Thermal power station real-time early warning system based on multi-dimensional data fusion

Abstract

The invention relates to the field of data processing, in particular to a thermal power station real-time early warning system based on multi-dimensional data fusion, and the system comprises a data collection device which is used for collecting image data and sensing data; the data processing module comprises a plurality of processing nodes for data processing; the analysis module is used for determining the type of the processing node based on the data acquisition change degree and the activity degree of the mobile equipment, and judging to execute consumption node selection or cloud processing unloading according to a neighborhood condition; the consumption control module is used for judging consumption nodes based on the evaluation balance value and the number of the key devices or the consumption evaluation value; the node regulation and control module is used for determining the consumption mode of the consumption node based on the distribution uniformity of the acquisition points corresponding to the acquisition range; the unloading module is used for executing cloud processing unloading so as to select the data acquisition equipment corresponding to the second-class node according to the priority selection value and distribute the data acquisition equipment to the cloud processing end; according to the invention, the processing efficiency of the early warning acquisition data is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to a real-time early warning system for thermal power plants based on multi-dimensional data fusion. Background Technology

[0002] In traditional thermal power plant intelligent operation and maintenance and real-time early warning systems, with the widespread application of sensor technology and image acquisition equipment, the data dimensions and scale on which the system relies are increasing day by day, forming a multi-source, heterogeneous, and high-concurrency data environment. To meet the real-time requirements of data processing, edge computing has been gradually introduced into industrial sites. By deploying edge nodes with certain computing capabilities near the data source, local processing of some data can be achieved, thereby reducing cloud load and improving response efficiency. However, in existing edge computing-based thermal power plant early warning systems, data processing strategies often adopt static or semi-static resource allocation mechanisms, which fail to fully consider the dynamic changes in node status, data quality, and equipment movement during actual operation. Especially in multi-dimensional data fusion scenarios, this leads to unreasonable allocation of data processing tasks, low resource utilization efficiency, and even affects the accuracy and timeliness of early warning.

[0003] Chinese Patent Publication No. CN120907594A discloses an intelligent early warning and fault diagnosis system for thermal power plants. The system includes a multi-source data acquisition module for real-time data acquisition; an edge computing node for noise filtering and outlier correction of the acquired data; a digital twin modeling unit for constructing a dynamic simulation model of the equipment based on a physical model and historical data; a hybrid analysis engine for locating early anomaly detection and fault root causes; and a visual early warning interface for dynamically displaying the health status and fault probability of the equipment and generating graded alarm signals. It is evident that the above technical solution suffers from the following problems: in multi-dimensional data fusion scenarios, it cannot adaptively adjust the node processing tasks according to the dynamic changes in data acquisition during actual operation, easily leading to poor early warning efficiency. Summary of the Invention

[0004] To address this issue, the present invention provides a real-time early warning system for thermal power plants based on multi-dimensional data fusion, which overcomes the problem in existing technologies that, when faced with multi-dimensional data fusion scenarios, cannot adaptively adjust node processing tasks according to the dynamic changes in data acquisition during actual operation, easily leading to poor early warning efficiency.

[0005] To achieve the above objectives, the present invention provides a real-time early warning system for thermal power plants based on multi-dimensional data fusion, comprising: Data acquisition equipment, including mobile devices for acquiring image data and sensing devices for acquiring sensor data; A data processing module, which is connected to the data acquisition device, includes several processing nodes for data processing. The analysis module, which is connected to the data processing module, is used to determine the category of processing nodes based on the degree of change in data collection and the activity of mobile devices, and to determine the neighborhood conditions based on the number of neighborhood first-class nodes corresponding to the second-class nodes in order to determine whether to select the execution of the absorption node or the cloud processing unloading. The absorption control module, which is connected to the analysis module and the data processing module, is used to perform absorption node selection based on the evaluation equilibrium value and to determine the absorption node based on the number of key equipment or the absorption evaluation value. The node control module, which is connected to the data processing module and the absorption control module, is used to determine the absorption method of the absorption node based on the comparison result of the uniformity of the distribution of the collection points corresponding to the collection range and the preset uniformity of the distribution of the collection points. The unloading module, which is connected to the analysis module and the data processing module, is used to perform cloud processing unloading to select the data acquisition devices corresponding to the second type of nodes and allocate them to the cloud processing terminal according to the priority selection value.

