc.

[0074] The learning unit (330) performs the function of generating a learning data set using fuel consumption characteristics information and operating information. The fuel consumption characteristics information is extracted by converting (interpolating) the original fuel consumption characteristics information into data (time series data with regular acquisition intervals) having the same format as the operating information, and calculating the fuel consumption characteristics supplied to each combustion burner at each point in time during the learning target time period by utilizing the loading amount variable (ton) of the silo (not shown) and the supply amount variable (ton / hr) of the fuel supplier (not shown). Finally, the learning data set is formed by merging it with operating information of a certain section.

[0075] The model generation unit (340) inputs the learning data set generated in the learning unit (330) into an artificial neural network and generates a virtual measurement model through artificial neural network-based learning. The artificial neural network may be a DNN (Deep Neural Network), a CNN (Convolution Neural Network), an RNN (Recurrent Neural Network), etc.

[0076] The prediction unit (350) performs the function of outputting data on each item of the current fuel consumption characteristics using a virtual measurement model.

[0077] The learning unit (330), model generation unit (340), and prediction unit (350) illustrated in FIG. 1 represent units that process at least one function or operation, and may be implemented in software and / or hardware. In hardware implementation, the units may be implemented in an ASIC (application specific integrated circuit), DSP (digital signal processing), PLD (programmable logic device), FPGA (field programmable gate array), processor, microprocessor, other electronic unit, or a combination thereof designed to perform the above-described functions.

[0078] In a software implementation, it may include software composition components (elements), object-oriented software composition components, class composition components, task composition components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, data, databases, data structures, tables, arrays, and variables. Software, data, etc. may be stored in memory and executed by a processor. The memory or processor may employ various means well known to those skilled in the art.

[0079] FIG. 4 is a conceptual diagram for calculating a combustion result using process situation variables, combustion-related measurement variables, and properties information of consumed fuel according to an embodiment of the present invention. Referring to FIG. 4, the problem is solved to some extent by accelerating combustion through manipulation of combustion-related operating variables (such as combustion air supply amount), while using real-time properties information of consumed fuel during boiler operation (step S410). In other words, by using real-time properties information of consumed fuel, it is possible to optimize combustion-related operating variables (step S420). Due to this optimization, a combustion result is obtained in which nitrogen oxides (NOx), carbon monoxide (CO), etc. are reduced (step S430).

[0080] Of course, in Fig. 4, it is illustrated that all process situation variables, combustion-related measurement variables, and consumption fuel properties information are used, but the process situation variables and combustion-related measurement variables can be used for reference only.

[0081] Figure 5 is a conceptual diagram illustrating a method for calculating thermal efficiency using information on power generation, fuel consumption properties, and incoming fuel properties according to an embodiment of the present invention. Referring to Figure 5, information on power generation and fuel consumption properties according to the operation of a steam power generation boiler (210) is utilized (step S510). Of course, if there is no fuel consumption properties information, incoming fuel properties information can be utilized, but in this case, there is a disadvantage in that it is difficult to accurately calculate thermal efficiency during boiler operation.

[0082] The boiler water supply amount, boiler increased production amount according to power generation amount, and the calorific value of consumed fuel according to consumed fuel properties information or incoming fuel properties information are calculated (step S520).

[0083] Finally, thermal efficiency is calculated (step S530). In detail, by calculating input energy from the calorific value of fuel consumed and calculating output energy from the boiler steam production and boiler feedwater amount, thermal efficiency can be expressed as thermal efficiency = output energy / input energy (%). Output / input = (power generation, in W) / (fuel consumption, in cal)

[0084] Meanwhile, as the concentration of pollutants in boiler exhaust gas is reduced and thermal efficiency is improved, the need for real-time information on the properties of consumed fuel, such as not only the calorific value of the consumed fuel but also the ash content, ash melting temperature, and sulfur content, is increasing. In other words, the consumed fuel properties information may include the calorific value, ash content, ash melting temperature, and sulfur content of the consumed fuel.

