Data prediction method and system applied to power plant equipment

Abstract

The application provides a data prediction method and system applied to a thermal power plant equipment, and relates to the technical field of artificial intelligence.In the application, operation data of a to-be-analyzed thermal power plant equipment is extracted, and key information mining operation is performed on the operation data to form a first key information description vector corresponding to the to-be-analyzed thermal power plant equipment; the operation data of the to-be-analyzed thermal power plant equipment is subjected to enhancement processing to form enhanced operation data of the to-be-analyzed thermal power plant equipment, and key information mining operation is performed on the enhanced operation data to form a second key information description vector corresponding to the to-be-analyzed thermal power plant equipment; and based on the first key information description vector and the second key information description vector corresponding to the to-be-analyzed thermal power plant equipment, the running state information corresponding to the to-be-analyzed thermal power plant equipment is predicted and associated.Based on the above, the reliability of data prediction can be improved to a certain extent.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to a data prediction method and system for thermal power plant equipment. Background Technology

[0002] In the operation monitoring of thermal power plant equipment, real-time monitoring and prediction of its operating status are necessary. Current monitoring can immediately handle accidents and prevent them from escalating, while status prediction is also crucial for proactive prevention and maintenance. However, in existing technologies, the prediction of operating status based on equipment operating data generally suffers from relatively low reliability. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a data prediction method and system for thermal power plant equipment, so as to improve the reliability of data prediction to a certain extent.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0005] A data prediction method for thermal power plant equipment includes:

[0006] The system extracts the operating data of the thermal power plant equipment to be analyzed, and performs key information mining operations on the operating data of the equipment to be analyzed to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed. The operating data of the equipment to be analyzed is formed by monitoring the operation process of the thermal power plant equipment to be analyzed.

[0007] The operating data of the equipment to be analyzed is enhanced to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed. In addition, key information mining operation is performed on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed.

[0008] Based on the first key information description vector and the second key information description vector corresponding to the thermal power plant equipment to be analyzed, the operating status information corresponding to the thermal power plant equipment to be analyzed is predicted by association.

[0009] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, include:

[0010] Using a key information mining model, the operating data of the device to be analyzed is subjected to key information mining operations to form a corresponding key information description vector cluster, which includes multiple key information description vectors.

[0011] Based on the key information description vector cluster, the enhanced operation key information description vector corresponding to the operation data of the device to be analyzed is determined;

[0012] Based on the enhanced operation key information description vector, the operation data of the equipment to be analyzed is enhanced to form the enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed.

[0013] The operational data of the enhanced equipment is subjected to key information mining operations to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed.

[0014] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the key information mining model includes multiple key information mining sub-models distributed in a node-like manner. The node-like distribution includes a distribution state where there is a connection between the data output side of at least one key information mining sub-model and the data input side of at least two key information mining sub-models. The key information description vector cluster includes the operational key information description vector mined by each of the multiple key information mining sub-models. The key information mining model includes multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction. The first direction and the second direction intersect. The first set of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models. The data output side of the a-th key information mining sub-model included in the first set of key information mining sub-models distributed in the first direction has a connection relationship with the data input side of the other sets of a-th key information mining sub-models distributed in the first direction.

[0015] The step of determining the enhanced operational key information description vector corresponding to the operational data of the device to be analyzed based on the key information description vector cluster includes:

[0016] Aggregate the running key information description vectors corresponding to each group of key information mining sub-models distributed in the first direction in the key information mining model to form a corresponding first direction aggregated description vector, thereby forming a corresponding first direction aggregated description vector cluster. The first direction aggregated description vector cluster includes a second number of first direction aggregated description vectors.

[0017] Aggregate the running key information description vectors corresponding to each set of key information mining sub-models distributed in the second direction in the key information mining model to form a corresponding second-direction aggregated description vector, thereby forming a corresponding second-direction aggregated description vector cluster. The second-direction aggregated description vector cluster includes a first number of second-direction aggregated description vectors.

[0018] Perform a row and column swap operation on the b-th aggregate description vector cluster in at least two aggregate description vector clusters to form corresponding row and column vector parameter swap data. The at least two aggregate description vector clusters include the first direction aggregate description vector cluster and the second direction aggregate description vector cluster.

[0019] The row and column vector parameter swapping data and the b-th aggregated description vector cluster are fused to form corresponding vector fused data;

[0020] The vector fusion data is loaded into the configured stimulus mapping output model to output the distribution of the b-th correlation parameter corresponding to the b-th aggregate description vector cluster.

[0021] The step of loading the vector fusion data into the configured stimulus mapping output model is executed multiple times to output the b-th correlation parameter distribution corresponding to the b-th aggregated descriptive vector cluster, thereby forming at least two correlation parameter distributions corresponding to the at least two aggregated descriptive vector clusters.

[0022] Based on the distribution of the bth correlation parameter corresponding to the bth aggregated description vector cluster in the at least two aggregated description vector clusters, the bth aggregated description vector cluster is multiplied to output the corresponding bth correlation fusion parameter distribution;

[0023] The step of performing the multiplication operation on the b-th correlation parameter distribution corresponding to the b-th aggregated description vector cluster in the at least two aggregated description vector clusters multiple times to output the corresponding b-th correlation fusion parameter distribution is used to form at least two correlation fusion parameter distributions corresponding to the at least two aggregated description vector clusters.