[0006] Furthermore, the data acquisition change rate corresponding to the first type of node is less than the data acquisition change rate threshold and the mobile device activity rate is less than the mobile device activity rate threshold. The data acquisition change rate corresponding to the two types of nodes is greater than the data acquisition change rate threshold or the mobile device activity rate is greater than the mobile device activity rate threshold.

[0007] Furthermore, the analysis module determines the node selection for elimination if the neighborhood condition is that the number of Class I nodes in the neighborhood is greater than the preset number of Class I nodes in the neighborhood.

[0008] Furthermore, when the absorption control module selects absorption nodes, it determines the evaluation equilibrium value based on the absorption evaluation value corresponding to each neighboring Class I node; If the evaluation equilibrium value is greater than the preset evaluation equilibrium value, the absorption control module determines to prioritize the neighborhood type I node with the smallest number of key equipment as the absorption node. If the evaluation equilibrium value is less than or equal to the preset evaluation equilibrium value, the absorption control module determines to prioritize selecting the neighborhood type I node with the largest absorption evaluation value as the absorption node.

[0009] Furthermore, the absorption evaluation value is determined by the absorption control module based on the mobile acquisition overlap rate and data anomaly reference value corresponding to the first type of neighboring nodes; The absorption evaluation value, the mobile acquisition overlap rate, and the data anomaly reference value are all negatively correlated.

[0010] Furthermore, under control conditions, the node control module determines the elimination method of the elimination node based on the comparison result between the uniformity of the distribution of the collection points corresponding to the collection range and the preset uniformity of the distribution of the collection points. If the uniformity of the collection point distribution is greater than the preset uniformity of the collection point distribution, the absorption adjustment method is range adjustment; If the uniformity of the collection point distribution is less than or equal to the preset uniformity of the collection point distribution, the absorption adjustment method is equipment absorption adjustment. The control condition is the completion of the selection of the absorption node corresponding to the second type of node.

[0011] Furthermore, the node control module adjusts the execution range by increasing the collection range corresponding to the absorption node based on the absorption evaluation value; The increase in the collection range is positively correlated with the absorption evaluation value.

[0012] Furthermore, the node control module performs equipment absorption adjustment, selecting data acquisition devices within the acquisition range of the second type of node and allocating them to the absorption node based on the priority selection value in descending order.

[0013] Furthermore, the analysis module determines to perform cloud processing unloading for Class II nodes whose neighborhood condition is that the number of Class I nodes in the neighborhood is less than or equal to the preset number of Class I nodes in the neighborhood.

[0014] Furthermore, the unloading module selects data acquisition devices within the acquisition range corresponding to the second type of node and assigns them to the cloud processing terminal according to the priority selection value from largest to smallest.

[0015] Compared with the prior art, the beneficial effect of the present invention is that the present invention reflects the dynamic changes of the data collected by the mobile device corresponding to the processing node through the data acquisition change degree, and reflects the dynamic movement of the current mobile device through the mobile device activity degree, and classifies the processing nodes accordingly, thereby obtaining two types of nodes with active data acquisition and one type of nodes with stable data acquisition, thus realizing the effective division of processing nodes.

[0016] In this invention, the analysis module determines the elimination node selection based on the result that the number of Class I nodes in the neighborhood is greater than the preset number of Class I nodes in the neighborhood. This avoids the chain reaction of collapse caused by the unstable node state due to too few Class I nodes when directly selecting elimination nodes.

[0017] In this invention, the absorption control module determines the evaluation equilibrium value based on the absorption evaluation value corresponding to each neighboring type of node. The evaluation equilibrium value characterizes the degree of balance between the absorption evaluation values ​​corresponding to each neighboring type of node, and selects different absorption node selection methods accordingly. This makes the selection method of absorption node more adaptable to specific application scenarios, thereby improving the task processing efficiency of this invention.

