Dynamic coal blending method based on improved back propagation neural network

Abstract

The invention relates to the technical field of intelligent coal blending, and discloses a dynamic coal blending method based on an improved back-propagation neural network, which comprises the following steps: acquiring coal blending historical data of different furnace types of a power plant and dividing the data into a plurality of categories; the method comprises the following steps: preprocessing coal blending historical data of different furnace types, dividing the coal blending historical data into a training set, a verification set and a test set, training a preset BP model to obtain a basic model, performing model testing on the basic model, and establishing a target model; raw coal information is collected to serve as input of the target model, and after analysis and calculation of the target model, a coal as fired proportioning scheme with the lowest comprehensive cost is output. According to the method, the BP model is optimized by utilizing the dynamic cuckoo search algorithm, the coal proportioning scheme with the lowest comprehensive cost of the coal as fired is obtained under the influence of factors such as a production process, boiler design and coal type characteristics, and the problems that the calculation index coefficient is fixed and cannot be further popularized are solved; and the problem that a global optimal solution cannot be obtained when the optimal solution is obtained by using an exhaustion method is solved.

Description

Technical Field

[0001] The invention relates to the technical field of intelligent coal blending, and in particular to a power coal blending method based on an improved back propagation neural network. Background Art

[0002] With the rapid development of social economy, energy supply is of vital importance. Since the index range of the coal entering the furnace required for power steam production is strict, coal from a single source cannot meet the needs of industrial production, which will cause problems such as low energy conversion rate and waste of resources; the factors affecting power coal blending mainly include production process, boiler design, and coal characteristics. Among them, the main indicators of coal types are calorific value, moisture, ash content, volatile matter, sulfur content, and in some furnace types, the Hardgrove grindability index and softening temperature need to be additionally considered. The comprehensive index of the coal entering the furnace needs to achieve the lowest comprehensive cost (comprehensive cost = raw material cost of coal entering the furnace + water treatment cost + sulfur treatment cost + ash treatment cost) under the premise that all indicators are met. In the process of proportion calculation, the sulfur and ash content of the coal entering the furnace must also be taken into account to ensure that the exhaust gas emissions meet the standards and reduce the cost of sulfur treatment and ash treatment.

[0003] Traditional power coal blending relies on experienced coal blending engineers and there are uncertainties caused by human factors. Coal blending based on power intelligent coal blending algorithm reduces human intervention and reliance on experience, making coal blending more scientific and efficient.

[0004] Coal blending is one of the important research areas in the power generation industry. Current research mainly focuses on mathematical models and mechanism models. Existing research has proposed a mathematical model for power coal blending, which uses a generalized regression neural network for prediction and an exhaustive method to select coal types with high cost performance. It aims to minimize the proportion of high-quality coal, maximize the proportion of low-quality coal, and minimize the unit price of raw coal. The model does not fully consider the actual operating conditions of the boiler, nor the deviations generated during the production process, and does not combine the professional experience of senior coal blending engineers. Therefore, only local optimal solutions can be found. In addition, there are studies that propose a power coal blending model for thermal power plants, which aims to determine the most economically efficient coal blending strategy. The model mainly guides the coal blending process by analyzing the mechanism model of the boiler, but the coefficients in some of the formulas are fixed, lacking self-learning and correction mechanisms. Therefore, there may be a risk of not being able to obtain the optimal blending solution. Summary of the invention

[0005] The present invention provides a method for power coal blending based on an improved back propagation neural network, which optimizes the BP model using a dynamic cuckoo search algorithm, and takes into account factors such as production technology, boiler design, and coal type characteristics, to obtain a coal blending scheme that meets various production index requirements and has the lowest comprehensive cost of coal entering the furnace, thereby solving the problem that the existing calculation index coefficients are fixed and cannot be further generalized, and the problem that the global optimal solution cannot be obtained when the exhaustive method is used to obtain the optimal solution.