[0085] Fig. 6 is an example of a virtual measurement model based on an artificial neural network according to an embodiment of the present invention. Referring to Fig. 6, a virtual measurement model is created by inputting past driving information and past fuel consumption characteristics information into an artificial neural network (610), and this virtual measurement model (620) is replicated to input real-time driving information to create real-time fuel consumption characteristics information, which is the final output result.

[0086] FIG. 7 is a conceptual diagram showing a procedure for constructing a learning data set according to an embodiment of the present invention. Referring to FIG. 7, the fuel consumption characteristics information is obtained by ① extracting original fuel consumption characteristics information data (i.e., past fuel consumption characteristics information), ② converting (interpolating) it into data (time series data with regular acquisition intervals) having the same format as the operation information (i.e., past operation information), and ③ calculating the fuel consumption characteristics supplied to each combustion burner at each point in the learning target time zone by utilizing the silo loading amount variable (unit: ton) and the fuel supply amount variable (unit: ton / hr) data. ④ Merging it with operation information of a certain section to construct a learning data set.

[0087] Figure 8 is an example showing an algorithm for calculating the sampling start time (ts) of consumed fuel supplied at a specific time (tb) according to one embodiment of the present invention. The variables in Figure 8 are defined as follows.

[0088] tb: boiler supply time of consumed fuel

[0089] ts: sampling time of the consumed fuel

[0090] SiloMass t : Amount of fuel consumed in the silo at time t (ton)

[0091] FeedRate t : Fuel consumption in the boiler at time t (ton / sec)

[0092] Define the While Loop termination condition. That is, define the condition as follows:

[0093] am.

[0094] Initialize the backward time step time by setting n = 0.

[0095] Determine whether the While Loop termination condition is satisfied while increasing the time.

[0096] FIG. 9 illustrates an example of the configuration of a learning data set according to an embodiment of the present invention. Referring to FIG. 9, the learning data set can be efficiently managed depending on the data processing environment and can be structured in a table format as illustrated in FIG. 9. The table is comprised of driving information data (910) indicating past driving information for a certain period of time and fuel consumption characteristics data (920) indicating past fuel consumption characteristics for a certain period of time.

[0097] The table entries include date times, air flow, air pressure, Damper Throttle, Higher Heating Value (HHV), ash values, Initial Decomposition Temperature (IDT), and Hemispherical Temperature (HT). IDT is the ash melting point, the temperature at which a spherical ash weight begins to melt when heated. HT is the hemispherical melting point, the temperature at which a spherical ash weight becomes half its original diameter when heated.

[0098] Fig. 10 is a conceptual diagram illustrating a case of utilizing a single virtual prediction model according to an embodiment of the present invention. Referring to Fig. 10, the structure of the virtual measurement model (1010) can be set to output all condition items by inputting a driving information data set into a single model, depending on the number or complexity (numerical range, etc.) of condition items to be predicted.

[0099] Fig. 11 is a conceptual diagram showing a case where a virtual measurement model for each property item is utilized according to another embodiment of the present invention. Referring to Fig. 11, a structure can be set so that one model outputs one property item data. In other words, each property of real-time consumed fuel is output using a virtual measurement model for each property item. That is, a calorific value virtual measurement model is configured to calculate real-time calorific value, an ash content virtual measurement model is configured to calculate real-time ash content, and an ash melting point virtual measurement model is configured to calculate real-time ash melting point.

[0100] Figure 12 is a conceptual diagram for processing input / output data of a virtual measurement model according to an embodiment of the present invention. Referring to Figure 12, by inputting a time-series operation information data set (1220) for a certain period from the current point in time into a virtual measurement model for which learning has been completed, the combustion burner-specific condition information data (1210) for the current point in time can be predicted.

[0101] Through a domestic thermal power plant demonstration, 192 virtual measurement models were learned for 32 properties items for 6 combustion burners from boiler operation information and consumption fuel properties information (48 cases), and by inputting an operation information data set unrelated to the learning period into the virtual measurement model, it was confirmed that the present invention can be applied in the field to predict real-time mixed consumption fuel properties that can be utilized in thermal power plant boiler combustion management practice with a prediction error rate of within 5%.