[0024] Based on the distribution of the at least two correlation fusion parameters, the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster are determined.

[0025] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the step of determining the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster based on the distribution of the at least two correlation fusion parameters includes:

[0026] Perform corresponding superposition operations on the at least two aggregated description vector clusters and the at least two correlation fusion parameter distributions respectively to output the corresponding correlation fusion description vector;

[0027] The correlation fusion description vector is subjected to vector dimension adjustment, and the correlation fusion description vector after vector dimension adjustment is subjected to filtering or convolution to form enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster.

[0028] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the key information mining model includes multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction.

[0029] The step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form corresponding key information description vector clusters includes:

[0030] The key information mining sub-models distributed in the first direction are used to perform key information mining operations on the operating data of the device to be analyzed, so as to form a first number of operating key information description vectors. The input data of the key information mining sub-models connected to the next key information mining sub-model in the first group of key information mining sub-models distributed in the first direction is the operating key information description vector mined by the adjacent key information mining sub-model. The first group of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models.

[0031] Each of the other key information mining sub-models distributed in the first direction performs key information mining operations on the first number of operational key information description vectors to form the corresponding first number of operational key information description vectors. The input data of the a-th key information mining sub-model of each of the other groups distributed in the first direction is the operational key information description vector mined by the a-th key information mining sub-model of the first group distributed in the first direction.

[0032] The third number of running key information description vectors mined by the multiple sets of key information mining sub-models distributed in the first direction and the multiple sets of key information mining sub-models distributed in the second direction are combined to form a key information description vector cluster. The third number is equal to the product of the first number and the second number. The multiple sets of key information mining sub-models distributed in the second direction are the second number.

[0033] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the first number of key information mining sub-models in each of the other groups of key information mining sub-models distributed in the first direction correspond to the first number of key information mining units. The key information mining units corresponding to the a-th key information mining sub-model are different among the different groups of key information mining sub-models included in each of the other groups of key information mining sub-models distributed in the first direction.

[0034] The steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, further include:

[0035] By using the key information mining unit corresponding to the a-th key information mining sub-model in each of the other key information mining sub-models distributed in the first direction, the running key information description vector mined by the a-th key information mining sub-model in the first group distributed in the first direction is subjected to expansion key information mining operation to form the corresponding a-th expansion key information mining data.

[0036] The step of performing key information mining operations on the first number of operational key information description vectors through each of the other sets of key information mining sub-models distributed in the first direction to form the corresponding first number of operational key information description vectors includes:

[0037] By using the a-th key information mining sub-model of each of the other sets of key information mining sub-models distributed in the first direction, the a-th expanded key information mining data is subjected to key information mining operation to form the corresponding a-th running key information description vector; the a-th key information mining sub-model is each key information mining sub-model in a set of key information mining sub-models to obtain the corresponding first number of running key information description vectors.

[0038] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the step of performing key information mining operations on the operating data of the equipment to be analyzed through a key information mining model to form a corresponding key information description vector cluster includes:

[0039] The feature space mapping unit performs feature space mapping on the operating data of the device to be analyzed to form a corresponding operating data space mapping result.

[0040] The key information mining model is used to perform key information mining operations on the spatial mapping results of the running data to form a corresponding key information description vector cluster.

[0041] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the step of enhancing the operating data of the equipment to be analyzed based on the enhanced operational key information description vector to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed includes:

[0042] The enhanced operational key information description vector is subjected to key information deep mining operation through the key information deep mining unit to form a corresponding deep key information description vector.

[0043] The depth key information description vector is interpolated by the information interpolation unit to output the corresponding interpolated key information description vector.

[0044] The key information restoration unit performs a key information restoration operation on the interpolated key information description vector to form enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed.

[0045] In some preferred embodiments, in the above-described data prediction method applied to thermal power plant equipment, the step of performing key information mining operations on the operating data of the equipment to be analyzed through a key information mining model to form a corresponding key information description vector cluster includes:

[0046] The feature space mapping unit performs feature space mapping on the operating data of the device to be analyzed to form a corresponding operating data space mapping result.

[0047] The key information mining model is used to perform key information mining operations on the spatial mapping results of the running data to form a corresponding key information description vector cluster.

[0048] The step of performing information interpolation on the depth key information description vector through the information interpolation unit to output the corresponding interpolated key information description vector includes:

[0049] The operation of superimposing the spatial mapping result of the running data and the depth key information description vector is performed to form a corresponding superimposed key information description vector;

[0050] The information interpolation unit performs an information interpolation operation on the superimposed key information description vector to output the corresponding interpolated key information description vector.

[0051] This invention also provides a data prediction system for thermal power plant equipment, including a processor and a memory. The memory stores a computer program, and the processor executes the computer program to implement the above-described data prediction method for thermal power plant equipment.

[0052] The data prediction method and system for thermal power plant equipment provided in this invention first extracts the operating data of the thermal power plant equipment to be analyzed, and performs key information mining operations on the operating data to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed; then, it performs enhancement processing on the operating data to form enhanced operating data corresponding to the thermal power plant equipment to be analyzed, and performs key information mining operations on the enhanced operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed; based on the first and second key information description vectors corresponding to the thermal power plant equipment to be analyzed, it predicts the operating status information corresponding to the thermal power plant equipment to be analyzed. Based on the above, because the operating data of the equipment to be analyzed is enhanced, the prediction of the operating status can fully combine both the operating data of the equipment to be analyzed and the enhanced operating data, thus improving the reliability of data prediction to a certain extent, thereby addressing the problem of relatively low reliability in data prediction in existing technologies.

