Multi-source sensor fused wind field monitoring data error compensation method and system

Abstract

The invention discloses a wind field monitoring data error compensation method and system based on multi-source sensor fusion, and belongs to the technical field of meteorological monitoring, and the method comprises the following steps: putting sensor nodes to each point location of a detection area, and collecting wind field monitoring data and communication signal intensity data; determining spatial position information of each sensor node according to the communication signal intensity data, and extracting a terrain shielding coefficient of each sensor node; calculating the wind field monitoring data of each sensor node according to the error correction coefficient, and outputting compensated wind field monitoring data; and in the process of outputting the compensated wind field monitoring data, continuously performing drift detection on the wind field monitoring data of each sensor node, and when drift abnormality is detected, performing replacement processing on the error correction coefficient of the abnormal node. According to the method, the three-dimensional coordinate system is established, and the terrain shielding coefficient is introduced on the basis of the three-dimensional coordinate system, so that the defect that a traditional compensation method only looks at data and does not look at the environment is overcome.

Description

Technical Field

[0001] This invention relates to the field of meteorological monitoring technology, specifically to a method and system for error compensation of wind field monitoring data based on multi-source sensor fusion. Background Technology

[0002] In the field of meteorological monitoring, the accuracy of wind field data directly impacts subsequent decision analysis. Existing wind field monitoring methods typically rely on sensor networks distributed across the target area. However, in complex geographical environments, the data collected by sensors often contains errors.

[0003] Traditional sensor compensation methods mostly rely on the correction of a single physical quantity and lack deep integration of the spatial characteristics of the monitored environment.

[0004] For example, in mountainous or densely built-up areas, subtle differences in micro-topography (such as one node being located behind a retaining wall while the other is in open ground) can cause significant nonlinear deviations in the wind speed data collected by two adjacent sensor nodes. Traditional linear compensation algorithms cannot identify this local wind field distortion caused by "terrain shading," and when sensors experience electrical characteristic drift over time, the system often has to crudely remove outliers, resulting in data vacuums in the monitoring network and hindering high-precision real-time supplementation and self-repair. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention provides a method and system for error compensation of wind field monitoring data by multi-source sensor fusion.

[0006] Therefore, the technical problem solved by this invention is that most traditional sensor compensation methods rely only on the correction of a single physical quantity and lack deep integration of the spatial characteristics of the monitoring environment.

[0007] To address the aforementioned technical problems, this invention provides the following technical solution: a method for error compensation of wind field monitoring data using multi-source sensor fusion, comprising: deploying sensor nodes to various points in the detection area to collect wind field monitoring data and communication signal strength data; determining the spatial location information of each sensor node based on the communication signal strength data, and extracting the terrain occlusion coefficient of each sensor node; calculating the error correction coefficient of each sensor node based on the spatial location information and the terrain occlusion coefficient; calculating the wind field monitoring data of each sensor node based on the error correction coefficient, and outputting the compensated wind field monitoring data; during the output of the compensated wind field monitoring data, continuously performing drift detection on the wind field monitoring data of each sensor node, and when an abnormal drift is detected, replacing the error correction coefficient of the abnormal node.

[0008] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the step of determining the spatial location information of each sensor node includes: constructing a multi-source sensor data fusion system, wherein the multi-source sensor data fusion system is wirelessly connected to each sensor node, and receives wind field monitoring data and communication signal strength data uploaded by each sensor node; the multi-source sensor data fusion system calculates the signal transmission attenuation coefficient from each sensor node to the core central monitoring node according to the logarithmic path loss model based on the communication signal strength data of each sensor node; selecting sensor nodes whose communication signal strength data meets a first preset threshold as core central monitoring nodes; taking the core central monitoring node as the coordinate origin, recursively calculating the three-dimensional coordinate distance of each sensor node relative to the core central monitoring node based on the difference between the signal transmission attenuation coefficient and the reference path loss corresponding to the reference distance; and determining the three-dimensional relative coordinates of each sensor node by combining the azimuth information of the communication signal arrival.

[0009] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the calculation formula for the signal transmission attenuation coefficient is expressed as follows: ; in, The signal transmission attenuation coefficient, The reference signal strength at the reference distance. Here, represents the measured communication signal strength data of the i-th sensor node, and n is the path loss exponent, determined by the terrain and environmental calibration of the detection area. Let be the three-dimensional coordinate distance from the i-th sensor node to the core central monitoring node. For reference distance.

