An artificial intelligence power operation behavior control method and system based on image recognition
Through the multi-linear data system and component gradient mapping algorithm, the video image data of power operations is converted from single discrete coordinate data to multi-linear vector data, solving the problem of single data source and form in power operations behavior monitoring, and achieving efficient intelligence Integrated control and artificial intelligence data adaptation have significantly improved monitoring accuracy and detailed correlation.
Patent Information
- Application Number
- CN202410093223.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-23
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-01-23
AI Technical Summary
The data source and data form in the existing power operation video image monitoring system is single, resulting in limited information and it is difficult to achieve efficient intelligent control of power operation behavior.
Through multilinear parameterization of data configuration and algorithm, a multilinear data system is built, and the discrete spatial coordinate data is converted into multilinear vector data using gradient iteration and component gradient mapping algorithms, which is adapted to artificial intelligence data training and iteration.
The accuracy and detailed correlation of power operation behavior monitoring have been improved, and the generated high-order data configuration can be effectively adapted to artificial intelligence platforms such as TensorFlow, significantly improving the monitoring level.
Abstract
Description
Technical Field
[0001] The present invention relates to a smart grid and information control technology for electric power operations, and in particular to an electric power operation behavior control system and method adapted to artificial intelligence based on data processing and algorithm improvement. Background Art
[0002] At present, from the current situation and development direction of power grid safety operation, it is of great significance to further strengthen the informatization identification of production site operation risks based on video and image data, improve the technical system of power production operation site risks, and build information and intelligent technical tools that are equivalent to the national grid system indicator system such as source prevention, hierarchical management and control, and result-oriented + process assessment, in order to effectively improve the standardization and safety level of power operation sites. From a technical perspective, the traditional algorithm data source and data form used in the intelligent monitoring of power operation sites based on video image acquisition and its subsequent safety prevention system are single, generally based on the dynamic images collected by the video system and the spatial position coordinate database of the subject and object of the power operation target generated thereby, and this spatial position coordinate database is generally the most basic discrete coordinate site data, that is, a discrete two-dimensional array or three-dimensional array. This primary unprocessed data can show very limited information. This means that the successful development of an intelligent control system for power operation behavior based on image recognition must be a multi-dimensional reconstruction of the underlying data. Summary of the invention
[0003] The technical problem to be solved by the present invention is to overcome the above-mentioned technical deficiencies and to provide an electric power operation behavior data system and method based on data processing and algorithm improvement and highly adapted to artificial intelligence.
[0004] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows.
[0005] An artificial intelligence-based multi-linear data intelligent control system for power operation behavior based on image recognition. Aiming at the current situation of single algorithm data source and single data form in conventional power operation video image monitoring scenarios, the system is based on existing power operation video image on-site recording data or a networked power operation video image platform that has been aggregated and uploaded. Guided by the multi-linear parameterization of data configuration and data algorithm, the system is oriented towards and adapted to artificial intelligence data feeding training and its intelligent iterative evolution under a multi-parameter data model, and constructs a multi-linear data system in the power operation video image data scenario. The system sets a core multi-linear data traversal allocation process, and performs gradient iteration on the discrete space coordinate data group in the existing power operation video image data system on a multi-linear dimensional basis, thereby obtaining a multi-linear vector data system for power operation behavior monitoring that is highly adapted to artificial intelligence data training and data iteration.
[0006] As a preferred technical solution of the present invention, when the discrete spatial coordinate data group in the existing power operation video image data system is plane or spatial vector data, multiple discrete spatial coordinate data groups belonging to the same power operation event are merged to obtain a 2×n / 3×n configuration array.
[0007] As a preferred technical solution of the present invention, when the discrete spatial coordinate data group in the existing power operation video image data system is multidimensional vector data, multiple discrete spatial coordinate data groups belonging to the same power operation event are merged to obtain a k×n configuration; for the k×n configuration, k=2 / 3+k′, k′ is: At present, some power operation behavior video monitoring systems have added other power or electrical monitoring data in addition to the spatial video information uploaded by the on-site monitoring equipment. At this time, the k′ additionally written and recorded other power or electrical monitoring data can be included in this multilinear data system for integrated data processing as long as the data constraints are met.
