Desulfurization system modeling method and device based on data drive
Through the data-driven desulfurization system modeling method, the local maintenance projection method and 1D-CNN model are used to solve the problem of untimely and inaccurate adjustment of desulfurization system in the existing technology, and efficient and accurate prediction and control of SO2 concentration are achieved, satisfying the satisfaction of Environmental protection and economic requirements.
Patent Information
- Application Number
- CN202111192367.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-13
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-10-13
AI Technical Summary
The existing desulfurization system lacks accuracy in regulating the slurry supply volume and slurry circulation pump, resulting in untimely and inaccurate SO2 emission control, making it difficult to meet increasingly stringent environmental protection requirements.
Using a data-driven desulfurization system modeling method, the training data set is constructed by determining the influencing factors related to the SO2 concentration of desulfurization outlet, and the dynamic prediction is established using local maintenance projection method and one-dimensional convolutional neural network (1D-CNN) model. Model, predict the SO2 concentration at the current moment.
It improves the accuracy of SO2 concentration prediction, realizes advance prediction and precise control, reduces operating costs, and has the significance of guiding the control of pollutant emissions from coal-fired units.
Smart Images

Figure CN113935236B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of flue gas desulfurization in coal-fired power plants, and in particular to a data-driven desulfurization system modeling method and device. Background Art
[0002] In the existing technology, SO2 is one of the main causes of acid rain and haze, and the pollutant emissions from coal-fired power plants in my country are the main source of SO2. A large amount of coal is burned in the production process of thermal power plants, and coal often contains a certain amount of sulfur. These sulfur elements will release a large amount of SO2 after combustion. If not controlled, it will cause great harm to the environment. It can be seen that it is imperative to control and reduce the emission of SO2 pollutants from coal-fired power plants.
[0003] Limestone / gypsum wet desulfurization has many advantages, such as mature technology, high reliability, wide source of desulfurizer, strong adaptability to coal types, and the ability to utilize desulfurization products. This desulfurization technology is increasingly favored by modern power plants. The desulfurization system generates a large amount of operating data during operation, which contains rich value for mining. However, many desulfurization systems currently rely on operators to adjust the slurry supply and slurry circulation pump based on experience, and the disadvantage is that the adjustment is not timely and accurate.
[0004] In order to more effectively control the emission of SO2 concentration at the desulfurization outlet, many scholars have carried out the application of machine learning methods and data mining methods in the control of desulfurization systems, especially the application of neural network methods. However, the traditional BP neural network is prone to fall into the local optimum, which leads to the research on neural network data processing stagnating after the initial research. At the same time, this method is a point-to-point prediction method with limited application. In recent years, deep learning is a new field in machine learning research, and its application in desulfurization systems has become a hot topic.
[0005] To this end, with increasingly stringent environmental protection requirements, thermal power plants are in urgent need of a set of intelligent desulfurization system modeling tools. Deep learning of the historical data of the desulfurization system is both an environmental protection requirement and a requirement for improving the economic efficiency of power plant desulfurization. Summary of the invention
[0006] The purpose of the present invention is to provide a data-driven desulfurization system modeling method and device, aiming to solve the above-mentioned problems in the prior art.
[0007] The present invention provides a desulfurization system modeling method based on data drive, comprising:
[0008] Determine the influencing factors related to the SO2 concentration at the desulfurization outlet, collect relevant historical operation data based on the influencing factors, and construct a training data set based on the relevant historical operation data;
[0009] Projecting the relevant historical operation data in the training data set into a low-dimensional subspace by using a locality preserving projection method (LPP);
[0010] Based on the training data set, a one-dimensional convolutional neural network 1D-CNN model is used to establish a dynamic prediction model for SO2 concentration at the desulfurization outlet;
[0011] The current operation data related to the SO2 concentration at the desulfurization outlet at the current moment is obtained, and the predicted value of the SO2 concentration at the desulfurization outlet at the current moment is determined based on the dynamic prediction model of the SO2 concentration at the desulfurization outlet.