[0006] The present invention provides a method for blending power coal based on an improved back propagation neural network, comprising:

[0007] Acquire historical coal blending data covering different furnace types provided by a power plant, and divide the historical coal blending data into multiple categories;

[0008] The coal blending history data of the different furnace types are sorted to form a data set for model training, and the data set is divided into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, and use the test set and the validation set to perform model testing on the basic model to establish a target model;

[0009] The raw coal information is collected as the input of the target model, and after analysis and calculation by the target model, a furnace coal proportioning scheme with the lowest comprehensive cost is output; wherein the raw coal information is inventory information or purchasable information.

[0010] Furthermore, after the step of obtaining the historical coal blending data covering different furnace types provided by the power plant and dividing the historical coal blending data into multiple categories, it also includes data cleaning of the historical coal blending data. The data cleaning method is to process missing values, that is, to organize the data by deletion, interpolation or filling with the mean / median to ensure the uniformity of the data in format, unit and range.

[0011] Furthermore, after the step of collating the coal blending history data of the different furnace types to form a data set for model training and dividing it into a training set, a validation set and a test set, it also includes:

[0012] Features that are helpful for model learning are extracted from the coal blending history data, screened, and analyzed to what extent the features contribute to the model prediction results.

[0013] Furthermore, the coal blending history data of different furnace types are collated to form a data set for model training, and the data set is divided into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, and use the test set and the validation set to perform model testing on the basic model. In the step of establishing the target model, the preset BP neural network model uses the coal information as the original input for feature training, and the objective function is defined as:

[0014] min W=P T X+P S +B

[0015] Among them, W is the comprehensive cost of coal entering the furnace, X is the quality of a single type of coal, P T is the unit price of a single type of coal, P S is the secondary cost after production, and B is the bias term of the coal blending model.

[0016] Furthermore, in the preset BP model, the dynamic cuckoo search algorithm is used for optimization, and the specific implementation of the algorithm is as follows:

[0017]

[0018] in, represents the kth iteration solution of the i-th nest in the cuckoo algorithm, α is the step size factor, and Lévy(θ) is the mathematical expression of the Lévy flight mode, which is:

[0019]

[0020] in, Represents the best solution of the kth iteration, μ and v satisfy the normal distribution, and the algorithm generates a solution in each iteration. The solution is generated by the position update equation, and the new solution is used to replace the original randomly generated solution.

[0021] Furthermore, in the process of optimizing the BP model by using the dynamic cuckoo search algorithm, the solution obtained by the dynamic cuckoo search algorithm optimization represents the initial weights and thresholds of the BP neural network structure, and by iteratively optimizing each layer of neurons in the BP neural network, the weights and thresholds of the connections are obtained, and the BP neural network structure is obtained after distribution; by inputting the training samples into the BP neural network, the neural network undergoes repeated learning iterations, and finally the error between the output value and the expected value meets the requirements of the coal blending conditions; at the same time, the dynamic cuckoo search algorithm will continue to optimize the BP neural network until the furnace coal ratio is found and then stop.

[0022] Furthermore, after the step of collecting raw coal information as input of the target model and outputting the coal type ratio scheme with the lowest input and output coal cost after calculation by the target model, the step further includes:

[0023] The collected actual production data is compared and analyzed with the preset BP model output data, and the information is fed back to the preset BP model. The preset BP model performs self-learning based on the information and continuously optimizes its own performance.

[0024] The present invention also provides a power coal blending device based on an improved back propagation neural network, comprising:

[0025] An acquisition module, used for acquiring historical coal blending data covering different furnace types provided by a power plant, and dividing the historical coal blending data into multiple categories;

[0026] A sorting module is used to sort the coal blending history data of the different furnace types to form a data set for model training, and divide it into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, use the test set and the validation set to perform model testing on the basic model, and establish a target model;

[0027] The output module is used to collect raw coal information as the input of the target model, and output the coal type ratio scheme with the lowest input and output coal cost after calculation by the target model; wherein the raw coal information is inventory information or purchasable information.