[0010] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the steps of calculating the three-dimensional coordinate distance and the three-dimensional relative coordinate include: The logarithmic path loss model is modified to inversely solve the three-dimensional coordinate distance using the difference between the signal transmission attenuation coefficient and the reference path loss. The modified formula is as follows: ; The three-dimensional coordinate distance of each sensor node relative to the core monitoring node is calculated recursively based on the deformation formula; the azimuth and elevation angles of the communication signals from each sensor node to the core monitoring node are obtained, and the three-dimensional relative coordinates of each sensor node are calculated by combining the three-dimensional coordinate distances. Specifically, it is expressed as: ; ; ; In the formula, It is the azimuth angle. The pitch angle.

[0011] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the step of extracting the terrain occlusion coefficient of each sensor node includes: acquiring terrain images through the visual monitoring cameras mounted on each sensor node; extracting the terrain structure contours of the terrain images, classifying the terrain structure contours according to preset analysis conditions, and filtering out potential occlusion areas; performing connected component analysis on the potential occlusion areas to extract continuous occlusion areas, and calculating the pixel area of ​​each occlusion area; summing the pixel areas of each occlusion area to obtain the total number of occlusion pixels, and calculating the terrain occlusion coefficient in combination with the total number of pixels in the image.

[0012] As a preferred embodiment of the multi-source sensor fusion wind field monitoring data error compensation method described in this invention, the step of calculating the error correction coefficient of each sensor node includes: calculating the height influence factor and distance attenuation factor based on the three-dimensional relative coordinates of each sensor node; and constructing an error correction model to calculate the error correction coefficient in conjunction with the terrain occlusion coefficient. The error correction model is expressed as follows: ; In the formula, , The preset weighting coefficients, This is the terrain occlusion coefficient. As a high-impact factor, For distance attenuation factor, This is the error correction factor; The original wind field monitoring data is compensated using the error correction coefficient to obtain compensated wind field monitoring data.

[0013] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the step of continuously detecting drift in the wind field monitoring data of each sensor node includes: acquiring the historical monitoring sequence of the target sensor node within a set sliding window period; calculating the mean of the historical monitoring sequence and introducing the data mean of neighboring sensor nodes as a reference benchmark; determining whether the data change rate of the target sensor node exceeds a preset abnormal threshold, and if it exceeds the threshold and the duration reaches the alarm duration, then determining that the sensor node has experienced a drift abnormality.

[0014] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the step of replacing the error correction coefficient of abnormal nodes includes: taking the sensor node experiencing drift anomaly as the center, selecting multiple neighboring sensor nodes with normal operating status within a preset communication radius; obtaining the three-dimensional relative coordinates, terrain occlusion coefficient, and currently effective error correction coefficient of each neighboring sensor node; calculating the straight-line distance between each neighboring sensor node and the abnormal node, and allocating the contribution weight of each neighboring node according to the reciprocal of the distance; comparing the terrain occlusion coefficients between each neighboring sensor node, eliminating neighboring nodes whose terrain occlusion coefficients differ from those of the abnormal node by more than a preset proportion, and readjusting the contribution weights; multiplying and summing the error correction coefficients of the remaining neighboring sensor nodes with their corresponding contribution weights to generate a replacement correction coefficient for the abnormal node; overwriting the original coefficient of the abnormal node with the replacement correction coefficient, and re-collecting the wind field monitoring data of the abnormal node for secondary verification.

[0015] As a preferred embodiment of the wind field monitoring data error compensation method based on multi-source sensor fusion described in this invention, the secondary verification step includes: extracting the real-time monitoring value of the abnormal node after applying the substitution correction coefficient, and calculating the residual between the value and the average monitoring value of the neighboring sensor node group; determining whether the residual converges to a preset error allowable range; if the residual does not converge, retrieving the historical health data features of the corresponding point of the abnormal node and adjusting the gain of the substitution correction coefficient; if the residual still does not converge after a preset number of fine adjustments, marking the abnormal node as a hardware fault state.

[0016] This invention provides a wind field monitoring data error compensation system based on multi-source sensor fusion.