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0054] Figure 1 This is a structural block diagram of a data prediction system for thermal power plant equipment provided in an embodiment of the present invention.

[0055] Figure 2 This is a flowchart illustrating the steps of the data prediction method for thermal power plant equipment provided in this embodiment of the invention.

[0056] Figure 3 This is a schematic diagram of the modules included in the data prediction device for thermal power plant equipment provided in an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0058] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0059] like Figure 1 As shown, this embodiment of the invention provides a data prediction system for use in thermal power plant equipment. The data prediction system may include a memory and a processor.

[0060] In detail, the memory and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, they can be electrically connected via one or more communication buses or signal lines. The memory may store at least one software functional module (computer program) that exists in the form of software or firmware. The processor can be used to execute the executable computer program stored in the memory, thereby implementing the data prediction method for thermal power plant equipment provided in this embodiment of the invention.

[0061] Alternatively, in some possible implementations, the memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0062] Alternatively, in some possible implementations, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0063] Alternatively, in some possible implementations, the data prediction system applied to thermal power plant equipment can be a server with data processing capabilities.

[0064] Combination Figure 2 This invention also provides a data prediction method for thermal power plant equipment, which can be applied to the aforementioned data prediction system for thermal power plant equipment. The method steps defined in the process of the data prediction method for thermal power plant equipment can be implemented by the data prediction system for thermal power plant equipment.

[0065] The following will be about Figure 2 The specific process shown will be explained in detail.

[0066] Step S110: Extract the operating data of the thermal power plant equipment to be analyzed, and perform key information mining operation on the operating data of the equipment to be analyzed to form the first key information description vector corresponding to the thermal power plant equipment to be analyzed.

[0067] In this embodiment of the invention, the data prediction system applied to thermal power plant equipment can extract the operating data of the thermal power plant equipment to be analyzed, and perform key information mining operations on the operating data to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed. The operating data of the equipment to be analyzed is formed by monitoring the operation process of the thermal power plant equipment to be analyzed, such as by extracting the equipment's operating log data. By performing convolution operations or encoding operations on the operating data of the equipment to be analyzed using a convolutional neural network, the first key information description vector corresponding to the thermal power plant equipment to be analyzed can be obtained.

[0068] Step S120: Enhance the operating data of the equipment to be analyzed to form enhanced operating data of the thermal power plant equipment to be analyzed; and perform key information mining operation on the enhanced operating data to form a second key information description vector of the thermal power plant equipment to be analyzed.

[0069] In this embodiment of the invention, the data prediction system applied to thermal power plant equipment can enhance the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and perform key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed. By performing convolution operations or encoding operations on the enhanced equipment operating data through a convolutional neural network, the second key information description vector corresponding to the thermal power plant equipment to be analyzed can be obtained.

[0070] Step S130: Based on the first key information description vector and the second key information description vector corresponding to the thermal power plant equipment to be analyzed, the operating status information corresponding to the thermal power plant equipment to be analyzed is predicted by association.

[0071] In this embodiment of the invention, the data prediction system applied to thermal power plant equipment can predict the operating status information of the thermal power plant equipment to be analyzed based on the first key information description vector and the second key information description vector corresponding to the equipment to be analyzed, such as indicating whether the equipment to be analyzed will experience a fault or abnormality. For example, the first key information description vector and the second key information description vector can be concatenated or superimposed to form a corresponding concatenated key information description vector. Then, the concatenated key information description vector can be analyzed and predicted to output the operating status information of the thermal power plant equipment to be analyzed. Furthermore, the extraction of the first key information description vector, the extraction of the second key information description vector, and the correlation prediction of the corresponding operating status information of the thermal power plant equipment to be analyzed can be achieved through a neural network. During the learning process, this neural network can learn from exemplary data and exemplary labels to obtain the mapping relationship between the two. The exemplary data may include exemplary equipment operating data and exemplary enhanced equipment operating data, and the exemplary labels may include the corresponding exemplary operating status information, enabling the neural network to possess the aforementioned processing capabilities. In addition, the optimization learning process of this neural network can be carried out separately from the learning process of the network model that enhances the operating data of the equipment to be analyzed. The network model can also be formed based on the exemplary equipment operating data and the exemplary enhanced equipment operating data as labels.

[0072] Based on the above, since the operating data of the equipment to be analyzed is enhanced, the operating status prediction can fully combine the operating data of the equipment to be analyzed and the enhanced operating data. Therefore, the reliability of data prediction can be improved to a certain extent, thereby improving the problem of relatively low reliability of data prediction in the existing technology.

[0073] Alternatively, in some possible implementations, the steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, i.e., step S120 in the above implementation, may further include the following specific implementation process:

[0074] The key information mining model is used to perform key information mining operations on the operating data of the device to be analyzed in order to form a corresponding key information description vector cluster. The key information description vector cluster includes multiple key information description vectors, which are formed through multiple key information mining operations.

[0075] Based on the key information description vector cluster, the enhanced operational key information description vector corresponding to the operating data of the device to be analyzed is determined, that is, the vector enhancement operation is performed based on the key information description vector cluster, which includes multiple key information description vectors.