[0028] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0029] The present invention also provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the steps of the above method when executed by a processor.

[0030] The beneficial effects of the present invention are:

[0031] The present invention aims to obtain the lowest cost coal blending for furnace entry, and proposes a back propagation neural network power coal blending method; the BP model is optimized by using a dynamic cuckoo search algorithm, and the coal blending scheme with the lowest comprehensive cost of furnace coal is obtained by taking into account factors such as production process, boiler design, and coal type characteristics. In the process of model construction, the professional knowledge and production experience of coal blending engineers are incorporated to improve the accuracy and scientificity of coal blending. In addition, the model takes into account the characteristics of different boilers and can use historical data and continuous production data for self-learning to enhance the adaptability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The figure is a schematic diagram of a method flow according to an embodiment of the present invention.

[0033] Figure 2 A hierarchical graph is constructed for a BP neural network according to an embodiment of the present invention.

[0034] Figure 3 FIG. 1 is a schematic diagram of a device structure according to an embodiment of the present invention.

[0035] Figure 4 The figure is a schematic diagram of the internal structure of a computer device according to an embodiment of the present invention.

[0036] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0037] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0038] The present invention relates to a process of using a variety of coal types to improve energy conversion rate and increase economic benefits in a power production steam link, and a power intelligent coal blending algorithm is specially designed.

[0039] The main purpose of the present invention is to propose a coal blending algorithm for power plants based on an optimized BP (back propagation network) neural network, which can obtain the lowest cost coal blending ratio for power plant production under the premise of meeting production indicators. The present invention can solve the problem that the calculation index coefficient is fixed and cannot be further generalized, and the problem that the global optimal solution cannot be obtained when the exhaustive method is used to obtain the optimal solution, and reduce the demand for hardware equipment for the exhaustive method.

[0040] like Figure 1 As shown, the present invention provides a method for power coal blending based on an improved back propagation neural network, comprising:

[0041] S1. Obtain historical coal blending data covering different furnace types provided by a power plant, and divide the historical coal blending data into multiple categories.

[0042] The original data comes from several large listed power plants, using the actual production data in recent years as the data source, which is highly representative of the industry. The original data (historical coal blending data) is purified, irrelevant and abnormal data are removed, and missing values ​​are processed by deletion, interpolation, or filling with mean / median. Ensure the uniformity of data in format, unit and range to prevent analysis errors caused by data inconsistency.

[0043] S2. Organize the historical data of coal blending of different furnace types to form a data set for model training, and divide it into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, use the test set and the validation set to test the basic model, and establish a target model.

[0044] Features that are helpful for model learning are extracted from the raw data, and the features are screened to analyze the contribution of the features to the model prediction results; the key point of the realization of the power intelligent coal blending algorithm is the construction of the model. The present invention uses a back propagation neural network for modeling. The back propagation neural network is a multi-layer feedforward network that is trained according to the error back propagation algorithm and uses historical data for learning and training to determine the mapping relationship between network input factors and prediction results. The BP neural network has the characteristics of self-learning, self-organization, self-adaptation, fast training speed, local approximation and global convergence. The preset BP neural network model uses coal information as the original input for feature training, and the objective function is defined as:

[0045] min W=P T X+P S +B

[0046] Among them, W is the comprehensive cost of coal entering the furnace, X is the quality of a single type of coal, P T is the unit price of a single type of coal, P S is the secondary cost after production (including water treatment cost, ash treatment cost and sulfur treatment cost), B is the bias term of the coal blending model, which includes the business experience of coal blending engineers, historical coal blending data learning experience and model self-learning experience.

[0047] In the preset BP model, the dynamic cuckoo search algorithm (DCS) is used to optimize the BP neural network to avoid the BP neural network from falling into the local minimum problem and to improve the speed of BP neural network training convergence. The specific implementation of the algorithm is as follows:

[0048]

[0049] in, represents the kth iteration solution of the i-th nest in the cuckoo algorithm, α is the step size factor, and Lévy(θ) is the mathematical expression of the Lévy flight mode, which is:

[0050]

[0051] in, Represents the best solution of the kth iteration, μ and v satisfy the normal distribution, and the algorithm generates a solution in each iteration. The solution is generated by the position update equation, and the new solution is used to replace the original randomly generated solution.