[0018] In this invention, the node control module determines the elimination method of the elimination node based on the comparison between the uniformity of the distribution of collection points corresponding to the collection range and the preset uniformity of the distribution of collection points under control conditions. It takes into account the distribution of actual collection devices within the collection range corresponding to the two types of nodes and selects the elimination method accordingly. This makes the selection of the elimination method fit the distribution of devices within the collection range, thereby improving the effectiveness of the elimination method and further improving the task allocation efficiency of the processing node in this invention. Attached Figure Description

[0019] Figure 1 This is a module connection diagram of the real-time early warning system for thermal power plants based on multi-dimensional data fusion according to the present invention; Figure 2 This is a flowchart illustrating how the present invention determines the category of processing nodes based on the degree of change in data acquisition and the activity level of mobile devices; Figure 3 This invention provides a flowchart for determining neighborhood conditions based on the number of neighboring nodes of type I corresponding to type II nodes, in order to decide whether to select a node for absorption or perform cloud processing unloading. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0022] Please see Figures 1 to 3 As shown, this invention provides a real-time early warning system for thermal power plants based on multi-dimensional data fusion, comprising: Data acquisition equipment, including mobile devices for acquiring image data and sensing devices for acquiring sensor data; A data processing module, which is connected to the data acquisition device, includes several processing nodes for data processing. The analysis module, which is connected to the data processing module, is used to determine the category of processing nodes based on the degree of change in data collection and the activity of mobile devices, and to determine the neighborhood conditions based on the number of neighborhood first-class nodes corresponding to the second-class nodes in order to determine whether to select the execution of the absorption node or the cloud processing unloading. The absorption control module, which is connected to the analysis module and the data processing module, is used to perform absorption node selection based on the evaluation equilibrium value and to determine the absorption node based on the number of key equipment or the absorption evaluation value. The node control module, which is connected to the data processing module and the absorption control module, is used to determine the absorption method of the absorption node based on the comparison result of the uniformity of the distribution of the collection points corresponding to the collection range and the preset uniformity of the distribution of the collection points. The unloading module, which is connected to the analysis module and the data processing module, is used to perform cloud processing unloading to select the data acquisition devices corresponding to the second type of nodes and allocate them to the cloud processing terminal according to the priority selection value.

[0023] This invention is applied to real-time early warning in thermal power plants. Specifically, the mobile device is an inspection device with at least mobility, image acquisition, and data transmission capabilities, such as an inspection robot. The inspection device moves along a path preset by the user and acquires images upon reaching a preset location. The acquired images are the image data. The sensing devices include, but are not limited to, temperature sensors for acquiring the temperature at the installation location, humidity sensors for acquiring the humidity at the installation location, vibration sensors for acquiring the vibration frequency at the installation location, gas composition analyzers for acquiring the concentrations of O2, CO, NOx, SO2, and CO2 at the installation location, and pressure sensors for acquiring the pressure at the installation location. Those skilled in the art already know how to set the sensor positions and select positions according to their own needs. This is common knowledge in the field and does not affect the content of the technical solution, so it will not be elaborated here.

[0024] In this invention, the processing node is a computing device with data processing capabilities. The processing node performs data processing, namely, monitoring and issuing early warnings for image data and sensor data through its built-in machine learning model. The location of the processing node is set by the user. It can be understood that the user can set the location of the processing node according to the actual distribution of equipment in the thermal power plant to ensure that each sensor device is within the acquisition range of at least one processing node. This is already known to those skilled in the art and will not be elaborated here.

[0025] The real-time early warning system for thermal power plants based on multi-dimensional data fusion described in this invention also includes a cloud processing terminal, which also has at least data processing capabilities and the ability to monitor and issue early warnings for image data and sensor data through machine learning models.

[0026] This invention also utilizes several historical records. Each historical record at least records the data acquisition variation rate, mobile device activity, overlap area, number of neighboring Class I nodes, evaluation equilibrium value, mobile acquisition overlap rate, data anomaly reference value, uniformity of acquisition point distribution, and the total number of selected data acquisition devices during the historical process of the thermal power plant. In addition, each historical record has a corresponding qualification mark, which records whether the historical record meets the user's needs. It is already known to those skilled in the art that the historical process of the thermal power plant corresponding to the historical record meets the user's needs based on self-defined indicator standards (e.g., the duration between the time of the fault occurrence and the time of the warning) and will not be elaborated further.