[0008] As a preferred technical solution of the present invention, the gradient iteration utilizes component gradient mapping to construct a new data configuration that is compatible with the subject and object of the power operation in the power operation event and the components obtained after they are expanded in multiple linear dimensions. This new data configuration covers the process data of the power operation event and the detail association data from local to global.
[0009] As a preferred technical solution of the present invention, the gradient iteration mines the data connotation of the original discrete space coordinate data group or its merged array based on the component gradient mapping algorithm; the obtained data has a configuration expanded in multiple dimensions; the component gradient mapping algorithm is specifically: for a 2×n / 3×n configuration or k×n configuration array, take the change value of 2 / 3 / k or n data in any row or column at a selected time interval in the time dimension, and at the same time, pre-select a row or column as a multilinear dimensional basis to act as an independent variable data, take the 2 / 3 / k or n data of this independent variable row or column in the time dimension The change value at the time interval selected on the scale is used as the change increment and the latter as the independent variable increment for gradient linear allocation under the selected data operation mode; wherein, since the independent variable increment and the change increment are both derived and obtained based on any row or column data, they each have a vector data configuration. The difference is that when performing component gradient mapping, any component of the change increment vector is allocated to each independent variable increment under the selected data operation mode. In this way, a change increment vector generates a two-dimensional, open multilinear data, i.e., a second-order tensor array, after undergoing a component gradient mapping.
[0010] As a preferred technical solution of the present invention, for a given original discrete spatial coordinate data set vector in an existing power operation video image data system, if it has been merged into a digital matrix, then its row / column vector is taken, and the application of the single component gradient mapping algorithm enables the original data to be expanded in r dimensions, r=2 / 3 / k or r=n, that is, a single scalar data is decomposed into r components carrying specific local detail information according to its selected dimensional basis and the specified data operation mode (usually number division), and these local component information has local correlation compared to the row or column of the mapped dimensional basis; then the application of the s-times component gradient mapping algorithm corresponds to decomposing the single scalar data into r components according to its selected dimensional basis and the specified data operation mode (usually number division). s components that carry specific local detail information, and each local component information also has local correlation; finally, when s traverses all given original discrete space coordinate data group vectors in the existing power operation behavior video image data system (specific invalid data with the main diagonal / sub-diagonal being 1 can be eliminated in advance), the final high-order open data configuration carries global correlation information, and its data configuration is directly adapted to the data feeding training and intelligent iteration of existing artificial intelligence platforms such as TensorFlow.
[0011] As a preferred technical solution of the present invention, the selection of rows or columns is based on the row and column layout originally adopted when constructing the matrix in claim 2, and the row and column correspondence or the row and column transposition correspondence two component gradient mapping data processing processes to generate two multilinear data models with different data connotations.
[0012] As a preferred technical solution of the present invention, for two groups of change values, when the time intervals are selected in the time dimension when the two groups of values are obtained, completely uniform and consistent event intervals are maintained. Uniformity means that the time intervals in different time periods are the same, and consistency means that the time intervals used when the two groups of values are obtained are the same.
[0013] As a preferred technical solution of the present invention, the selected data operation mode is usually scalar number division, and in some special cases, scalar subtraction or other operations are used. At this time, a specific component gradient mapping tensor or a variation component gradient mapping tensor is obtained, because the gradient is classically defined as being obtained based on multiple scalar number division operations.
[0014] As a preferred technical solution of the present invention, if the original discrete space coordinate data group of the component gradient mapping has been merged into a two-dimensional array in advance, then in fact, a component gradient mapping can obtain a multilinear data array spread out in three linear dimensions, that is, a third-order tensor array.
[0015] As a preferred technical solution of the present invention, if the original discrete space coordinate data set itself has vector properties, the gradient iteration of the above discrete space coordinate data set on a multi-linear dimensional basis obtains second-order tensor data after one iteration.
[0016] As a preferred technical solution of the present invention, multiple iterations are performed to obtain high-order data and the order of the high-order data set in a single data system for monitoring a single target power operation should be kept globally consistent, that is, all discrete space coordinate data groups monitored by the target power operation behavior should be subjected to gradient iterations of the same number of times and in the same order.