[0052] The introduction of the cuckoo search algorithm can retain the optimal solution finally obtained through continuous iteration. During the iterative optimization process, it is necessary to carefully search in a small range, make short-distance attempts, and occasionally walk long distances to reach another range to continue searching. It avoids the algorithm from falling into a local optimal solution and prevents repeated searches in the same range. Short-distance search can improve the quality of the solution, and long-distance search can avoid local optimization of the solution and improve the diversity of the solution.

[0053] BP neural network construction level Figure 2 As shown in the figure, in the process of citing the dynamic cuckoo search algorithm to optimize the BP model (DCS-BP algorithm), the solution optimized by the dynamic cuckoo search algorithm (DCS algorithm) represents the initial weight and threshold of the BP neural network structure, and obtains the connection weight and threshold by iteratively optimizing each layer of neurons in the BP neural network, and obtains the BP neural network structure after distribution; by inputting the training samples into the BP neural network, the neural network undergoes repeated learning iterations, and finally makes the error between the output value and the expected value meet the requirements of the coal blending conditions; at the same time, the dynamic cuckoo search algorithm will continue to optimize the BP neural network until the furnace coal ratio is found and then stop.

[0054] After establishing the objective function and the overall algorithm framework, it is necessary to adjust the model's hyperparameters, including selecting a suitable activation function, setting an appropriate learning rate, determining the loss function, etc. At the same time, the practical experience and professional business knowledge of coal blending experts should be structured and integrated into the training process of the model bias term. This can ensure that the professional knowledge of business personnel is effectively utilized, thereby improving the reliability and accuracy of the model.

[0055] S3. After the model training is completed, the raw coal information (which may be inventory information or purchase information) is collected as the input of the target model, and after analysis and calculation by the target model, the coal ratio scheme for the furnace with the lowest comprehensive cost is output.

[0056] S4. After the production process is completed, the collected actual production data is compared and analyzed with the preset BP model output data, and this information is fed back to the preset BP model. The preset BP model performs self-learning based on this information and continuously optimizes its own performance to better adapt to the production needs of specific furnace types, thereby reducing the prediction deviation caused by differences in furnace types.

[0057] Based on the above core technology, the present invention provides a power coal blending algorithm based on optimized feedforward neural network, which can obtain the coal type blending scheme with the lowest comprehensive furnace coal cost. It has been put into practice in a large power plant and significantly reduced the furnace coal cost.

[0058] The present invention aims to obtain the lowest cost coal blending for the furnace, and proposes a back propagation neural network power coal blending method; the dynamic cuckoo search algorithm is used to optimize the BP model, and the production process, boiler design, coal type characteristics and other factors are taken into consideration to obtain a coal blending scheme with the lowest comprehensive cost of coal for the furnace and various production indicators. In the process of model construction, the professional knowledge and production experience of coal blending engineers are incorporated to improve the accuracy and scientificity of coal blending. In addition, the model takes into account the characteristics of different boilers, and can use historical data and continuous production data for self-learning to enhance the adaptability of the model, thereby improving the accuracy and scientificity of coal blending.

[0059] The present invention adopts a method based on optimizing BP neural network to construct an optimized coal blending model for coal-to-steam production, which is trained based on historical production data of power coal blending. At the same time, the model has the ability to self-learn according to actual production data to correct the prediction deviation caused by different furnace types. The algorithm model is used to guide the power coal blending work to ensure that the coal index entering the furnace is within the specified range. At the same time, by comprehensively considering the production-related secondary costs such as water treatment, sulfur treatment, and ash treatment, the coal blending plan with the lowest comprehensive cost of coal entering the furnace can be output.