[0027] Specifically, the data collection change rate corresponding to the first type of node is less than the data collection change rate threshold and the mobile device activity rate is less than the mobile device activity rate threshold. The data acquisition change rate corresponding to the two types of nodes is greater than or equal to the data acquisition change rate threshold or the mobile device activity rate is greater than or equal to the mobile device activity rate threshold.

[0028] For a single edge node, its corresponding data acquisition change rate is the average of the image change rate of each mobile device within the acquisition range corresponding to the edge node in the most recent monitoring period. Its corresponding mobile device activity rate is the average of the displacement reference values ​​of each mobile device within the acquisition range corresponding to the edge node in the most recent monitoring period. For a single mobile device, the method for confirming the image change rate of the most recent monitoring period is to obtain each image acquired by the mobile device in the most recent monitoring period, detect the pixel average value of each image, and record the difference between the maximum and minimum values ​​of the pixel average value as the image change rate of the mobile device. For a single mobile device, the method for confirming the displacement reference value of the most recent monitoring period is to obtain the movement path of the mobile device in the most recent monitoring period and record the area of ​​the smallest circle that can completely include the movement path as the displacement reference value of the mobile device. For a single edge node, its corresponding acquisition range is a circular area centered on the location of the edge node and with a radius of 60% of the maximum data transmission distance of the edge node.

[0029] The values ​​for the data collection variability threshold and the mobile device activity threshold are determined by the understanding that higher values ​​indicate more unstable data collection. Therefore, the greater the user's demand for stable data collection, the lower the values ​​of these thresholds should be. One approach is to extract the data collection variability and mobile device activity values ​​corresponding to historical records that meet user needs, remove outliers from both values, and then record the average values ​​of these values ​​as the data collection variability threshold and the mobile device activity threshold, respectively. Outlier removal methods include, but are not limited to, the 3σ criterion or the IQR method.

[0030] This invention applies a continuous cyclical monitoring cycle. The duration of a single monitoring cycle can be set by the user according to actual needs. It can be understood that the greater the user's need for monitoring accuracy of data collection changes and mobile device activity, the shorter the duration of a single monitoring cycle will be. One possible value is a duration of 6 minutes for a single monitoring cycle.

[0031] Specifically, the analysis module determines and executes the selection of elimination nodes for Class II nodes whose neighborhood condition is that the number of Class I nodes in the neighborhood is greater than the preset number of Class I nodes in the neighborhood.

[0032] For a single Class II node, the method for identifying its corresponding Class I neighboring nodes is as follows: detect Class I nodes whose maximum acquisition range overlaps with the acquisition range of the Class II node by a preset area, and record them as Class I neighboring nodes. The maximum acquisition range is a circular area constructed with the position of the Class I node as the center and the maximum data transmission distance of the Class I node as the radius. The number of Class I neighboring nodes is the total number of Class I neighboring nodes corresponding to the Class II node. The maximum data transmission distance is the farthest distance that the processing node itself can allow for data transmission. This is common knowledge and will not be elaborated here.

[0033] The values ​​for the preset area and the preset number of Class I neighboring nodes are understood to be such that a larger overlapping area indicates a stronger receiving capability of Class I neighboring nodes for data acquisition devices within Class II nodes. Therefore, the greater the user's demand for the data receiving capability of Class I neighboring nodes, the larger the preset area should be. One possible value is an overlapping area of ​​30% of the area of ​​the acquisition range corresponding to the Class II nodes, rounded up to the nearest integer. Furthermore, a larger number of Class I neighboring nodes indicates a higher probability of the existence of Class I nodes capable of receiving data from data acquisition devices within Class II nodes. Therefore, the greater the user's demand for the effectiveness of node selection, the larger the preset number of Class I neighboring nodes should be. One possible value is a preset number of Class I neighboring nodes of 3. One method for setting these values ​​involves extracting the overlapping area and the number of Class I neighboring nodes corresponding to historical records that meet the user's needs, removing outliers from both the overlapping area and the number of Class I neighboring nodes, and then recording the average of the overlapping area and the number of Class I neighboring nodes after removing outliers as the preset area and the preset number of Class I neighboring nodes, respectively.