[0076] Based on the enhanced operation key information description vector, the operation data of the equipment to be analyzed is enhanced to form the enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed.

[0077] The enhanced equipment operation data is subjected to key information mining operations (as described above) to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed.

[0078] Alternatively, in some possible implementations, the step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form a corresponding key information description vector cluster may further include the following specific implementation process:

[0079] The feature space mapping unit performs feature space mapping on the operating data of the device to be analyzed to form a corresponding operating data space mapping result. In other words, discrete operating data of the device to be analyzed can be mapped into the feature space and represented in the form of continuous vectors to obtain the operating data space mapping result.

[0080] The key information mining model is used to perform key information mining operations on the spatial mapping results of the running data to form a corresponding key information description vector cluster. The specific processing process of the key information mining model can be referred to in the relevant description below. For example, the processing of the running data of the device to be analyzed in the following text can be changed to the processing of the spatial mapping results of the running data.

[0081] Optionally, in some possible implementations, the key information mining model may include multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction. Based on this, the step of performing key information mining operations on the operating data of the device to be analyzed through the key information mining model to form corresponding key information description vector clusters may further include the following specific implementation process:

[0082] By using a first set of key information mining sub-models distributed in a first direction, the operating data of the device to be analyzed is subjected to key information mining operations to form a first number of operating key information description vectors. The input data of the key information mining sub-models connected to the next key information mining sub-model in the first set of key information mining sub-models distributed in the first direction is the operating key information description vector mined by the adjacent key information mining sub-model. The first set of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models. Thus, a first number of operating key information description vectors can be mined through the first number of key information mining sub-models. For example, the first key information mining sub-model in the first set of key information mining sub-models distributed in the first direction can mine the first operating key information description vector, the second key information mining sub-model in the first set of key information mining sub-models distributed in the first direction can mine the second operating key information description vector, the third key information mining sub-model in the first set of key information mining sub-models distributed in the first direction can mine the third operating key information description vector, and the fourth key information mining sub-model in the first set of key information mining sub-models distributed in the first direction can mine the fourth operating key information description vector.

[0083] Each of the other key information mining sub-models distributed along the first direction performs key information mining operations on the first number of operational key information description vectors to form corresponding first number of operational key information description vectors. The input data of the a-th key information mining sub-model in each of the other groups distributed along the first direction is the operational key information description vector mined by the a-th key information mining sub-model in the first group distributed along the first direction. For example, the input data of the first key information mining sub-model in the second group of key information mining sub-models distributed along the first direction is the operational key information description vector mined by the 1-th key information mining sub-model in the first group distributed along the first direction. The input data for the second key information mining sub-model in the second group of key information mining sub-models distributed in one direction is the operational key information description vector mined by the second key information mining sub-model in the first group distributed in the first direction. The input data for the first key information mining sub-model in the third group of key information mining sub-models distributed in the first direction is the operational key information description vector mined by the first key information mining sub-model in the first group distributed in the first direction. The input data for the second key information mining sub-model in the third group of key information mining sub-models distributed in the first direction is the operational key information description vector mined by the second key information mining sub-model in the first group distributed in the first direction.

[0084] The third number of operational key information description vectors mined by the multiple sets of key information mining sub-models distributed in the first direction and the multiple sets of key information mining sub-models distributed in the second direction are combined to form a key information description vector cluster. The third number is equal to the product of the first number and the second number. The multiple sets of key information mining sub-models distributed in the second direction are the second number. That is to say, the key information description vector cluster includes all the operational key information description vectors mined, that is, the operational key information description vectors mined by the third number of key information mining sub-models.

[0085] Optionally, in some possible implementations, each of the first number of key information mining sub-models in each of the other groups of key information mining sub-models distributed in the first direction corresponds to a first number of key information mining units. Among the different groups of key information mining sub-models included in each of the other groups of key information mining sub-models distributed in the first direction, the key information mining unit corresponding to the a-th key information mining sub-model is different. For example, the key information mining unit corresponding to the first key information mining sub-model in the second group of key information mining sub-models distributed in the first direction is different from the key information mining unit corresponding to the first key information mining sub-model in the third group of key information mining sub-models distributed in the first direction, and the key information mining unit corresponding to the second key information mining sub-model in the second group of key information mining sub-models distributed in the first direction is different from the key information mining unit corresponding to the second key information mining sub-model in the third group of key information mining sub-models distributed in the first direction. Furthermore, each of the other groups of key information mining sub-models distributed in the first direction may refer to other groups of key information mining sub-models besides the first group of key information mining sub-models distributed in the first direction. Based on this, the steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, i.e., the above-mentioned step S120, may further include the following specific implementation process:

[0086] By using the key information mining unit corresponding to the a-th key information mining sub-model in each of the other key information mining sub-models distributed in the first direction, the running key information description vector mined by the a-th key information mining sub-model in the first group distributed in the first direction is subjected to dilated key information mining operation to form the corresponding a-th dilated key information mining data. Here, the key information mining unit corresponding to the a-th key information mining sub-model in each of the other key information mining sub-models distributed in the first direction can be a dilated convolution unit (such as Dilated Convolution, which increases the receptive field of the output unit without increasing the number of parameters).