[0102] The items of the fuel consumption characteristics information for the virtual measurement target during the demonstration are as follows.

[0103] Item Analysis Condition UnitTotal Moisture Content ARwt%Intrinsic Moisture Content ADwt%Ash Content AR, AD, Dwt%Volatile Matter Content AR, AD, D, DMMFwt%Fixed Carbon Content AR, AD, Dwt%Total Sulfur Content AR, AD, Dwt%Combustible Sulfur Content Dwt%Carbon Content Dwt%Hydrogen Content Dwt%Nitrogen Content DAFwt%Oxygen Content Dwt%Higher Calorific Value AR, AD, Dkcal / kgLower Calorific Value ARkcal / kgAsh Melting Temperature (Oxidizing and Reducing Environment) IDT, ST, HT, FT˚CAsh Detailed Components (Fe2O3, etc.) -%

[0104] AR: As Received (This is the concept of arrival criteria, and is a condition for analyzing the properties of the sample in its original state without any preprocessing.)

[0105] AD: Air Dried (This is the concept of air-dried standard, and is the condition for removing moisture from a sample by drying it naturally or in an environment slightly higher than room temperature and analyzing its properties.)

[0106] D: Dried (This is the concept of anhydrous standard, and is a condition for analyzing the properties by artificially completely removing moisture from the sample)

[0107] DMMF: Dry Mineral Matter Free (This is a concept of dry mineral matter exclusion criteria, and is a condition for completely removing moisture and mineral matter from a sample and analyzing its properties.)

[0108] DAF: Dry Ash Free (This is the concept of an ash-free standard, and is a condition for completely removing moisture and ash from a sample and analyzing its properties.)

[0109]

[0110] Additionally, the steps of the method or algorithm described in connection with the embodiments disclosed herein may be implemented in the form of program instructions that can be executed by various computer means, such as a microprocessor, processor, CPU (Central Processing Unit), etc., and recorded on a computer-readable medium. The computer-readable medium may include program (command) codes, data files, data structures, etc., either singly or in combination.

[0111] The program (command) code recorded on the above medium may be specially designed and configured for the present invention, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium may include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and Blu-rays, and semiconductor memory devices specially configured to store and execute program (command) codes such as ROMs (Read Only Memory), RAMs (Random Access Memory), and flash memories.

[0112] Here, examples of program (instruction) code include not only machine language code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

Claims

1. A fuel management device (120) that generates fuel consumption characteristics information for a certain section using the input analysis results; A driving management device (140) that generates driving information for a certain section using input driving operation information; and A real-time fuel consumption property prediction device (150) that generates a virtual measurement model based on an artificial neural network by referring to the above fuel consumption property information and the trend of the driving information, and generates a prediction result of real-time fuel consumption property information using the virtual measurement model; A fuel consumption property prediction system characterized by including:

2. In paragraph 1, The above real-time fuel consumption property prediction device (150) is A data acquisition unit (310) that acquires the above fuel consumption characteristics information and the above driving information; A learning unit (330) that generates a learning data set to be input into the artificial neural network by using a certain section of past fuel consumption characteristics information among the above fuel consumption characteristics information and a certain section of past driving information among the above driving information; A model generation unit (340) that inputs the above learning data set into the artificial neural network and generates the virtual measurement model through learning; and A fuel consumption property prediction system characterized by including a prediction unit (350) that applies real-time driving information among the above driving information to the virtual measurement model to generate a prediction result of the real-time fuel consumption property information.

3. In paragraph 2, A fuel consumption property prediction system characterized in that the above fuel consumption property information is produced by extracting original fuel consumption property information, converting it into data having the same format as the past operation information, and calculating the fuel consumption property supplied to each combustion burner at each point in the learning target time zone by utilizing the silo loading amount variable data and the fuel supply amount variable data.