[0034] Specifically, when the absorption control module selects absorption nodes, it determines the evaluation equilibrium value for a single Class II node based on the absorption evaluation values ​​of each neighboring Class I node. If the evaluation equilibrium value is greater than the preset evaluation equilibrium value, the absorption control module determines to prioritize the neighboring Class I node with the smallest number of critical equipment as the absorption node corresponding to the Class II node. If the evaluation equilibrium value is less than or equal to the preset evaluation equilibrium value, the absorption control module determines to prioritize selecting the neighborhood type I node with the largest absorption evaluation value as the absorption node corresponding to the type II node.

[0035] For a single Class II node, the evaluation equilibrium value of its corresponding neighboring Class I nodes is calculated as follows: in, To evaluate the equilibrium value, For the first The absorption evaluation value corresponding to each type of node in the neighborhood. This represents the average absorption evaluation value corresponding to each type of node in the neighborhood. This represents the total number of Class I nodes in the neighborhood corresponding to the Class II node.

[0036] The preset evaluation equilibrium value can be understood as follows: the larger the evaluation equilibrium value, the greater the balance between the absorption evaluation values ​​corresponding to each neighboring type of node. Therefore, when the evaluation equilibrium value is greater than the preset evaluation equilibrium value, it reflects a poor effectiveness in selecting absorption nodes based on absorption evaluation values. Thus, the selection of absorption nodes is based on the number of key equipment. Therefore, the greater the user's acceptance of the balance between absorption evaluation values, the larger the preset evaluation equilibrium value. One possible value is provided: the preset evaluation equilibrium value is... 80% of the value is used to provide a value selection method, which extracts the evaluation equilibrium value corresponding to the historical records that meet the user's needs, and records the average value of the evaluation equilibrium value after removing outliers as the preset evaluation equilibrium value.

[0037] The key equipment refers to sensing devices installed on boilers, steam turbines, generators, or water pumps. For a single neighborhood type 1 node, the number of key equipment is the number of key equipment within the acquisition range corresponding to that neighborhood type 1 node.

[0038] Specifically, the absorption evaluation value is determined by the absorption control module based on the mobile acquisition overlap rate and data anomaly reference value corresponding to the first type of neighboring nodes; The absorption evaluation value, the mobile acquisition overlap rate, and the data anomaly reference value are all negatively correlated.

[0039] When selecting a corresponding absorption node for a single Class II node, the absorption evaluation value of a single neighboring Class I node = (preset mobile acquisition overlap rate / mobile acquisition overlap rate) + (preset data anomaly reference value / data anomaly reference value). For a single neighborhood Class I node, the corresponding mobile acquisition overlap rate is calculated as the number of mobile devices in the overlapping area between the maximum acquisition range of the Class I node and the acquisition range of the Class II node, divided by the overlapping area, with the unit being devices / m². The data anomaly reference value is the total number of faults of the target equipment within the last three monitoring cycles recorded by the user within the neighborhood Class I node. The target equipment includes boilers, steam turbines, and generators. The existence of boilers, steam turbines, and generators in thermal power plants and the recording of fault counts are all knowledgeable to those skilled in the art and will not be elaborated upon here.

[0040] The preset mobile acquisition overlap rate and preset data anomaly reference value are set with the understanding that the larger the mobile acquisition overlap rate and the data anomaly reference value, the lower the stability of data acquisition for a certain type of node in the neighborhood. Therefore, the greater the user's demand for data acquisition stability, the smaller the preset mobile acquisition overlap rate and preset data anomaly reference value should be. A value setting method is provided to extract the mobile acquisition overlap rate and data anomaly reference value corresponding to the historical records that meet the user's needs, remove the outliers from the mobile acquisition overlap rate and data anomaly reference value respectively, and record the average values ​​of the mobile acquisition overlap rate and data anomaly reference value after removing the outliers as the preset mobile acquisition overlap rate and preset data anomaly reference value respectively.