[0087] The step of performing key information mining operations on the first number of operational key information description vectors through each of the other sets of key information mining sub-models distributed in the first direction to form the corresponding first number of operational key information description vectors includes:

[0088] By using the a-th key information mining sub-model of each of the other sets of key information mining sub-models distributed in the first direction, the a-th expanded key information mining data is subjected to key information mining operation to form the corresponding a-th running key information description vector; the a-th key information mining sub-model is each key information mining sub-model in a set of key information mining sub-models (that is, each key information mining sub-model in a set of key information mining sub-models is used sequentially or in parallel as the a-th key information mining sub-model to perform key information mining operation), so as to obtain the corresponding first number of running key information description vectors.

[0089] Optionally, in some possible implementations, the step of performing key information mining operations on the a-th expanded key information mining data through the a-th key information mining sub-model of each of the other sets of key information mining sub-models distributed in the first direction to form the corresponding a-th running key information description vector may further include the following specific implementation process:

[0090] For the a-th key information mining sub-model of the second group of key information mining sub-models distributed in the first direction (the processing process of other key information mining sub-models can be the same), the a-th expanded key information mining data is filtered using the first filtering unit (which can be composed of at least one filtering matrix) included in the a-th key information mining sub-model to obtain a first filtered description vector. Then, the first nonlinear mapping unit included in the a-th key information mining sub-model is used to perform a nonlinear mapping operation on the first filtered description vector (such as through a sigmoid function) to obtain a corresponding first nonlinear mapping vector.

[0091] Using the second filtering unit included in the a-th key information mining sub-model, a filtering operation is performed on the superposition result of the first nonlinear mapping vector and the a-th expanded key information mining data to obtain a second filtered description vector; and using the second nonlinear mapping unit included in the a-th key information mining sub-model, a nonlinear mapping operation is performed on the second filtered description vector to obtain the corresponding second nonlinear mapping vector.

[0092] Using the third filtering unit included in the a-th key information mining sub-model, a filtering operation is performed on the superposition result of the second nonlinear mapping vector, the first nonlinear mapping vector, and the a-th expansion key information mining data to obtain a third filtering description vector. Also, using the third nonlinear mapping unit included in the a-th key information mining sub-model, a nonlinear mapping operation is performed on the third filtering description vector to obtain the corresponding third nonlinear mapping vector.

[0093] Using the linear integration unit (FC, Fully Connected) included in the a-th key information mining sub-model, a linear integration operation is performed on the superposition result of the third nonlinear mapping vector, the second nonlinear mapping vector, the first nonlinear mapping vector, and the a-th expanded key information mining data to obtain the corresponding linear integration vector. And using the fourth filtering unit included in the a-th key information mining sub-model, a filtering operation is performed on the linear integration vector to obtain the corresponding fourth filtering description vector.

[0094] The a-th expansion key information mining data and the fourth filter description vector are superimposed to form the corresponding a-th operation key information description vector.

[0095] Alternatively, in some possible implementations, the key information mining model may include multiple key information mining sub-models distributed in a node-like manner. The node-like distribution includes a distribution state where there is a connection between the data output side of at least one key information mining sub-model and the data input side of at least two key information mining sub-models. The key information description vector cluster includes the operational key information description vector mined by each of the multiple key information mining sub-models. The key information mining model includes multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction. The first direction and the second direction intersect; for example, they may present a multi-row, multi-column distribution. The first group of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models (the first group of key information mining sub-models distributed in the second direction includes a second number of cascaded key information mining sub-models). The data output side of the a-th key information mining sub-model in the first group of key information mining sub-models distributed in the first direction has a connection relationship with the data input side of the a-th key information mining sub-model in other groups distributed in the first direction (enabling data input and output). Based on this, the step of determining the enhanced operational key information description vector corresponding to the operational data of the device to be analyzed based on the key information description vector cluster can further include the following specific implementation process:

[0096] The key information mining model performs an aggregation operation on the running key information description vectors corresponding to each set of key information mining sub-models distributed in the first direction to form a corresponding first direction aggregated description vector, thereby forming a corresponding first direction aggregated description vector cluster. The first direction aggregated description vector cluster includes a second number of first direction aggregated description vectors. Each first direction aggregated description vector can aggregate a first number of running key information description vectors, such as {running key information description vector 1, running key information description vector 2, ..., running key information description vector n}. That is, the aggregation operation can refer to splicing.

[0097] The key information mining model performs an aggregation operation on the running key information description vectors corresponding to each set of key information mining sub-models distributed in the second direction to form a corresponding second-direction aggregated description vector, thereby forming a corresponding second-direction aggregated description vector cluster. The second-direction aggregated description vector cluster includes a first number of second-direction aggregated description vectors. Each second-direction aggregated description vector can aggregate a second number of running key information description vectors, such as {running key information description vector 1, running key information description vector 2, ..., running key information description vector m}. That is, the aggregation operation can refer to splicing.

[0098] A row-column swap operation is performed on the b-th aggregated description vector cluster in at least two aggregated description vector clusters to form corresponding row-column vector parameter swap data. The at least two aggregated description vector clusters include the first direction aggregated description vector cluster and the second direction aggregated description vector cluster. That is, it can include only the first direction aggregated description vector cluster and the second direction aggregated description vector cluster, or it can be aggregated in other ways based on including the first direction aggregated description vector cluster and the second direction aggregated description vector cluster to form other aggregated description vector clusters. Other aggregation methods can be one or more. In addition, the at least two aggregated description vector clusters can be used as the b-th aggregated description vector cluster sequentially or in parallel. Furthermore, the row-column swap operation can refer to swapping the rows and columns of the vector parameters in the vector.