[0012] The present invention provides a desulfurization system modeling device based on data drive, comprising:
[0013] A collection module, used to determine the influencing factors related to the SO2 concentration at the desulfurization outlet, collect relevant historical operation data based on the influencing factors, and construct a training data set based on the relevant historical operation data;
[0014] A projection module, used for projecting the relevant historical operation data in the training data set into a low-dimensional subspace by adopting a locality preserving projection method (LPP);
[0015] Establishing a module for establishing a dynamic prediction model of SO2 concentration at a desulfurization outlet using a one-dimensional convolutional neural network 1D-CNN model based on the training data set;
[0016] The prediction module is used to obtain the current operation data related to the SO2 concentration at the desulfurization outlet at the current moment, and determine the predicted value of the SO2 concentration at the desulfurization outlet at the current moment based on the dynamic prediction model of the SO2 concentration at the desulfurization outlet.
[0017] An embodiment of the present invention also provides a data-driven desulfurization system modeling device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the above-mentioned data-driven desulfurization system modeling method when executed by the processor.
[0018] An embodiment of the present invention further provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps of the above-mentioned data-driven desulfurization system modeling method are implemented.
[0019] By adopting the embodiment of the present invention, the SO2 concentration at the desulfurization outlet of the desulfurization system can be predicted in advance, and the 1D-CNN method with better time series data processing effect can be adopted to effectively improve the accuracy of the prediction of the SO2 concentration at the desulfurization outlet, which has guiding significance for the control of pollutant emissions from coal-fired units.
[0020] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0022] Figure 1 is a flow chart of a desulfurization system modeling method based on data-driven according to an embodiment of the present invention;
[0023] Figure 2 1D-CNN model is a schematic diagram of the specific structure of the embodiment of the present invention;
[0024] Figure 3 Schematic diagram of a desulfurization system modeling device based on data drive according to the first embodiment of the device of the present invention;
[0025] Figure 4 It is a schematic diagram of a desulfurization system modeling device based on data drive according to the second embodiment of the device of the present invention. DETAILED DESCRIPTION
[0026] In order to solve the problems in the prior art, an embodiment of the present invention discloses a desulfurization system modeling method and device based on data-driven, specifically: by analyzing the combustion principle of the coal-fired unit boiler and the mechanism of the desulfurization system, the influencing factors related to the SO2 concentration at the desulfurization outlet are determined, and relevant historical operation data are collected to construct a training data set; the historical operation data is projected into a low-dimensional subspace by adopting the Locality Preserving Projections (LPP) method to achieve data dimensionality reduction while retaining the local structure; a one-dimensional convolutional neural network (1D-CNN) model is used to establish a dynamic prediction model for SO2 at the desulfurization outlet; the operation data related to the SO2 concentration at the desulfurization outlet at the current moment is obtained, and the predicted value of the SO2 concentration at the desulfurization outlet at the current moment is determined.
[0027] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0028] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.
[0029] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0030] Method Embodiment
[0031] According to an embodiment of the present invention, a desulfurization system modeling method based on data driving is provided. Figure 1 is a flow chart of a data-driven desulfurization system modeling method according to an embodiment of the present invention, such as Figure 1 As shown, the data-driven desulfurization system modeling method according to an embodiment of the present invention specifically includes:
[0032] Step 101, determine the influencing factors related to the SO2 concentration at the desulfurization outlet, collect relevant historical operation data based on the influencing factors, and construct a training data set according to the relevant historical operation data; in an embodiment of the present invention, the above-mentioned influencing factors specifically include at least one of the following: coal-fired unit load, flue gas volume, SO2 concentration at the absorption tower inlet, absorption tower inlet temperature, absorption tower inlet oxygen content, slurry pH value, slurry circulation pump current, slurry supply, slurry density, liquid-gas ratio, calcium-sulfur ratio, absorption tower outlet SO2 concentration, absorption tower outlet temperature, absorption tower outlet oxygen content.