[0017] The beneficial effects of adopting the above technical solution are as follows: see the specific implementation method below for details. The present invention solves the problem of single data source and data form of traditional video image monitoring algorithms. The data configuration is opened by sequentially and interactively (in a formulated operation mode such as division) data component allocation between related data groups under multi-linear dimensions. The resulting high-order data configuration contains interactive correlation detail information from local to global within the specified event of the power operation behavior. In particular, the resulting high-order data configuration is highly compatible with artificial intelligence platforms such as TensorFlow, and can collaborate in data feeding training and (based on data accumulation) intelligent iteration. Compared with the traditional single parameter comparison (monitoring parameter and the specified threshold parameter or threshold parameter range) algorithm, it can greatly improve the monitoring level of power operation behavior. DETAILED DESCRIPTION
[0018] The following examples illustrate the present invention in detail. Various raw materials and equipment used in the present invention are conventional commercial products, which can be directly obtained by market purchase. In the description of the following embodiments, specific details such as specific system structures, technologies, etc. are proposed for illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted to prevent unnecessary details from hindering the description of the present application. It should be understood that when used in the present application specification and the attached claims, the term "including" indicates the presence of the described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections. It should also be understood that the term "and / or" used in the present application specification and the attached claims refers to any combination of one or more of the items listed in association and all possible combinations, and includes these combinations. As used in the specification of the present application and the appended claims, the term "if" can be interpreted as "when..." or "once" or "in response to determination" or "in response to detection" according to the context. Similarly, the phrase "if it is determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once it is determined" or "in response to determination" or "once [the described condition or event] is detected" or "in response to detection [the described condition or event]" according to the context. In addition, in the description of the specification of the present application and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. The reference "one embodiment" or "some embodiments" described in the specification of the present application means that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Thus, the sentences "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "include", "comprising", "having" and variations thereof all mean "including but not limited to", unless specifically emphasized otherwise. Embodiment 0
[0019] From a technical perspective, the existing intelligent monitoring of power operation sites based on video image acquisition and its subsequent safety prevention system use traditional algorithm data sources and data formats that are single, and are generally based on dynamic images acquired by the video system and the spatial position coordinate database of the power operation target subject and object generated thereby. This spatial position coordinate database is generally the most basic discrete coordinate site data, that is, a discrete two-dimensional array or three-dimensional array of bi-data pairs / tri-data pairs; this primary unprocessed data can only display very limited information.
[0020] On the one hand, in the data analysis model before the artificial intelligence system is connected, in order to solve the problem of data roughness of the spatial position coordinate database, it is generally only possible to adjust the data discriminant threshold Dis-criteria (Discriminant-criteria) of the data discriminator from a point value or a narrow range of values to a very wide range of values. This adjustment process can only be done through continuous manual trial and error to gradually find a target value range that seems to be the most suitable; obviously, the wider the value range of the discriminant threshold Dis-criteria, the less target data it can identify and discriminate, because only those outside the value range of the discriminant threshold Dis-criteria can be identified and discriminated. Only the data that falls within the Dis-criteria range can be processed, while the target data within the Dis-criteria range can only be marked as pending data and then manually inspected and verified. It can be seen that the narrower the Dis-criteria range, the more accurate the data discriminability, and the wider the Dis-criteria range, the more ambiguous the data discriminability. Conversely, due to the coarseness of the basic data, if the Dis-criteria range is reduced, although the scale of discriminable data is increased, the error rate of the discrimination result will be greatly increased, which is even more unacceptable. In short, the coarseness of the basic data directly leads to the ambiguity in identifying and discriminating the various on-site subjects and objects of power operation behaviors.
[0021] On the other hand, when these data information are used for the intervention of artificial intelligence systems, they are fed as training data to artificial intelligence models such as TensorFlow, and often fail to achieve the expected training results. For example, the butterfly effect of data input and output is prone to occur, that is, the artificial intelligence model trained with the above-mentioned rough elementary coordinate data often produces very different data outputs when the approximate data input is used during trial use, which often means that the data training objectives of the artificial intelligence model are not achieved. Fundamentally, in addition to the adjustment of the artificial intelligence model data training parameters, the main focus is on providing the data and quality of the training data. Here, the quantity and scale of the training data generally use the existing video and image database of the power system, and there is often no more room for improvement; therefore, the most important research and development improvement direction is to improve the quality of the basic database of power operation behavior. Example 1
[0022] The discrete space coordinate data group in the existing power operation video image data system is plane or space vector data, and multiple discrete space coordinate data groups belonging to the same power operation event are combined to obtain a 2×n / 3×n configuration array;
[0023] The selection of the 2×n or 3×n configuration is associated with the discrete spatial coordinate data set recorded in the existing power operation video data system, that is, a two-dimensional data bi-data structure or a three-dimensional data tri-data structure.