[0060] like Figure 3 As shown, the present invention provides a power coal blending device based on an improved back propagation neural network, comprising:

[0061] An acquisition module 1 is used to acquire historical coal blending data covering different furnace types provided by a power plant, and divide the historical coal blending data into multiple categories;

[0062] Arrangement module 2, used for arranging the coal blending history data of the different furnace types to form a data set for model training, and dividing it into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, use the test set and the validation set to perform model testing on the basic model, and establish a target model;

[0063] The output module 3 is used to collect raw coal information as the input of the target model, and output the coal type ratio scheme with the lowest input and output coal cost after calculation by the target model; wherein the raw coal information is inventory information or purchasable information.

[0064] In one embodiment, it further includes:

[0065] The data cleaning module is used to clean the coal blending historical data. The data cleaning method is to process the missing values, that is, to sort them out by deletion, interpolation or filling with mean / median values, so as to ensure the uniformity of data in format, unit and range.

[0066] In one embodiment, it further includes:

[0067] A screening module is used to extract features that are helpful for model learning from the coal blending history data, and to screen and analyze the contribution of the features to the model prediction results;

[0068] In one embodiment, in the sorting module 2, the preset BP neural network model performs feature training by taking the coal information as the original input, and the objective function is defined as:

[0069] min W=P T X+P S +B

[0070] Among them, W is the comprehensive cost of coal entering the furnace, X is the quality of a single type of coal, P T is the unit price of a single type of coal, P S is the secondary cost after production, and B is the bias term of the coal blending model.

[0071] In one embodiment, in the sorting module 2, the preset BP model uses a dynamic cuckoo search algorithm to optimize, so as to avoid the BP neural network from falling into the local minimum problem, and at the same time improve the speed of BP neural network training convergence. The specific implementation of the algorithm is as follows:

[0072]

[0073] in, represents the kth iteration solution of the i-th nest in the cuckoo algorithm, α is the step size factor, and Lévy(θ) is the mathematical expression of the Lévy flight mode, which is:

[0074]

[0075] in, Represents the best solution of the kth iteration, μ and v satisfy the normal distribution, and the algorithm generates a solution in each iteration. The solution is generated by the position update equation, and the new solution is used to replace the original randomly generated solution.

[0076] In one embodiment, in the sorting module 2, the solution obtained by the dynamic cuckoo search algorithm optimization represents the initial weights and thresholds of the BP neural network structure, and by iteratively optimizing each layer of neurons in the BP neural network, the connection weights and thresholds are obtained, and the BP neural network structure is obtained after distribution; by inputting the training samples into the BP neural network, the neural network undergoes repeated learning iterations, and finally the error between the output value and the expected value reaches the requirement that meets the coal blending conditions; at the same time, the dynamic cuckoo search algorithm will continue to optimize the BP neural network until the furnace coal ratio is found and then stop.

[0077] In one embodiment, it further includes:

[0078] The optimization module is used to compare and analyze the collected actual production data with the preset BP model output data, and feed back this information to the preset BP model. The preset BP model performs self-learning based on this information and continuously optimizes its own performance.

[0079] The above modules are used to execute the corresponding steps in the above-mentioned method for power coal blending based on improved back propagation neural network. The specific implementation method thereof is described in the above-mentioned method embodiment and will not be repeated here.

[0080] like Figure 4 As shown, the present invention also provides a computer device, which can be a server, and its internal structure can be as shown in Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store all data required for the process of the improved back-propagation neural network power coal blending method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the improved back-propagation neural network power coal blending method is implemented.

[0081] Those skilled in the art will understand that Figure 4 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0082] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for power coal blending based on an improved back propagation neural network.

[0083] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0084] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0085] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for blending power coal based on improved back propagation neural network, characterized in that: include: Acquire historical coal blending data covering different furnace types provided by a power plant, and divide the historical coal blending data into multiple categories; The coal blending history data of the different furnace types are sorted to form a data set for model training, and the data set is divided into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, and use the test set and the validation set to perform model testing on the basic model to establish a target model; The raw coal information is collected as the input of the target model, and after analysis and calculation by the target model, a furnace coal proportioning scheme with the lowest comprehensive cost is output; wherein the raw coal information is inventory information or purchasable information.