[0033] In step 101, determining the influencing factors related to the SO2 concentration at the desulfurization outlet, collecting relevant historical operation data based on the influencing factors, and constructing a training data set based on the relevant historical operation data specifically includes:
[0034] By analyzing the combustion principle of coal-fired boilers and the mechanism of desulfurization systems, the influencing factors related to the SO2 concentration at the desulfurization outlet are determined;
[0035] Based on the influencing factors, collecting relevant historical operation data at predetermined time intervals from a distributed control system DCS system of the on-site desulfurization system;
[0036] The training data set is determined according to the characteristic parameters and the output value, wherein the characteristic parameters are influencing factors related to the SO2 concentration at the desulfurization outlet, and the output value is a time series signal composed of the SO2 concentration at the desulfurization outlet.
[0037] Step 102, projecting the relevant historical operation data in the training data set into a low-dimensional subspace by using a locality preserving projection method (LPP); step 102 specifically includes the following processing:
[0038] The relevant historical operation data in the training data set is subjected to an unsupervised linear dimensionality reduction algorithm, namely, the Locality Preserving Projections (LPP) method, and a graph with a close and distant relationship is constructed to reduce the dimensionality of the data:
[0039] Given R n There are m data points x1, x2, ..., x m , Locality Preserving Projections (LPP) maps these m data points to R by finding a transformation matrix Q t (t<<n) data points y1, y2, ..., y m :
[0040] yi =A T i , i=1,2,…,m Formula 1;
[0041] Among them, the matrix A is obtained by minimizing the objective function:
[0042]
[0043] Among them, W ij is a weight matrix representing two data points y1, y2, ..., y m The relationship between.
[0044] Step 103, based on the training data set, a one-dimensional convolutional neural network 1D-CNN model is used to establish a dynamic prediction model for SO2 concentration at the desulfurization outlet; in an embodiment of the present invention, step 103 specifically includes:
[0045] Build the input layer, convolution layer, pooling layer, and fully connected layer of the 1D-CNN model framework, wherein the input layer is a time series signal composed of the influencing factors; the convolution layer is used to extract the input data features, and its convolution operation * represents a one-dimensional convolution operation; c represents a convolution layer operation; k is the convolution kernel index; b is the number of convolution kernels; is the kth convolution kernel; s is the convolution kernel size, that is, the local window width; Represented by the convolution kernel The kth feature generated; b is the bias; f(x) is the activation function; the pooling layer is used to extract features twice, optimize the features extracted by the convolution layer, extract the main features, and reduce the dimension of the input data, including mean pooling and maximum pooling, pooling operation p represents the pooling layer operation; is the kth feature after pooling. After maximum pooling, its sequence length is halved. The fully connected layer consists of multiple hidden layers, which are used to further combine the extracted time series features. The final output is fc represents the fully connected layer operation; α fc is the output of the fully connected layer; W fc ∈q is the weight matrix of the fully connected layer; q is the number of neurons in the hidden layer; is the output result of the i-th sequence;
[0046] The constructed 1D-CNN model is trained to find the optimal parameters to minimize the loss function and characterize the deviation between the predicted value and the actual value. During training, the root mean square error is used to determine the loss of 1D-CNN. In addition, the Dropout strategy is used to modify the connection method of the 1D-CNN model itself so that some units do not update the weights during the back propagation process during the 1D-CNN model training. The Adagrad optimization algorithm is used to accelerate the training process of the 1D-CNN model and establish a dynamic prediction model for SO2 concentration at the desulfurization outlet.
[0047] Step 104, obtaining current operation data related to the SO2 concentration at the desulfurization outlet at the current moment, and determining a predicted value of the SO2 concentration at the desulfurization outlet at the current moment based on the dynamic prediction model for the SO2 concentration at the desulfurization outlet.
[0048] The above technical solution of the embodiment of the present invention is described in detail below.