[0024] Among them, n is associated with the subject and object of the current target power operation event and the power operation behavior contained in its constraints. Example 2
[0025] The discrete space coordinate data group in the existing power operation video image data system is multi-dimensional vector data. Multiple discrete space coordinate data groups belonging to the same power operation event are combined to obtain a k×n configuration matrix.
[0026] Among them, for the k×n configuration, k=2 / 3+k′, k′ is: At present, some power operation behavior video monitoring systems have added other power or electrical monitoring data in addition to the spatial video information uploaded by the on-site monitoring equipment. At this time, the k′ additionally written and recorded other power or electrical monitoring data can be included in this multi-linear data system for integrated data processing as long as the data constraints are met.
[0027] Among them, n is associated with the subject and object of the current target power operation event and the power operation behavior contained in its constraints, which remains unchanged. Example 3
[0028] The component gradient mapping algorithm mines the data content of the original discrete space coordinate data group or its merged array. The gradient iteration uses component gradient mapping to construct a new data configuration that is compatible with the subject and object of the power operation in the power operation event and the components obtained after expanding it in multiple linear dimensions. This new data configuration covers the process data of the power operation event and the detailed association data from local to global.
[0029] The data process of the component gradient mapping algorithm includes:
[0030] T1. For a 2×n / 3×n configuration or k×n configuration number array, take the change value of 2 / 3 / k or n data in any row or column at the selected time interval in the time dimension;
[0031] T2. At the same time, a row or column is selected in advance as the multilinear dimension basis to serve as the independent variable data, and the change value of 2 / 3 / k or n data of this independent variable row or column at the selected time interval on the time dimension is taken;
[0032] T3, using the former as a variable increment and the latter as an independent variable increment to perform gradient linear distribution under the selected data operation mode; wherein the selected data operation mode is usually scalar number division, and scalar subtraction or other operations are used in some special cases, and a specific component gradient mapping tensor or a variation component gradient mapping tensor is obtained at this time, because the gradient is classically defined as being obtained based on multiple scalar number division operations; wherein, since the independent variable increment and the variable increment are both derived and obtained based on any row or column data, they each have a vector data configuration, and the difference is that when performing component gradient mapping, any component of the variable increment vector is respectively distributed to each independent variable increment under the selected data operation mode;
[0033] It is worth noting here that: in t1, the selection of rows or columns is based on the original row and column layout used when constructing the number array in Example 1 or 2. The two are consistent to form a component gradient mapping data processing process, and the intersection of the two constitutes another component gradient mapping data processing process; the above two component gradient mapping data processing processes have data connotation mining and data detail display effects at different angles and different levels, that is, the two component gradient mappings are unequal; therefore, it is allowed to traverse the above two component gradient mapping data processing processes simultaneously or successively, so that the amount of basic data that can be fed into the artificial intelligence system as training data is doubled; further, since the data connotation mining and data detail display effects of the two component gradient mapping data processing are unequal, the artificial intelligence training effect of the bilinear data system is better than that of the single data system training, and the degree of effect optimization is much higher than the two-fold relationship;
[0034] In addition, in t1, when a single data system needs to be used for training due to other reasons such as system capacity or computing power limitations, it is preferred to use a multilinear system generated when the row or column layout remains consistent;
[0035] In t1 and t2, the event intervals are completely uniform and consistent when the time intervals are selected in the time dimension when the two sets of values are obtained. Uniformity means that the time intervals of different periods are the same, and consistency means that the time intervals used when the two sets of values are obtained are the same.