[0017] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a wind field monitoring data error compensation system based on multi-source sensor fusion, comprising: an acquisition module for acquiring wind field monitoring data and communication signal strength data; The terrain occlusion coefficient extraction module is used to determine the spatial location information of each sensor node based on the communication signal strength data, and to extract the terrain occlusion coefficient of each sensor node. The calculation module is used to calculate the error correction coefficient of each sensor node based on the spatial location information and the terrain occlusion coefficient. The compensation module is used to calculate the wind field monitoring data of each sensor node based on the error correction coefficient and output the compensated wind field monitoring data.

[0018] The beneficial effects of this invention are as follows: By establishing a three-dimensional coordinate system and introducing a terrain occlusion coefficient extracted through visual monitoring, this invention overcomes the shortcomings of traditional compensation methods that only consider data and not the environment. By calculating the height influence factor and distance attenuation factor, the micro-topographic environment of each node can be identified, thereby generating targeted error correction coefficients and improving the accuracy of wind field monitoring under complex terrain. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 The above is a flowchart of an error compensation method for wind field monitoring data based on multi-source sensor fusion, provided as an embodiment of the present invention. Detailed Implementation

[0021] To make the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a method for error compensation of wind field monitoring data fusion from multiple sources, including the following steps: S1. Deploy sensor nodes to various locations in the detection area to collect wind field monitoring data and communication signal strength data.

[0023] S2. Determine the spatial location information of each sensor node based on the communication signal strength data, and extract the terrain occlusion coefficient of each sensor node.

[0024] S3. Based on the spatial location information and terrain occlusion coefficient, calculate the error correction coefficient for each sensor node.

[0025] S4. Calculate the wind field monitoring data of each sensor node based on the error correction coefficient, and output the compensated wind field monitoring data.

[0026] S5. During the output of compensated wind field monitoring data, drift detection is continuously performed on the wind field monitoring data of each sensor node. When an abnormal drift is detected, the error correction coefficient of the abnormal node is replaced.

[0027] It should be noted that in wind field monitoring in complex mountainous areas or densely built-up urban areas, wind distribution exhibits strong local nonlinear characteristics due to topographical undulations, vegetation obstruction, or building obstruction. Traditional sensors, after deployment, often fail to reflect the true micro-meteorological environment simply by relying on single electrical calibration. Furthermore, long-term exposure to harsh outdoor environments, coupled with temperature and humidity changes and hardware aging, makes sensors highly susceptible to zero-point drift or gain failure, causing the collected raw wind speed data to deviate from the true value. In addition, due to the dispersed nature of monitoring points, traditional manual inspections and offline calibrations are not only costly but also unable to capture the impact of instantaneous environmental changes on monitoring accuracy.

[0028] Therefore, to address the aforementioned issues of environmental interference coupling and long-term sensor operational stability, multi-source data acquisition in step S1 provides a foundation for subsequent positioning and compensation. Steps S2 and S3 utilize wireless signal strength to inversely calculate the three-dimensional relative coordinates of nodes and combine this with visual image recognition technology to quantify the degree of terrain occlusion, constructing an error correction model deeply coupled with the spatial environment to achieve personalized "correction" of wind field data for each monitoring point. Step S4 outputs high-precision compensated wind field data. Simultaneously, based on the sliding window drift detection and neighborhood consistency replacement mechanism in step S5, when a node exhibits anomalies, logical fitting and coefficient coverage are performed using neighboring node data with similar terrain features, ensuring the monitoring system's real-time self-healing capability and long-term monitoring accuracy in complex and variable environments.

[0029] Example 2, refer to Figure 1 As an embodiment of the present invention, based on the previous embodiment, a method for compensating for errors in wind field monitoring data through multi-source sensor fusion is provided, comprising: S1. Deploy sensor nodes to various locations in the detection area to collect wind field monitoring data and communication signal strength data.

[0030] In this embodiment, for a given wind field area to be measured, it is assumed to contain 20 sensor nodes, which are randomly deployed at various locations on the mountain slope, ridge, and valley floor. Each sensor node synchronously starts its data acquisition task. The wind field monitoring data, including real-time wind speed and direction, is acquired by the ultrasonic probes mounted on the nodes; the communication signal strength data is extracted in real time by the node's wireless communication module during data interaction.

[0031] S2. Determine the spatial location information of each sensor node based on the communication signal strength data, and extract the terrain occlusion coefficient of each sensor node.