[0049] The data-driven desulfurization system modeling method provided in accordance with an embodiment of the present invention specifically comprises the following steps:
[0050] Step 1: By analyzing the combustion principle of the coal-fired unit boiler and the mechanism of the desulfurization system, determine the influencing factors related to the SO2 concentration at the desulfurization outlet, collect relevant historical operation data, and construct a training data set;
[0051] Step 2: Project the historical operation data into a low-dimensional subspace by using the Locality Preserving Projections (LPP) method to achieve data dimensionality reduction while retaining the local structure;
[0052] Step 3, using a one-dimensional convolutional neural network (1D-CNN) model to establish a dynamic prediction model for SO2 concentration at the desulfurization outlet;
[0053] Step 4, obtaining the operating data related to the SO2 concentration at the desulfurization outlet at the current moment, and determining the predicted value of the SO2 concentration at the desulfurization outlet at the current moment.
[0054] The specific implementation process is as follows:
[0055] The influencing factors related to the SO2 concentration at the desulfurization outlet are mainly: coal-fired unit load, flue gas volume, SO2 concentration at the absorber inlet, absorber inlet temperature, absorber inlet oxygen content, slurry pH value, slurry circulation pump current, slurry supply, slurry density, liquid-gas ratio, calcium-sulfur ratio, absorber outlet SO2 concentration, absorber outlet temperature, absorber outlet oxygen content, etc. The acquisition of relevant historical data is mainly based on the mechanism analysis to determine the influencing factors of SO2 concentration, and the historical data with an interval of 1s is collected from the on-site DCS system. The training data set is determined by characteristic parameters and output values, the characteristic parameters are mainly the influencing factors related to the SO2 concentration at the desulfurization outlet, and the output is mainly a time series signal composed of the SO2 concentration at the desulfurization outlet.
[0056] The Locality Preserving Projections (LPP) method is an unsupervised linear dimensionality reduction algorithm that achieves data dimensionality reduction by constructing a graph with a close relationship. Specifically,
[0057] Given R n There are m data points x1,x2,...,x m , locality preserving projections (LPP) is used to find a transformation matrix Q to map these m data points to R t (t<<n) data points y1,y2,...,y m .
[0058] y i =A T x i , i=1,2,…,m Formula 1
[0059] The matrix A in the formula can be obtained by minimizing the objective function,
[0060]
[0061] Where W ij is a weight matrix representing two data points y1, y2, ..., y m The relationship between.
[0062] Figure 2 1D-CNN model establishment mainly includes the construction and training of the 1D-CNN model framework.
[0063] (1) 1D-CNN model framework construction
[0064] The 1D-CNN mainly includes input layer, convolution layer, pooling layer and fully connected layer. The input layer is the time series signal composed of relevant influencing factors; the convolution layer mainly realizes the extraction of input data features, and its convolution operation * represents a one-dimensional convolution operation; c represents a convolution layer operation; k is the convolution kernel index; b is the number of convolution kernels; is the kth convolution kernel; s is the convolution kernel size, that is, the local window width; Represented by the convolution kernel The kth feature generated; b is the bias; f(x) is the activation function, and commonly used activation functions include ReLu function, Sigmoid function and Tanh function; the pooling layer is mainly used to extract features for the second time, optimize the features extracted by the convolution layer, extract the main features, and reduce the dimension of the input data. It mainly includes mean pooling and maximum pooling. Pooling operation p represents the pooling layer operation; is the kth feature after pooling. After maximum pooling, its sequence length is halved. The fully connected layer consists of multiple hidden layers, which further combine the extracted time series features, and its final output is fc represents the fully connected layer operation; α fc is the output of the fully connected layer; W fc ∈q is the weight matrix of the fully connected layer; q is the number of neurons in the hidden layer; is the output result of the i-th sequence. So far, the entire 1D-CNN network has been built.
[0065] (2) 1D-CNN model training
[0066] The goal of the 1D-CNN neural network training is to find the best parameters (weights, biases, etc.) to minimize the loss function, which is used to characterize the deviation between the predicted value and the actual value. For the SO2 concentration regression problem at the desulfurization outlet, the root mean square error is often used to determine the loss of 1D-CNN.