4. In paragraph 3, A fuel consumption characteristics prediction system characterized in that the above learning data set is generated by merging past driving information of the above period and the fuel consumption characteristics supplied to each combustion burner at each point in time.

5. In paragraph 3, A fuel consumption property prediction system characterized in that the above transformation is performed in the same time series as the above past driving information.

6. In paragraph 3, A fuel consumption characteristics prediction system characterized in that the above learning data set is structured in a table form, and the table is composed of driving information data (910) representing past driving information of a certain section and consumption fuel characteristics information data (920) representing past consumption fuel characteristics information of a certain section.

7. In paragraph 6, A system for predicting the properties of consumed fuel, characterized in that the items of the above table include date times, air flow, air pressure, damper throttle, HHV (Higher Heating Value) values, ash values, IDT (Initial Decomposition Temperature) values, and HT (Hemispherical Temperature) values, wherein the IDT is an ash melting point, a temperature at which a round ash weight begins to melt when heated, and the HT is a hemispherical melting point, a temperature at which a round ash weight becomes half the diameter of the original weight when heated.

8. In paragraph 1, A consumption fuel properties prediction system characterized in that the above consumption fuel properties information includes the calorific value, ash content, ash melting temperature, and sulfur content of the consumption fuel.

9. In paragraph 1, A fuel consumption property prediction system characterized in that the prediction result of the real-time fuel consumption property information is calculated by inputting real-time driving information among the driving information into one virtual measurement model and calculating each property item of the real-time fuel consumption property information.

10. In paragraph 1, A fuel consumption property prediction system characterized in that the prediction result of the real-time fuel consumption property information is input into each virtual measurement model generated for each property item of the real-time fuel consumption property information among the driving information, and each property item of the real-time fuel consumption property information is produced. 11.(a) A step of generating fuel consumption characteristics information of a certain section using the analysis results input to the fuel management device (120); (b) a step of generating driving information for a certain section using the driving operation information input by the driving management device (140); and (c) A step of generating a virtual measurement model based on an artificial neural network by using the trend of the fuel consumption status information and the driving information as a reference by the real-time fuel consumption status prediction device (150), and generating a prediction result of real-time fuel consumption status information using the virtual measurement model; A method for predicting the properties of fuel consumption, characterized in that it includes:

12. In paragraph 11, Step (c) above, (c-1) A step in which the data acquisition unit (310) acquires the fuel consumption characteristics information and the driving information; (c-2) A step in which a learning unit (330) generates a learning data set to be input into the artificial neural network by using a certain section of past fuel consumption characteristics information among the above fuel consumption characteristics information and a certain section of past driving information among the above driving information; (c-3) A step in which the model generation unit (340) inputs the learning data set into the artificial neural network and generates the virtual measurement model through learning; and (c-4) A method for predicting fuel properties, characterized in that it includes a step in which the prediction unit (350) applies real-time driving information among the driving information to the virtual measurement model to generate a prediction result of the real-time fuel properties information.

13. In paragraph 12, A method for predicting fuel consumption properties, characterized in that the above fuel consumption properties information is produced by extracting original fuel consumption properties information, converting it into data having the same format as the past operation information, and calculating the fuel consumption properties supplied to each combustion burner at each point in time during the learning target time period by utilizing the silo loading amount variable data and the fuel supply amount variable data.

14. In paragraph 13, A method for predicting fuel consumption characteristics, characterized in that the above learning data set is generated by merging past driving information of the above-described section and the fuel consumption characteristics supplied to each combustion burner at each point in time.

15. In paragraph 13, A method for predicting fuel consumption characteristics, characterized in that the above transformation is performed in the same time series as the above past driving information.

16. In paragraph 13, A method for predicting fuel consumption characteristics, characterized in that the above learning data set is structured in a table form, and the table is composed of driving information data (910) representing past driving information of a certain section and consumption fuel characteristics information data (920) representing past consumption fuel characteristics information of a certain section.

Images