[0032] The steps for determining the spatial location information of each sensor node include S2.1 to S2.4: S2.1 Build a multi-source sensor data fusion system. The multi-source sensor data fusion system is wirelessly connected to each sensor node and receives wind field monitoring data and communication signal strength data uploaded by each sensor node.

[0033] The multi-source sensor data fusion system establishes connections with all nodes in the area via LoRa or 5G wireless base stations, and stores the communication signal strength data and wind speed values ​​uploaded by each node in the background database according to timestamp alignment.

[0034] S2.2 The multi-source sensor data fusion system calculates the signal transmission attenuation coefficient from each sensor node to the core monitoring node based on the communication signal strength data of each sensor node and in accordance with the logarithmic path loss model. For a node group with a certain tiered distribution, obtain the reference signal strength at a reference distance, and then use the formula for calculating the signal transmission attenuation coefficient: ; in, The signal transmission attenuation coefficient, The reference signal strength at the reference distance. Here, represents the measured communication signal strength data of the i-th sensor node, and n is the path loss exponent, determined by the terrain and environmental calibration of the detection area. Let be the three-dimensional coordinate distance from the i-th sensor node to the core central monitoring node. For reference distance.

[0035] The signal transmission attenuation coefficient of the i-th node is calculated. It should be noted that the path loss index is affected by environmental factors such as vegetation density and air humidity in the detection area, and is obtained through calibration using previously measured data in actual engineering.

[0036] The steps for calculating the three-dimensional coordinate distance include: transforming the logarithmic path loss model, and using the difference between the signal transmission attenuation coefficient and the reference path loss to solve for the three-dimensional coordinate distance. The transformed formula is as follows: ; The three-dimensional coordinate distance of each sensor node relative to the core central monitoring node is calculated recursively based on the deformation formula.

[0037] S2.3. Select sensor nodes whose communication signal strength data meets the first preset threshold as core center monitoring nodes. Using the core center monitoring node as the coordinate origin, calculate the three-dimensional coordinate distance of each sensor node relative to the core center monitoring node recursively based on the difference between the signal transmission attenuation coefficient and the reference path loss corresponding to the reference distance.

[0038] Specifically, nodes with consistently stable communication signal strength data greater than -50dBm (such as node 1) are selected as core monitoring nodes, and their three-dimensional coordinates are set to (0,0,0). Using the modified formula described above, the distance sequence of the remaining 19 nodes relative to node 1 is recursively calculated.

[0039] S2.4 Determine the three-dimensional relative coordinates of each sensor node by combining the azimuth information of the communication signal arrival.

[0040] The steps for calculating three-dimensional relative coordinates include: The azimuth and elevation angles of the communication signals from each sensor node reaching the core monitoring node are obtained. Combined with the three-dimensional coordinate distance, the three-dimensional relative coordinates of each sensor node are calculated. Specifically, it is expressed as: ; ; ; In the formula, It is the azimuth angle. The pitch angle.

[0041] Taking node 5 in this embodiment as an example, if the calculated straight-line distance azimuth Pitch angle Then its relative coordinates are calculated as follows: ; ; .

[0042] The steps for extracting the terrain occlusion coefficient of each sensor node include A1~A4: A1. Terrain images are collected through visual monitoring cameras mounted on each sensor node.

[0043] A2. Extract the terrain structure contours from the terrain image, classify the terrain structure contours according to preset analysis conditions, and filter out potential occlusion areas.

[0044] Specifically, the preset analysis conditions include edge gradient threshold conditions and horizontal baseline height conditions; during the extraction process, operators (such as the Canny operator) are first used to identify edges in the image where pixel brightness changes abruptly, and to construct the terrain structure outline; Subsequently, contours with solid features are filtered out based on the edge gradient threshold, and combined with the horizontal baseline determined by the sensor installation height, closed or semi-closed contour areas located above the horizontal baseline and with a slope change rate exceeding a preset threshold are classified and determined as potential occlusion areas.

[0045] A3. Perform connected component analysis on the potential occlusion regions, extract continuous occlusion regions, and calculate the pixel area of ​​each occlusion region.

[0046] Specifically, connected component analysis refers to using a pixel neighborhood check algorithm to traverse and label pixels within a potential occlusion area; clustering spatially adjacent pixels with the same masking properties into an independent topological set, thereby transforming scattered edges into continuous occlusion areas representing mountains or buildings. It is also important to know that the pixel area of ​​each occluded region is calculated as follows: the total number of pixels contained in each continuous occluded region is used as the apparent area of ​​the occluded entity.