[0067] The biggest problem with convolutional neural networks is that they are prone to overfitting. To prevent this problem, the Dropout strategy is used. This method modifies the connection mode of the 1D-CNN model itself. During the training of the 1D-CNN model, some units do not update the weights during the back propagation process, thereby reducing the complexity of the model. At the same time, in order to speed up the training process of 1D-CNN, the Adagrad optimization algorithm is usually used.
[0068] The operating data related to the SO2 concentration emission at the current moment is obtained, and the parameters at the current moment are input into the trained 1D-CNN neural network model to obtain the predicted value of the SO2 concentration at the outlet of the desulfurization system at the current moment.
[0069] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0070] 1. The embodiment of the present invention predicts the SO2 concentration value at the desulfurization outlet through multi-index data, and can predict the outlet SO2 concentration in advance to achieve the goal of early action.
[0071] 2. The embodiment of the present invention adopts the 1D-CNN method, which has the characteristics of fast calculation speed and strong calculation power. It has a strong characterization ability for desulfurization SO2 concentration data with time series characteristics and can improve the accuracy of outlet SO2 concentration prediction.
[0072] 3. The embodiment of the present invention provides a method combining Locality Preserving Projections (LPP) with 1D-CNN, which reduces the dimension of data and improves the efficiency and accuracy of desulfurization system modeling.
[0073] 4. The method provided in the embodiment of the present invention provides an effective desulfurization system modeling method, which can reduce material consumption during the operation of the desulfurization system and reduce operating costs.
[0074] Device Example 1
[0075] According to an embodiment of the present invention, a desulfurization system modeling device based on data drive is provided. Figure 3 Schematic diagram of a desulfurization system modeling device based on data drive according to the first embodiment of the device of the present invention, such as Figure 3 As shown, the data-driven desulfurization system modeling device according to an embodiment of the present invention specifically includes:
[0076] The acquisition module 30 is used to determine the influencing factors related to the SO2 concentration at the desulfurization outlet, collect relevant historical operation data based on the influencing factors, and construct a training data set based on the relevant historical operation data; the influencing factors specifically include at least one of the following:
[0077] Coal-fired unit load, flue gas volume, SO2 concentration at the absorber inlet, absorber inlet temperature, absorber inlet oxygen content, slurry pH value, slurry circulation pump current, slurry supply volume, slurry density, liquid-gas ratio, calcium-sulfur ratio, absorber outlet SO2 concentration, absorber outlet temperature, absorber outlet oxygen content.
[0078] The acquisition module 30 is specifically used for:
[0079] By analyzing the combustion principle of coal-fired boilers and the mechanism of desulfurization systems, the influencing factors related to the SO2 concentration at the desulfurization outlet are determined;
[0080] Based on the influencing factors, relevant historical operation data at predetermined time intervals are collected from the on-site DCS system.
[0081] The training data set is determined according to the characteristic parameters and the output value, wherein the characteristic parameters are influencing factors related to the SO2 concentration at the desulfurization outlet, and the output value is a time series signal composed of the SO2 concentration at the desulfurization outlet.