[0099] The row and column vector parameter swap data and the b-th aggregated descriptive vector cluster are fused to form corresponding vector fused data. For example, the fusion operation can refer to the cross product operation, that is, the row and column vector parameter swap data and the b-th aggregated descriptive vector cluster are cross-producted. The aggregated descriptive vectors included in the b-th aggregated descriptive vector cluster can be distributed in a matrix manner. Thus, the vector fused data can also be a matrix.

[0100] The vector fusion data is loaded into the configured incentive mapping output model to output the distribution of the b-th correlation parameter corresponding to the b-th aggregate description vector cluster. The incentive mapping output model may include functions such as sigmoid.

[0101] The step of loading the vector fusion data into the configured excitation mapping output model and outputting the b-th correlation parameter distribution corresponding to the b-th aggregated descriptive vector cluster is executed multiple times to form at least two correlation parameter distributions corresponding to the at least two aggregated descriptive vector clusters. For example, when there are two aggregated descriptive vector clusters, this step can be executed twice. For example, the first execution of this step can obtain the correlation parameter distribution corresponding to the first direction aggregated descriptive vector cluster, and the second execution of this step can obtain the correlation parameter distribution corresponding to the second direction aggregated descriptive vector cluster.

[0102] Based on the distribution of the bth correlation parameter corresponding to the bth aggregate description vector cluster in the at least two aggregate description vector clusters, the bth aggregate description vector cluster is multiplied (e.g., the parameters at corresponding positions in the matrix are multiplied) to output the corresponding bth correlation fusion parameter distribution, such as outputting the first correlation fusion parameter distribution and the second correlation fusion parameter distribution.

[0103] The step of performing the multiplication operation on the b-th correlation parameter distribution corresponding to the b-th aggregated description vector cluster in the at least two aggregated description vector clusters multiple times to output the corresponding b-th correlation fusion parameter distribution is used to form at least two correlation fusion parameter distributions corresponding to the at least two aggregated description vector clusters.

[0104] Based on the distribution of the at least two correlation fusion parameters, the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster are determined. In other words, enhancement can be performed based on the distribution of the at least two correlation fusion parameters.

[0105] Alternatively, in some possible implementations, the step of determining the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster based on the distribution of the at least two correlation fusion parameters may further include the following specific implementation process:

[0106] The at least two aggregated description vector clusters and the at least two correlation fusion parameter distributions are respectively subjected to corresponding superposition operations to output the corresponding correlation fusion description vector. For example, the first aggregated description vector cluster and the correlation fusion parameter distribution corresponding to the first aggregated description vector cluster can be superimposed (both are matrices, and the matrix parameters at the corresponding positions can be summed; or, the vector dimensions can be adjusted first, and then the superposition operation can be performed) to form the first correlation fusion description vector. The second aggregated description vector cluster and the correlation fusion parameter distribution corresponding to the second aggregated description vector cluster can be superimposed to form the second correlation fusion description vector.

[0107] The correlation fusion description vector is subjected to a vector dimension adjustment operation (which can be adjusted to a target dimension, which can be configured according to actual needs). The correlation fusion description vector after the vector dimension adjustment operation is subjected to a filtering operation or a convolution operation, such as processing through a corresponding filtering matrix, to form an enhanced running key information description vector corresponding to multiple running key information description vectors in the key information description vector cluster.

[0108] Alternatively, in some possible implementations, the step of enhancing the operating data of the equipment to be analyzed based on the enhanced operational key information description vector to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed may further include the following specific implementation process:

[0109] The key information deep mining unit performs key information deep mining operation on the enhanced operation key information description vector to form a corresponding deep key information description vector. For example, the key information deep mining unit may include multiple convolution kernels to perform convolution operation on the enhanced operation key information description vector to obtain the deep key information description vector.

[0110] The information interpolation unit performs information interpolation on the depth key information description vector to output the corresponding interpolated key information description vector. The specific interpolation parameters of the information interpolation unit can be formed in the optimization learning of the corresponding neural network.

[0111] The key information restoration unit performs a key information restoration operation on the interpolated key information description vector to form the enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed. The key information restoration unit can be a decoding network used to decode the interpolated key information description vector to form the corresponding decoded equipment operation data. In this way, the decoded equipment operation data can be used as the corresponding enhanced equipment operation data.

[0112] Alternatively, in some possible implementations, as mentioned above, the step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form a corresponding key information description vector cluster can further include the following specific implementation process: performing a feature space mapping operation on the operational data of the device to be analyzed using a feature space mapping unit to form a corresponding operational data space mapping result; and performing key information mining operations on the operational data space mapping result using the key information mining model to form a corresponding key information description vector cluster. Based on this, the step of performing information interpolation operations on the deep key information description vector using an information interpolation unit to output a corresponding interpolated key information description vector can further include the following specific implementation process:

[0113] The operation is performed by superimposing the running data space mapping result and the depth key information description vector to form a corresponding superimposed key information description vector. As mentioned above, the running data space mapping result can be represented as a vector. Therefore, the two vectors can be superimposed.

[0114] The information interpolation unit performs information interpolation on the superimposed key information description vector to output the corresponding interpolated key information description vector. In this way, by combining the initially obtained running data space mapping result, the gradient problem caused by deep mining can be resolved.