[0041] Specifically, under control conditions, the node control module determines the elimination method of the elimination node based on the comparison result between the uniformity of the distribution of the collection points corresponding to the collection range and the preset uniformity of the distribution of the collection points. If the uniformity of the collection point distribution is greater than the preset uniformity of the collection point distribution, the absorption adjustment method is range adjustment; If the uniformity of the collection point distribution is less than or equal to the preset uniformity of the collection point distribution, the absorption adjustment method is equipment absorption adjustment.

[0042] The control condition is the completion of the selection of the absorption node corresponding to the second type of node.

[0043] For a single Class II node, the method for confirming the uniformity of the distribution of collection points within its corresponding collection range is as follows: the collection range is evenly divided into several sector areas of equal area, and the total number of mobile devices and sensing devices in each sector area is recorded as the number of devices in the corresponding sector area. The uniformity of the distribution of collection points corresponding to this Class II node = 1 / the absolute value of the difference between the maximum and minimum values ​​of the number of devices in the sector area. It is worth noting that, as a special case, if the absolute value of the difference between the maximum and minimum values ​​of the number of devices in the sector area = 0, then the uniformity of the distribution of collection points is directly recorded as the preset uniformity of the distribution of collection points + 1.

[0044] The preset value for the uniformity of data collection point distribution is understood to mean that the greater the uniformity of the data collection point distribution, the more evenly the data collection devices are distributed. Therefore, when the uniformity of the data collection point distribution is greater than the preset uniformity, the range adjustment method is selected to avoid the problem of poor adjustment effect caused by the uneven distribution of data collection devices. So the higher the user's requirements for the uniformity of the distribution of data collection devices, the larger the value of the preset uniformity of data collection point distribution. A value selection method is provided to extract the uniformity of data collection point distribution corresponding to the historical records that meet the user's needs, and the average value of the uniformity of data collection point distribution after removing outliers is recorded as the preset uniformity of data collection point distribution.

[0045] Specifically, the node control module adjusts the execution range by increasing the collection range corresponding to the absorption node based on the absorption evaluation value. The increase in the collection range is positively correlated with the absorption evaluation value.

[0046] For a single Class II node, the expanded collection range of its corresponding absorption node remains a circular area with the center still at the location corresponding to the absorption node. The area of ​​the expanded collection range = the area of ​​the unexpanded collection range × (absorption evaluation value / baseline absorption evaluation value). The baseline absorption evaluation value is the average of the baseline absorption evaluation values ​​of each neighboring Class I node corresponding to the Class II node. It is worth noting that the area of ​​the expanded collection range is an integer rounded up, with the unit fixed at m², and the area of ​​the expanded collection range must not exceed the maximum collection range.

[0047] Specifically, the node control module performs equipment absorption adjustment, selecting data acquisition devices within the acquisition range of the second type of node and allocating them to the absorption node based on the priority selection value in descending order.

[0048] For a single data acquisition device, the priority value is the amount of data transmitted to the edge node in the previous monitoring cycle of that data acquisition device, in MB.

[0049] In this process, when selecting data acquisition devices within the collection range corresponding to the two types of nodes based on the priority selection value from largest to smallest, the selection stops when the total number of selected data acquisition devices reaches the preset total number. The value of the preset total number can be understood as follows: the larger the total number of selected data acquisition devices, the greater the impact on the stability of the absorption node. Once a load fluctuation occurs, it may affect the operation of the absorption node itself. Therefore, the greater the user's demand for the stability of the data acquisition device control, the smaller the preset total number. One method is to extract the total number of selected data acquisition devices corresponding to the historical records that meet the user's needs, and record the average of the total number after removing outliers as the preset total number.

[0050] Specifically, the analysis module determines to perform cloud processing unloading for Class II nodes whose neighborhood condition is that the number of Class I nodes in the neighborhood is less than or equal to the preset number of Class I nodes in the neighborhood.

[0051] Specifically, the unloading module selects data acquisition devices within the acquisition range corresponding to the second type of node and assigns them to the cloud processing terminal according to the priority selection value from largest to smallest.

[0052] When the total number of data acquisition devices selected to be allocated to the cloud processing terminal is the preset total number, the selection process stops.