[0082] The projection module 32 is used to project the relevant historical operation data in the training data set into a low-dimensional subspace by using a locality preserving projection method (LPP); the projection module 32 is specifically used to:
[0083] The relevant historical operation data in the training data set is subjected to an unsupervised linear dimensionality reduction algorithm, namely, the Locality Preserving Projections (LPP) method, and a graph with a close and distant relationship is constructed to reduce the dimensionality of the data:
[0084] Given R n There are m data points x1,x2,...,x m , locality preserving projections (LPP) is used to find a transformation matrix Q to map these m data points to R t (t<<n) data points y1,y2,...,y m :
[0085] y i =A T x i , i=1,2,…,m Formula 1;
[0086] Among them, the matrix A is obtained by minimizing the objective function:
[0087]
[0088] Among them, W ij is a weight matrix representing two data points y1, y2, ..., y m the relationship between;
[0089] Establishing module 34, for establishing a dynamic prediction model of SO2 concentration at the desulfurization outlet based on the training data set by using a one-dimensional convolutional neural network 1D-CNN model; the establishing module 34 is specifically used for:
[0090] Build the input layer, convolution layer, pooling layer, and fully connected layer of the 1D-CNN model framework, wherein the input layer is a time series signal composed of the influencing factors; the convolution layer is used to extract the input data features, and its convolution operation * represents a one-dimensional convolution operation; c represents a convolution layer operation; k is the convolution kernel index; b is the number of convolution kernels; is the kth convolution kernel; s is the convolution kernel size, that is, the local window width; Represented by the convolution kernel The kth feature generated; b is the bias; f(x) is the activation function; the pooling layer is used to extract features twice, optimize the features extracted by the convolution layer, extract the main features, and reduce the dimension of the input data, including mean pooling and maximum pooling, pooling operation p represents the pooling layer operation; is the kth feature after pooling. After maximum pooling, its sequence length is halved. The fully connected layer consists of multiple hidden layers, which are used to further combine the extracted time series features. The final output is fc represents the fully connected layer operation; α fe is the output of the fully connected layer; W fc ∈q is the weight matrix of the fully connected layer; q is the number of neurons in the hidden layer; is the output result of the i-th sequence;
[0091] The constructed 1D-CNN model is trained to find the optimal parameters to minimize the loss function and characterize the deviation between the predicted value and the actual value. During training, the root mean square error is used to determine the loss of 1D-CNN. In addition, the Dropout strategy is used to modify the connection method of the 1D-CNN model itself so that some units do not update the weights during the back propagation process during the 1D-CNN model training. The Adagrad optimization algorithm is used to accelerate the training process of the 1D-CNN model and establish a dynamic prediction model for SO2 concentration at the desulfurization outlet.
[0092] The prediction module 36 is used to obtain the current operation data related to the SO2 concentration at the desulfurization outlet at the current moment, and determine the predicted value of the SO2 concentration at the desulfurization outlet at the current moment based on the dynamic prediction model of the SO2 concentration at the desulfurization outlet.
[0093] The embodiment of the present invention is a device embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.
[0094] Device Example 2
[0095] The embodiment of the present invention provides a desulfurization system modeling device based on data drive, such as Figure 4 As shown, it includes: a memory 40, a processor 42, and a computer program stored in the memory 40 and executable on the processor 42. When the computer program is executed by the processor 42, the steps described in the method embodiment are implemented.
[0096] Device Example 3
[0097] An embodiment of the present invention provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by the processor 42, the steps described in the method embodiment are implemented.
[0098] The computer-readable storage medium in this embodiment includes, but is not limited to, ROM, RAM, magnetic disk or optical disk, etc.
[0099] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0100] In the 1930s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.
[0101] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.
[0102] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0103] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0104] It should be understood by those skilled in the art that one or more embodiments of this specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0105] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0106] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0107] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0108] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0109] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0110] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0111] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0112] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.
[0113] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0114] The above description is only an embodiment of this document and is not intended to limit this document. For those skilled in the art, this document may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this document should be included in the scope of the claims of this document.