[0115] Combination Figure 3This invention also provides a data prediction device for thermal power plant equipment, which can be applied to the aforementioned data prediction system for thermal power plant equipment. The data prediction device for thermal power plant equipment may include:

[0116] The first key information mining module is used to extract the operating data of the thermal power plant equipment to be analyzed, and to perform key information mining operations on the operating data of the equipment to be analyzed to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed. The operating data of the equipment to be analyzed is formed by monitoring the operating process of the thermal power plant equipment to be analyzed.

[0117] The second key information mining module is used to enhance the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and to perform key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed.

[0118] The operation status association prediction module is used to predict the operation status information of the thermal power plant equipment to be analyzed based on the first key information description vector and the second key information description vector corresponding to the equipment to be analyzed.

[0119] In summary, the data prediction method and system for thermal power plant equipment provided by this invention can first extract the operating data of the thermal power plant equipment to be analyzed, and then perform key information mining operations on the operating data to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed; then perform enhancement processing on the operating data to form enhanced operating data corresponding to the thermal power plant equipment to be analyzed, and then perform key information mining operations on the enhanced operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed; based on the first and second key information description vectors corresponding to the thermal power plant equipment to be analyzed, the operating status information corresponding to the thermal power plant equipment to be analyzed is predicted. Based on the above, because the operating data to be analyzed is enhanced, the prediction of the operating status can fully combine both the operating data of the equipment to be analyzed and the enhanced operating data, thus improving the reliability of data prediction to a certain extent, thereby addressing the problem of relatively low reliability in data prediction in existing technologies.

[0120] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A data prediction method applied to thermal power plant equipment, characterized in that, include: The system extracts the operating data of the thermal power plant equipment to be analyzed, and performs key information mining operations on the operating data of the equipment to be analyzed to form a first key information description vector corresponding to the thermal power plant equipment to be analyzed. The operating data of the equipment to be analyzed is formed by monitoring the operation process of the thermal power plant equipment to be analyzed. The operating data of the equipment to be analyzed is enhanced to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed. In addition, key information mining operation is performed on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed. Based on the first key information description vector and the second key information description vector corresponding to the thermal power plant equipment to be analyzed, the operating status information corresponding to the thermal power plant equipment to be analyzed is predicted by association. The steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, include: Using a key information mining model, the operating data of the device to be analyzed is subjected to key information mining operations to form a corresponding key information description vector cluster, which includes multiple key information description vectors. Based on the key information description vector cluster, the enhanced operation key information description vector corresponding to the operation data of the device to be analyzed is determined; Based on the enhanced operation key information description vector, the operation data of the equipment to be analyzed is enhanced to form the enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed. The key information mining operation is performed on the enhanced equipment operation data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed; The key information mining model includes multiple key information mining sub-models distributed in a node-like manner. The node-like distribution includes a distribution state where there is a connection between the data output side of at least one key information mining sub-model and the data input side of at least two key information mining sub-models. The key information description vector cluster includes the operational key information description vector mined by each of the multiple key information mining sub-models. The key information mining model includes multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction. The first direction and the second direction intersect. The first set of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models. The data output side of the a-th key information mining sub-model in the first set of key information mining sub-models distributed in the first direction is connected to the data input side of the a-th key information mining sub-model in the other sets distributed in the first direction. The step of determining the enhanced operational key information description vector corresponding to the operational data of the device to be analyzed based on the key information description vector cluster includes: Aggregate the running key information description vectors corresponding to each group of key information mining sub-models distributed in the first direction in the key information mining model to form a corresponding first direction aggregated description vector, thereby forming a corresponding first direction aggregated description vector cluster. The first direction aggregated description vector cluster includes a second number of first direction aggregated description vectors. Aggregate the running key information description vectors corresponding to each set of key information mining sub-models distributed in the second direction in the key information mining model to form a corresponding second-direction aggregated description vector, thereby forming a corresponding second-direction aggregated description vector cluster. The second-direction aggregated description vector cluster includes a first number of second-direction aggregated description vectors. Perform a row and column swap operation on the b-th aggregate description vector cluster in at least two aggregate description vector clusters to form corresponding row and column vector parameter swap data. The at least two aggregate description vector clusters include the first direction aggregate description vector cluster and the second direction aggregate description vector cluster. The row and column vector parameter swapping data and the b-th aggregated description vector cluster are fused to form corresponding vector fused data; The vector fusion data is loaded into the configured stimulus mapping output model to output the distribution of the b-th correlation parameter corresponding to the b-th aggregate description vector cluster. The step of loading the vector fusion data into the configured stimulus mapping output model is executed multiple times to output the b-th correlation parameter distribution corresponding to the b-th aggregated descriptive vector cluster, thereby forming at least two correlation parameter distributions corresponding to the at least two aggregated descriptive vector clusters. Based on the distribution of the bth correlation parameter corresponding to the bth aggregated description vector cluster in the at least two aggregated description vector clusters, the bth aggregated description vector cluster is multiplied to output the corresponding bth correlation fusion parameter distribution; The step of performing the multiplication operation on the b-th correlation parameter distribution corresponding to the b-th aggregated description vector cluster in the at least two aggregated description vector clusters multiple times to output the corresponding b-th correlation fusion parameter distribution is used to form at least two correlation fusion parameter distributions corresponding to the at least two aggregated description vector clusters. Based on the distribution of the at least two correlation fusion parameters, the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster are determined.