[0053] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments, and the above solutions can be implemented in whole or in part by software, hardware, firmware, or other arbitrary combinations. When implemented in software, it can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will recognize that the modules and algorithm steps disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0054] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A real-time early warning system for thermal power plants based on multi-dimensional data fusion, characterized in that, The application relates to a data processing method, which comprises the following steps: collecting image data by using a mobile device and collecting sensing data by using a sensing device; connecting a data processing module with the data collecting device, wherein the data processing module comprises a plurality of processing nodes for data processing; connecting an analysis module with the data processing module, wherein the analysis module is used for determining the category of the processing nodes based on the data collection variation degree and the mobile device activity degree, and determining the neighborhood condition of the two categories of nodes to determine whether to execute the accommodation node selection or cloud processing unloading; connecting an accommodation control module with the analysis module and the data processing module, wherein the accommodation control module is used for executing the accommodation node selection to determine the accommodation node selection based on the evaluation balance value and the key device quantity or the accommodation evaluation value; connecting a node regulation module with the data processing module and the accommodation control module, wherein the node regulation module is used for determining the accommodation mode of the accommodation node based on the comparison result of the collection point distribution uniformity corresponding to the collection range and the preset collection point distribution uniformity; connecting an unloading module with the analysis module and the data processing module, wherein the unloading module is used for executing the cloud processing unloading to select the data collecting device corresponding to the two categories of nodes to be distributed to the cloud processing end according to the priority selection value. The data collection variation degree of the first category of nodes is less than a data collection variation degree threshold value, and the mobile device activity degree is less than a mobile device activity degree threshold value. The data collection variation degree of the second category of nodes is greater than or equal to the data collection variation degree threshold value or the mobile device activity degree is greater than or equal to the mobile device activity degree threshold value. The analysis module determines to execute the accommodation node selection for the second category of nodes with the neighborhood condition of the neighborhood first category of nodes being greater than a preset neighborhood first category of nodes. When the accommodation control module executes the accommodation node selection, the evaluation balance value is determined based on the accommodation evaluation value corresponding to each neighborhood first category of node. If the evaluation balance value is greater than a preset evaluation balance value, the accommodation control module determines to preferentially select the neighborhood first category of node with the minimum key device quantity as the accommodation node. If the evaluation balance value is less than or equal to the preset evaluation balance value, the accommodation control module determines to preferentially select the neighborhood first category of node with the maximum accommodation evaluation value as the accommodation node.

2. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 1, characterized in that, The accommodation evaluation value is determined by the accommodation control module based on the mobile collection overlap rate and the data anomaly reference value corresponding to the neighborhood first category of node. The accommodation evaluation value, the mobile collection overlap rate and the data anomaly reference value are in negative correlation.

3. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 2, characterized in that, The node regulation module determines the accommodation mode of the accommodation node based on the comparison result of the collection point distribution uniformity corresponding to the collection range and the preset collection point distribution uniformity under the regulation condition.

4. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 3, characterized in that, If the collection point distribution uniformity is greater than the preset collection point distribution uniformity, the accommodation regulation mode is range regulation. If the collection point distribution uniformity is less than or equal to the preset collection point distribution uniformity, the accommodation regulation mode is device accommodation regulation. The regulation condition is that the selection of the accommodation node corresponding to the second category of nodes is completed.

5. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 4, characterized in that, The node regulation module executes the range regulation, and the accommodation evaluation value is used for increasing regulation of the collection range corresponding to the accommodation node. The increasing value of the collection range and the accommodation evaluation value are in positive correlation.

6. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 5, characterized in that, ​ ​ ​ ​ 7. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 6, characterized in that, ​ ​ 8. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 6, characterized in that, The node regulation module performs equipment accommodation adjustment, and selects data collection equipment in the collection range corresponding to the second-type node in descending order of the priority selection value and allocates the data collection equipment to the accommodation node.

9. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 2, characterized in that, The analysis module determines to execute cloud processing offloading for the second-type node whose neighborhood condition is that the number of neighborhood first-type nodes is less than or equal to the preset number of neighborhood first-type nodes.

10. The real-time warning system for thermal power plants based on multi-dimensional data fusion according to claim 9, characterized in that, The offloading module selects data collection equipment in the collection range corresponding to the second-type node in descending order of the priority selection value and allocates the data collection equipment to the cloud processing end.

Images