Claims
1. A data-driven desulfurization system modeling method, It is characterized in that include: Determine the SO2 2 concentration-related influencing factors, collecting relevant historical operating data based on the influencing factors, and constructing a training data set based on the relevant historical operating data; Projecting the relevant historical operation data in the training data set into a low-dimensional subspace by using the locality preserving projection method (LPP); specifically comprising: The relevant historical operation data in the training data set is subjected to an unsupervised linear dimensionality reduction algorithm, and a graph with a close and distant relationship is constructed to reduce the dimensionality of the data: Given R n There are m data points x in 1 ,x 2 ,...,x m , the local preservation projection method LPP is used to find a transformation matrix Q to map these m data points to R t Data point y on (t<<n) 1 ,y 2 ,...,y m : y i = A T x i , i = 1, 2, …, m Equation 1; Among them, the matrix A is obtained by minimizing the objective function: Among them, W ij is the weight matrix, representing two data points y 1 ,y 2 ,...,y m the relationship between; Based on the training data set, a one-dimensional convolutional neural network 1D-CNN model is used to establish the desulfurization export SO 2 Concentration dynamic prediction model; specifically including: Build the input layer, convolution layer, pooling layer, and fully connected layer of the 1D-CNN model framework, wherein the input layer is a time series signal composed of the influencing factors; the convolution layer is used to extract the input data features, and its convolution operation * represents a one-dimensional convolution operation; c represents a convolution layer operation; k is the convolution kernel index; b is the number of convolution kernels; is the kth convolution kernel; s is the convolution kernel size, that is, the local window width; Represented by the convolution kernel The kth feature generated; b is the bias; f(x) is the activation function; the pooling layer is used to extract features twice, optimize the features extracted by the convolution layer, extract the main features, and reduce the dimension of the input data, including mean pooling and maximum pooling, pooling operation p represents the pooling layer operation; is the kth feature after pooling. After maximum pooling, its sequence length is halved. The fully connected layer consists of multiple hidden layers, which are used to further combine the extracted time series features. The final output is fc represents the fully connected layer operation; α fc is the output of the fully connected layer; W fc ∈q is the weight matrix of the fully connected layer; q is the number of neurons in the hidden layer; is the output result of the i-th sequence; The constructed 1D-CNN model is trained to find the best parameters to minimize the loss function, characterizing the deviation between the predicted value and the actual value. During training, the root mean square error is used to determine the loss of 1D-CNN, and the Dropout strategy is used to modify the connection mode of the 1D-CNN model itself, so that some units do not update the weights during the back propagation process during the training of the 1D-CNN model. The Adagrad optimization algorithm is used to accelerate the training process of the 1D-CNN model, and the desulfurization export SO 2 Concentration dynamic prediction model; Get the current time and desulfurization export SO 2 The current operating data related to the concentration, based on the desulfurization outlet SO 2 The concentration dynamic prediction model determines the SO2 at the desulfurization outlet at the current moment. 2 Predicted concentration.
2. The method according to claim 1, It is characterized in that The influencing factors specifically include at least one of the following: Coal-fired unit load, flue gas volume, absorption tower inlet SO 2 concentration, absorption tower inlet temperature, absorption tower inlet oxygen content, slurry pH value, slurry circulation pump current, slurry supply, slurry density, liquid-gas ratio, calcium-sulfur ratio, absorption tower outlet SO 2 concentration, absorber outlet temperature, and absorber outlet oxygen content.
3. The method according to claim 1, It is characterized in that Determine the SO2 content at the desulfurization outlet 2 Influencing factors related to concentration, collecting relevant historical operation data based on the influencing factors, and constructing a training data set based on the relevant historical operation data specifically include: By analyzing the combustion principle of coal-fired boilers and the mechanism of desulfurization systems, it is determined that the SO 2 Concentration-related influencing factors; Based on the influencing factors, collecting relevant historical operation data at predetermined time intervals from a distributed control system DCS system of the on-site desulfurization system; The training data set is determined according to the characteristic parameters and the output value, wherein the characteristic parameters are related to the desulfurization outlet SO 2 Concentration-related influencing factors, the output value is the desulfurization outlet SO 2 The time series signal of concentration composition.