[0047] A4. Sum the pixel areas of each occluded region to obtain the total number of occluded pixels, and calculate the terrain occlusion coefficient by combining it with the total number of pixels in the image.

[0048] Sum the pixel areas of all the continuously occluded regions identified in step A3 to obtain the total number of occluded pixels. Subsequently, the total number of pixels in the image is obtained by multiplying the resolutions of the entire terrain image. ; Using the ratio formula The terrain shading coefficient of the node was calculated. .

[0049] S3. Based on the spatial location information and terrain occlusion coefficient, calculate the error correction coefficient for each sensor node.

[0050] The steps for calculating the error correction coefficients for each sensor node include S3.1 to S3.3: S3.1 Calculate the height influence factor and distance attenuation factor based on the three-dimensional relative coordinates of each sensor node.

[0051] Specifically, high impact factor The contribution of vertical height to wind pressure distribution is characterized by obtaining the height component in the three-dimensional relative coordinates of the i-th sensor node. Through the exponential distribution model It is confirmed that, among them, This is a preset reference altitude (e.g., the local average altitude is 500m).

[0052] Distance decay factor The method used to characterize the energy loss of signal and wind field propagation due to horizontal spatial distance is as follows: the horizontal projected distance between the node and the core node is calculated using the horizontal coordinate components. Then the distance decay factor ,in, The unit reference distance is 100m. By introducing a logarithmic function, the correction coefficient exhibits a non-linear, smooth decay characteristic at long distances.

[0053] S3.2. Based on the terrain occlusion coefficient, construct an error correction model and calculate the error correction coefficient. The error correction model is expressed as follows: ; In the formula, , The preset weighting coefficients, This is the terrain occlusion coefficient. As a high-impact factor, For distance attenuation factor, This is the error correction factor.

[0054] It should be noted that when the node is at a high altitude (i.e., the altitude influence factor increases) or is severely blocked by the surrounding area (i.e., the terrain blocking coefficient increases), the wind force is significantly reduced due to the obstruction of the terrain, and the calculated error correction coefficient increases. However, when the node is far from the core point, which increases environmental uncertainty (i.e., the distance attenuation factor increases), the denominator is adjusted to prevent overcorrection.

[0055] S3.3. The original wind field monitoring data is compensated using the error correction coefficient to obtain the compensated wind field monitoring data.

[0056] Specifically, the raw wind speed data collected synchronously from each node is retrieved and multiplied by the corresponding error correction coefficient. For example, if the raw wind speed collected by a node is 10.5 m / s and the calculated error correction coefficient is 1.08, then the wind field monitoring data at that point is 11.34 m / s.

[0057] S4. Calculate the wind field monitoring data of each sensor node based on the error correction coefficient, and output the compensated wind field monitoring data.

[0058] S5. During the output of compensated wind field monitoring data, drift detection is continuously performed on the wind field monitoring data of each sensor node. When an abnormal drift is detected, the error correction coefficient of the abnormal node is replaced.

[0059] The steps for continuously detecting drift in the wind field monitoring data of each sensor node include S5.1~S5.3: S5.1 Obtain the historical monitoring sequence of the target sensor node within the set sliding window period.

[0060] Specifically, a 10-minute sliding window is assigned to each node, and wind speed monitoring values ​​for that node within that time period are extracted in real time with a step size of 1 second, forming a time series dataset. .

[0061] S5.2 Calculate the mean of the historical monitoring sequence and introduce the data mean of the neighboring sensor nodes as a reference benchmark.

[0062] Calculate the mean value of the target node within the current window. Simultaneously, extract the real-time monitoring mean values ​​of all neighboring nodes within a 300-meter physical distance of the target node, and calculate their arithmetic mean as a reference benchmark.

[0063] S5.3 Determine whether the data change rate of the target sensor node exceeds the preset abnormal threshold. If it exceeds the threshold and the duration reaches the alarm duration, then the sensor node is determined to have drifted abnormally.

[0064] Specifically, by calculating the rate of change of data . For reference, The average value is the data. When the rate of change of the data exceeds the preset abnormal threshold (set to 25% in this embodiment), and the duration of the deviation reaches the alarm duration (set to 300 seconds in this embodiment), after excluding environmental fluctuation factors such as sudden gusts of wind, it is determined that the node has experienced an abnormality caused by sensor zero drift or hardware aging.