2. The method for blending coal for power generation based on an improved back propagation neural network according to claim 1, characterized in that: After the steps of obtaining the historical coal blending data covering different furnace types provided by the power plant and dividing the historical coal blending data into multiple categories, the method also includes data cleaning of the historical coal blending data. The data cleaning method is to process missing values, that is, to organize the data by deletion, interpolation or filling with the mean / median to ensure the uniformity of the data in format, unit and range.

3. The method for blending coal for power generation based on improved back propagation neural network according to claim 1, characterized in that: After the step of collating the coal blending history data of the different furnace types to form a data set for model training and dividing it into a training set, a validation set and a test set, the method further includes: Features that are helpful for model learning are extracted from the coal blending history data, screened, and analyzed to what extent the features contribute to the model prediction results.

4. The method for blending coal for power generation based on improved back propagation neural network according to claim 1, characterized in that: The coal blending history data of different furnace types are collated to form a data set for model training, and divided into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, and use the test set and the validation set to perform model testing on the basic model. In the step of establishing the target model, the preset BP neural network model uses the coal information as the original input for feature training, and the objective function is defined as: min W=P T X+P S +B Among them, W is the comprehensive cost of coal entering the furnace, X is the quality of a single type of coal, P T is the unit price of a single type of coal, P S is the secondary cost after production, and B is the bias term of the coal blending model.

5. The coal blending method based on improved back propagation neural network power according to claim 4 is characterized in that: In the preset BP model, the dynamic cuckoo search algorithm is used for optimization, and the specific implementation of the algorithm is as follows: in, represents the kth iteration solution of the i-th nest in the cuckoo algorithm, α is the step size factor, and Lévy(θ) is the mathematical expression of the Lévy flight mode, which is: in, Represents the best solution of the kth iteration, μ and v satisfy the normal distribution, and the algorithm generates a solution in each iteration. The solution is generated by the position update equation, and the new solution is used to replace the original randomly generated solution.

6. The coal blending method based on improved back propagation neural network power according to claim 5, characterized in that: In the process of optimizing the BP model by using the dynamic cuckoo search algorithm, the solution obtained by the dynamic cuckoo search algorithm represents the initial weights and thresholds of the BP neural network structure, and obtains the connection weights and thresholds by iteratively optimizing each layer of neurons in the BP neural network, and obtains the BP neural network structure after distribution; by inputting the training samples into the BP neural network, the neural network undergoes repeated learning iterations, and finally makes the error between the output value and the expected value meet the requirements of the coal blending conditions; at the same time, the dynamic cuckoo search algorithm will continue to optimize the BP neural network until the furnace coal ratio is found and then stop.

7. The method for blending coal for power generation based on improved back propagation neural network according to claim 1, characterized in that: After the step of collecting raw coal information as the input of the target model and outputting the coal type ratio scheme with the lowest input and output coal cost after calculation by the target model, the method further includes: The collected actual production data is compared and analyzed with the preset BP model output data, and the information is fed back to the preset BP model. The preset BP model performs self-learning based on the information and continuously optimizes its own performance.

8. A power coal blending device based on an improved back propagation neural network, characterized in that: include: An acquisition module, used for acquiring historical coal blending data covering different furnace types provided by a power plant, and dividing the historical coal blending data into multiple categories; A sorting module is used to sort the coal blending history data of the different furnace types to form a data set for model training, and divide it into a training set, a validation set and a test set, so as to use the training set to train the preset BP model to obtain a basic model, use the test set and the validation set to perform model testing on the basic model, and establish a target model; The output module is used to collect raw coal information as the input of the target model, and output the coal type ratio scheme with the lowest input and output coal cost after calculation by the target model; wherein the raw coal information is inventory information or purchasable information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images