2. The data prediction method for thermal power plant equipment as described in claim 1, characterized in that, The step of determining the enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster based on the distribution of the at least two correlation fusion parameters includes: Perform corresponding superposition operations on the at least two aggregated description vector clusters and the at least two correlation fusion parameter distributions respectively to output the corresponding correlation fusion description vector; The correlation fusion description vector is subjected to vector dimension adjustment, and the correlation fusion description vector after vector dimension adjustment is subjected to filtering or convolution to form enhanced operational key information description vectors corresponding to multiple operational key information description vectors in the key information description vector cluster.

3. The data prediction method for thermal power plant equipment as described in claim 1, characterized in that, The key information mining model includes multiple sets of key information mining sub-models distributed in a first direction and multiple sets of key information mining sub-models distributed in a second direction. The step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form a corresponding key information description vector cluster includes: The key information mining sub-models distributed in the first direction are used to perform key information mining operations on the operating data of the device to be analyzed, so as to form a first number of operating key information description vectors. The input data of the key information mining sub-models connected to the next key information mining sub-model in the first group of key information mining sub-models distributed in the first direction is the operating key information description vector mined by the adjacent key information mining sub-model. The first group of key information mining sub-models distributed in the first direction includes a first number of cascaded key information mining sub-models. Each of the other key information mining sub-models distributed in the first direction performs key information mining operations on the first number of operational key information description vectors to form the corresponding first number of operational key information description vectors. The input data of the a-th key information mining sub-model of each of the other groups distributed in the first direction is the operational key information description vector mined by the a-th key information mining sub-model of the first group distributed in the first direction. The third number of running key information description vectors mined by the multiple sets of key information mining sub-models distributed in the first direction and the multiple sets of key information mining sub-models distributed in the second direction are combined to form a key information description vector cluster. The third number is equal to the product of the first number and the second number. The multiple sets of key information mining sub-models distributed in the second direction are the second number.

4. The data prediction method for thermal power plant equipment as described in claim 3, characterized in that, The first number of key information mining sub-models in each of the other groups of key information mining sub-models distributed in the first direction correspond to the first number of key information mining units. The key information mining units corresponding to the a-th key information mining sub-model are different among the different groups of key information mining sub-models included in each of the other groups of key information mining sub-models distributed in the first direction. The steps of enhancing the operating data of the equipment to be analyzed to form enhanced equipment operating data corresponding to the thermal power plant equipment to be analyzed, and performing key information mining operations on the enhanced equipment operating data to form a second key information description vector corresponding to the thermal power plant equipment to be analyzed, further include: By using the key information mining unit corresponding to the a-th key information mining sub-model in each of the other key information mining sub-models distributed in the first direction, the running key information description vector mined by the a-th key information mining sub-model in the first group distributed in the first direction is subjected to expansion key information mining operation to form the corresponding a-th expansion key information mining data. The step of performing key information mining operations on the first number of operational key information description vectors through each of the other sets of key information mining sub-models distributed in the first direction to form the corresponding first number of operational key information description vectors includes: By using the a-th key information mining sub-model of each of the other sets of key information mining sub-models distributed in the first direction, the a-th expanded key information mining data is subjected to key information mining operation to form the corresponding a-th running key information description vector; the a-th key information mining sub-model is each key information mining sub-model in a set of key information mining sub-models to obtain the corresponding first number of running key information description vectors.

5. The data prediction method for thermal power plant equipment as described in claim 1, characterized in that, The step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form a corresponding key information description vector cluster includes: The feature space mapping unit performs feature space mapping on the operating data of the device to be analyzed to form a corresponding operating data space mapping result. The key information mining model is used to perform key information mining operations on the spatial mapping results of the running data to form a corresponding key information description vector cluster.

6. The data prediction method for thermal power plant equipment as described in claim 1, characterized in that, The step of enhancing the operating data of the equipment to be analyzed based on the enhanced key information description vector to form enhanced operating data corresponding to the thermal power plant equipment to be analyzed includes: The enhanced operational key information description vector is subjected to key information deep mining operation through the key information deep mining unit to form a corresponding deep key information description vector. The depth key information description vector is interpolated by the information interpolation unit to output the corresponding interpolated key information description vector. The key information restoration unit performs a key information restoration operation on the interpolated key information description vector to form the enhanced equipment operation data corresponding to the thermal power plant equipment to be analyzed.

7. The data prediction method for thermal power plant equipment as described in claim 6, characterized in that, The step of performing key information mining operations on the operational data of the device to be analyzed using a key information mining model to form a corresponding key information description vector cluster includes: The feature space mapping unit performs feature space mapping on the operating data of the device to be analyzed to form a corresponding operating data space mapping result. The key information mining model is used to perform key information mining operations on the spatial mapping results of the running data to form a corresponding key information description vector cluster. The step of performing information interpolation on the depth key information description vector through the information interpolation unit to output the corresponding interpolated key information description vector includes: The operation of superimposing the spatial mapping result of the running data and the depth key information description vector is performed to form a corresponding superimposed key information description vector; The information interpolation unit performs an information interpolation operation on the superimposed key information description vector to output the corresponding interpolated key information description vector.

8. A data prediction system for use in thermal power plant equipment, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, and the processor being used to execute the computer program to implement the data prediction method for thermal power plant equipment as described in any one of claims 1-7.

Images