[0065] The steps for replacing the error correction coefficients of abnormal nodes include B1 to B6: B1. Using the sensor node that has experienced drift anomalies as the center, select multiple neighboring sensor nodes that are operating normally within a preset communication radius.

[0066] For example, using the abnormal node 8 as the center, nodes 7, 9, 12, and 15, which are currently marked as healthy, are selected within a 500-meter communication radius.

[0067] B2. Obtain the three-dimensional relative coordinates, terrain occlusion coefficient, and currently effective error correction coefficient of each of the neighboring sensor nodes.

[0068] Retrieve the coordinates, occlusion coefficients, and correction coefficients of the aforementioned neighboring nodes from the database. Assume node 8 has coordinates (100, 100, 15), and its terrain occlusion coefficient... .

[0069] B3. Calculate the straight-line distance between each of the neighboring sensor nodes and the abnormal nodes, and assign the contribution weight of each neighboring node according to the reciprocal of the distance.

[0070] Specifically, calculate the straight-line distance between each neighboring node and node 8. .

[0071] Assuming the calculation yields the following distance: Node 7 ; Distance of node 9 ; Distance of node 12 ; Distance of node 15 .

[0072] The initial contribution weight is calculated based on the reciprocal of the distance (1 / d), with higher weights for closer nodes. The initial weight ratio for each node is calculated to be 4:2:1:0.8.

[0073] B4. Compare the terrain occlusion coefficients among the neighboring sensor nodes, remove neighboring nodes whose terrain occlusion coefficients differ from those of abnormal nodes by more than a preset ratio, and readjust the contribution weights.

[0074] Set the preset ratio threshold to 25%.

[0075] Given that the occlusion coefficient of node 8 is 0.20, the following comparisons are made with neighboring nodes: node 7 has a coefficient of 0.21, which is 5% different from node 8; node 9 has a coefficient of 0.19, which is 5% different from node 8; node 12 has a coefficient of 0.23, which is 15% different from node 8; and node 15 has a coefficient of 0.40, which is 100% different from node 8.

[0076] Because the difference ratio of node 15 exceeds 25%, its micro-topography is deemed inconsistent with that of node 8, and therefore it is removed. The contribution weights of the remaining nodes 7, 9, and 12 are then redistributed. After normalization, the final contribution weights are output again. , , .

[0077] B5. Multiply the error correction coefficients of the remaining neighboring sensor nodes by their corresponding contribution weights and sum them to generate alternative correction coefficients for the abnormal node.

[0078] Specifically, obtain the currently effective correction coefficients for the remaining nodes, assuming... , , .

[0079] Calculate the substitution correction factor: The replacement correction factor for node 8 is 1.10.

[0080] B6. Overwrite the original coefficient of the abnormal node with the replacement correction coefficient, and re-collect the wind field monitoring data of the abnormal node for secondary verification.

[0081] Specifically, 1.10 is written into the calculation configuration item of node 8, overwriting its original anomaly coefficient. Then, the wind speed value output from node 8 after applying the new coefficient is collected. .

[0082] The specific verification process is as follows: Calculation ,in The average real-time monitoring values ​​for nodes 7, 9, and 12.

[0083] The secondary verification steps include C1~C4: C1. Extract the real-time monitoring value of the abnormal node after applying the substitution correction coefficient, and calculate the residual between it and the monitoring mean of the neighboring sensor node group.

[0084] Specifically, the new wind speed monitoring value output by node 8 after applying a replacement correction factor of 1.10 is obtained. Simultaneously, the real-time monitoring values ​​of neighboring nodes 7, 9, and 12 at the same sampling time are extracted, and their average value is calculated, assuming the average value is 10.2 m / s.

[0085] Then calculate the residual between the two. .

[0086] C2. Determine whether the residual has converged to the preset error allowable range.

[0087] The preset error tolerance range is ±5% of the neighborhood mean, which means the maximum allowable residual threshold is... .

[0088] In this embodiment, since the calculated residual of 0.6m / s > 0.51m / s, it is determined that the residual has not converged to the allowable error range, and the system triggers further gain adjustment logic.

[0089] C3. If the residuals do not converge, retrieve the historical health data features of the points corresponding to the abnormal nodes and adjust the gain of the substitution correction coefficient.