[0036] In t3, it can be seen that a variable increment vector generates a two-dimensional, open multilinear data after undergoing a component gradient mapping, that is, a second-order tensor array; if the original discrete space coordinate data group of the component gradient mapping has been merged into a two-dimensional array in advance, then in fact, a component gradient mapping will obtain a multilinear data array opened in three linear dimensions; among them, the row selection corresponds to 2 / 3 / k, and the column selection corresponds to n; this correspondence allows mutual conversion according to the initial row and column designation;
[0037] T1-3 iterations, for the original discrete space coordinate data set vector given in the existing power operation video image data system, (if it has been merged into a matrix, take its row / column vector), the application of the single component gradient mapping algorithm makes the original data expand in r dimensions, r=2 / 3 / k or r=n, that is, the single scalar data is decomposed into r components carrying specific local detail information according to its selected dimensional basis and the specified data operation mode (usually number division), and these local component information has local correlation compared to the mapped dimensional basis row or column; then the application of the s-component gradient mapping algorithm corresponds to the decomposition of the single scalar data into r components according to its selected dimensional basis and the specified data operation mode (usually number division). s components that carry specific local detail information, and each local component information also has local correlation; finally, when s traverses all given original discrete space coordinate data group vectors in the existing power operation behavior video image data system (specific invalid data with the main diagonal / sub-diagonal being 1 is eliminated in advance as needed), the final high-order open data configuration carries global correlation information, and its data configuration is efficiently adapted to the data feeding training and intelligent iteration of existing artificial intelligence platforms such as TensorFlow. Example 4
[0038] As mentioned above, in a single component gradient mapping, for the 2 / 3 / k or n data of a row or column selected as the multilinear dimensional basis and thus acting as the independent variable, it will also be expanded into a second-order array according to itself in the component gradient mapping data process. This second-order data corresponds to x / x=1 in the scalar operation, and it actually does not have any data connotation; by analogy with x / x=1 in the scalar operation, its data configuration feature is that the values on the main diagonal / sub-diagonal are all 1 (the main diagonal / sub-diagonal is based on whether the selection of rows or columns in the component gradient mapping data process is consistent with the row and column layout adopted when the two-dimensional array is originally merged. If they are consistent, the values of the main diagonal are all 1, otherwise the values of the sub-diagonal are all 1); according to this data configuration feature, a data elimination process can be set.
[0039] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0040] It can be seen that compared with the prior art, the present invention solves the problem of single data source and data form of traditional video image monitoring algorithms, and expands the data configuration through sequential and interactive (in formulating operation modes such as division) data component allocation between related data groups in multiple linear dimensions. The resulting high-order data configuration contains detailed information on interactive correlation from local to global within the specified event of the power operation behavior. In particular, the resulting high-order data configuration is highly compatible with artificial intelligence platforms such as TensorFlow, and can collaborate in data feeding training and intelligent iteration (based on data accumulation).
[0041] In each embodiment, the hardware implementation of the technology can directly adopt existing intelligent devices, including but not limited to industrial computers, PCs, smart phones, handheld stand-alone machines, floor-standing stand-alone machines, etc. Its input device preferably adopts a screen keyboard, its data storage and calculation module adopts existing memory, calculator, controller, its internal communication module adopts existing communication ports and protocols, and its remote communication adopts existing GPRS networks, World Wide Internet, etc. It can be clearly understood by technicians in the relevant field that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of each functional unit and module are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the above method embodiments, which will not be repeated here.
[0042] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.
[0043] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.