4. A data-driven desulfurization system modeling device, It is characterized in that include: Acquisition module, used to determine the SO 2 concentration-related influencing factors, collecting relevant historical operating data based on the influencing factors, and constructing a training data set based on the relevant historical operating data; A projection module is used to project the relevant historical operation data in the training data set into a low-dimensional subspace through LPP; the projection module is specifically used to: The relevant historical operation data in the training data set is subjected to an unsupervised linear dimensionality reduction algorithm, and a graph with a close and distant relationship is constructed to reduce the dimensionality of the data: Given R n There are m data points x in 1 ,x 2 ,...,x m , the local preservation projection method LPP is used to find a transformation matrix Q to map these m data points to R t Data point y on (t<<n) 1 ,y 2 ,...,y m : y i = A T x i , i = 1, 2, …, m Formula 1; Among them, the matrix A is obtained by minimizing the objective function: Among them, W ij is the weight matrix, representing two data points y 1 ,y 2 ,...,y m the relationship between; Establish a module for establishing a desulfurization outlet SO based on the training data set using a one-dimensional convolutional neural network 1D-CNN model 2 Concentration dynamic prediction model; the establishment module is specifically used for: Build the input layer, convolution layer, pooling layer, and fully connected layer of the 1D-CNN model framework, wherein the input layer is a time series signal composed of the influencing factors; the convolution layer is used to extract the input data features, and its convolution operation * represents a one-dimensional convolution operation; c represents a convolution layer operation; k is the convolution kernel index; b is the number of convolution kernels; is the kth convolution kernel; s is the convolution kernel size, that is, the local window width; Represented by the convolution kernel The kth feature generated; b is the bias; f(x) is the activation function; the pooling layer is used to extract features twice, optimize the features extracted by the convolution layer, extract the main features, and reduce the dimension of the input data, including mean pooling and maximum pooling, pooling operation p represents the pooling layer operation; is the kth feature after pooling. After maximum pooling, its sequence length is halved. The fully connected layer consists of multiple hidden layers, which are used to further combine the extracted time series features. The final output is fc represents the fully connected layer operation; α fc is the output of the fully connected layer; W fc ∈q is the weight matrix of the fully connected layer; q is the number of neurons in the hidden layer; is the output result of the i-th sequence; The constructed 1D-CNN model is trained to find the best parameters to minimize the loss function, characterizing the deviation between the predicted value and the actual value. During training, the root mean square error is used to determine the loss of 1D-CNN, and the Dropout strategy is used to modify the connection mode of the 1D-CNN model itself, so that some units do not update the weights during the back propagation process during the training of the 1D-CNN model. The Adagrad optimization algorithm is used to accelerate the training process of the 1D-CNN model, and the desulfurization export SO 2 Concentration dynamic prediction model; Prediction module, used to obtain the current time and desulfurization outlet SO 2 The current operating data related to the concentration, based on the desulfurization outlet SO 2 The concentration dynamic prediction model determines the SO at the desulfurization outlet at the current moment. 2 Predicted concentration.
5. The device according to claim 4, It is characterized in that The influencing factors specifically include at least one of the following: Coal-fired unit load, flue gas volume, absorption tower inlet SO 2 concentration, absorption tower inlet temperature, absorption tower inlet oxygen content, slurry pH value, slurry circulation pump current, slurry supply, slurry density, liquid-gas ratio, calcium-sulfur ratio, absorption tower outlet SO 2 concentration, absorber outlet temperature, and absorber outlet oxygen content; The acquisition module is specifically used for: By analyzing the combustion principle of coal-fired boilers and the mechanism of desulfurization systems, it is determined that the SO 2 Concentration-related influencing factors; Based on the influencing factors, collecting relevant historical operation data at predetermined time intervals from the on-site DCS system; The training data set is determined according to the characteristic parameters and the output value, wherein the characteristic parameters are related to the desulfurization outlet SO 2 Concentration-related influencing factors, the output value is the desulfurization outlet SO 2 The time series signal of concentration composition.
6. A data-driven desulfurization system modeling device, It is characterized in that include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the data-driven desulfurization system modeling method according to any one of claims 1 to 3 are implemented.
7. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores an implementation program for information transmission, and when the program is executed by a processor, the steps of the data-driven desulfurization system modeling method according to any one of claims 1 to 3 are implemented.