Claims
1. An artificial intelligence-based multi-linear data intelligent control system for power operation behavior based on image recognition. Aiming at the current situation of single algorithm data source and single data form in conventional power operation video image monitoring scenarios, the system is based on existing power operation video image field recording data or the networked power operation video image platform that has been aggregated and uploaded, guided by the multi-linear parameterization of data configuration and data algorithm, oriented to and adapted to artificial intelligence data feeding training and its intelligent iterative evolution under a multi-parameter data model, and constructs a multi-linear data system in the power operation video image data scenario; the characteristics are: The core multi-linear data traversal distribution process is set to perform gradient iteration on the multi-linear dimension basis on the discrete spatial coordinate data group in the existing power operation video image data system, thereby obtaining a multi-linear vector data system for power operation behavior monitoring that is highly adaptable to artificial intelligence data training and data iteration; When the discrete space coordinate data group in the existing power operation video image data system is plane or space vector data, multiple discrete space coordinate data groups belonging to the same power operation event are combined to obtain a 2×n / 3×n configuration array; When the discrete spatial coordinate data group in the existing power operation video image data system is multidimensional vector data, multiple discrete spatial coordinate data groups belonging to the same power operation event are combined to obtain a k×n configuration; for the k×n configuration, k=2 / 3+k′, k′ is: At present, some power operation behavior video monitoring systems have added other power or electrical monitoring data in addition to the spatial video information uploaded by the on-site monitoring equipment. At this time, the k′ additionally written and recorded other power or electrical monitoring data can be included in this multilinear data system for integrated data processing as long as the data constraints are met; The gradient iteration utilizes component gradient mapping to construct a new data configuration that is compatible with the subject and object of the power operation in the power operation event and the components obtained after the subject and object are expanded in multiple linear dimensions. The new data configuration covers the process data of the power operation event and the detailed association data from local to global. The gradient iteration mines the data connotation of the original discrete space coordinate data set or its merged array based on the component gradient mapping algorithm; the obtained data has a configuration expanded in multiple dimensions; the component gradient mapping algorithm is specifically: for a 2×n / 3×n configuration or k×n configuration array, take the change value of 2 / 3 / k or n data in any row or column at a time interval selected on the time dimension, and at the same time, select a row or column in advance as a multilinear dimensional basis to serve as an independent variable data, take the 2 / 3 / k or n data of this independent variable row or column at a time interval selected on the time dimension The changing value under the interval, the former is used as the changing increment and the latter as the independent variable increment for gradient linear distribution under the selected data operation mode; wherein, since the independent variable increment and the changing increment are derived based on any row or column data, they each have a vector data configuration. The difference is that when performing component gradient mapping, any component of the changing increment vector is allocated to each independent variable increment under the selected data operation mode. In this way, a changing increment vector generates a two-dimensional, open multilinear data, i.e., a second-order tensor array, after undergoing a component gradient mapping.
2. According to claim 1, an artificial intelligence-based multi-linear data intelligent control system for electric power operation behavior based on image recognition, characterized in that: For the original discrete space coordinate data set vector given in the existing power operation video image data system, if it has been merged into a matrix, then take its row / column vector, and the application of the single component gradient mapping algorithm makes the original data unfold in r dimensions, r=2 / 3 / k or r=n, that is, decompose the single scalar data into r components carrying specific local detail information according to its selected dimensional basis and the specified data operation mode, and these local component information has local correlation compared to the mapped dimensional basis row or column; then the application of the s-component gradient mapping algorithm corresponds to decomposing the single scalar data into r components according to its selected dimensional basis and the specified data operation mode. s components that carry specific local detail information, and each local component information also has local correlation; finally, when s traverses all given original discrete space coordinate data group vectors in the existing power operation behavior video image data system, the final high-order opened data configuration carries global correlation information, and its data configuration is directly adapted to the data feeding training and intelligent iteration of the existing artificial intelligence platform.
3. According to claim 1, an artificial intelligence-based multi-linear data intelligent control system for electric power operation behavior based on image recognition, characterized in that: The selection of rows or columns is based on the original row and column layout used when constructing the matrix in claim 2. The row and column correspondence or the row and column transposition correspondence to the two component gradient mapping data processing processes, generating two multilinear data models with different data connotations.
4. According to claim 1, an artificial intelligence-based multi-linear data intelligent control system for electric power operation behavior based on image recognition, characterized in that: For two groups of change values, the time interval selected in the time dimension when the two groups of values are obtained should be completely uniform and consistent. Uniformity means that the time intervals in different time periods are the same, and consistency means that the time intervals used when the two groups of values are obtained are the same.
5. According to claim 1, an artificial intelligence-based multi-linear data intelligent control system for electric power operation behavior based on image recognition, characterized in that: in, If the original discrete space coordinate data set itself has vector properties, the gradient iteration of the above discrete space coordinate data set on the multi-linear dimensional basis is repeated once to obtain the second-order tensor data; Among them, if the original discrete space coordinate data group of the component gradient mapping has been merged into a two-dimensional array in advance, then in fact, a component gradient mapping will obtain a multilinear data array spread out in three linear dimensions, that is, a third-order tensor array; Among them, multiple iterations are performed to obtain high-order data and the order of high-order data set in a single data system for monitoring a single target power operation should be kept globally consistent, that is, all discrete space coordinate data groups monitored by the target power operation behavior should be subjected to gradient iterations of the same number of times and in the same order.