[0090] By retrieving the health operation data characteristics of node 8 in the 24 hours before the drift occurred from the database (such as the historical average correction factor of 1.05 for this point under a specific wind direction).

[0091] Introducing feedback gain factor (In this embodiment, the value is 0.95), for the current substitution correction coefficient. Perform fine-tuning calculations: The fine-tuned gain coefficient The data was sent back to node 8, and the changes in its output wind speed were observed again, attempting to bring the residual closer to within 0.51 m / s.

[0092] C4. If the residual still fails to converge after a preset number of fine-tuning operations, the abnormal node will be marked as a hardware fault.

[0093] In this embodiment, the maximum number of fine-tuning attempts is set to 3.

[0094] If, after three iterations with different gain factors, the output residual of node 8 remains above 0.51 m / s, or if the data does not change with the coefficient, the system determines that the deviation cannot be resolved by environmental correction and officially marks node 8 as a hardware failure in the topology map.

[0095] Meanwhile, the system sets node 8 as an invalid source in the multi-source sensor data fusion system, so it no longer participates in global wind field modeling.

[0096] Example 3 is an embodiment of the present invention. This embodiment provides a wind field monitoring data error compensation system based on multi-source sensor fusion, including a data acquisition module for acquiring wind field monitoring data and communication signal strength data. The terrain occlusion coefficient extraction module is used to determine the spatial location information of each sensor node based on the communication signal strength data, and to extract the terrain occlusion coefficient of each sensor node. The calculation module is used to calculate the error correction coefficient of each sensor node based on the spatial location information and the terrain occlusion coefficient. The compensation module is used to calculate the wind field monitoring data of each sensor node based on the error correction coefficient and output the compensated wind field monitoring data.

[0097] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for compensating for errors in wind field monitoring data fusion from multiple sensor sources, characterized in that: Includes the following steps: Sensor nodes are deployed to various locations within the detection area to collect wind field monitoring data and communication signal strength data. Based on the communication signal strength data, the spatial location information of each sensor node is determined, and the terrain occlusion coefficient of each sensor node is extracted. Based on the spatial location information and terrain occlusion coefficient, the error correction coefficient of each sensor node is calculated; The wind field monitoring data of each sensor node are calculated based on the error correction coefficient, and the compensated wind field monitoring data is output. During the output of compensated wind field monitoring data, drift detection is continuously performed on the wind field monitoring data of each sensor node. When an abnormal drift is detected, the error correction coefficient of the abnormal node is replaced.

2. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 1, characterized in that: The steps to determine the spatial location information of each sensor node include: A multi-source sensor data fusion system is established, which is wirelessly connected to each sensor node and receives wind field monitoring data and communication signal strength data uploaded by each sensor node. The multi-source sensor data fusion system calculates the signal transmission attenuation coefficient from each sensor node to the core monitoring node based on the communication signal strength data of each sensor node and according to the logarithmic path loss model. Sensor nodes whose communication signal strength data meets the first preset threshold are selected as core central monitoring nodes. Taking the core central monitoring node as the origin of the coordinate system, the three-dimensional coordinate distance of each sensor node relative to the core central monitoring node is recursively calculated based on the difference between the signal transmission attenuation coefficient and the reference path loss corresponding to the reference distance. The three-dimensional relative coordinates of each sensor node are determined by combining the azimuth information of the communication signal arrival.

3. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 2, characterized in that: The formula for calculating the signal transmission attenuation coefficient is as follows: ； in, The signal transmission attenuation coefficient, The reference signal strength at the reference distance. Here, represents the measured communication signal strength data of the i-th sensor node, and n is the path loss exponent, determined by the terrain and environmental calibration of the detection area. Let be the three-dimensional coordinate distance from the i-th sensor node to the core central monitoring node. For reference distance.

4. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 3, characterized in that: The steps for calculating 3D coordinate distance and 3D relative coordinates include: The logarithmic path loss model is modified to inversely solve the three-dimensional coordinate distance using the difference between the signal transmission attenuation coefficient and the reference path loss. The modified formula is as follows: ； The three-dimensional coordinate distance of each sensor node relative to the core central monitoring node is calculated recursively based on the deformation formula. The azimuth and elevation angles of the communication signals from each sensor node reaching the core monitoring node are obtained. Combined with the three-dimensional coordinate distance, the three-dimensional relative coordinates of each sensor node are calculated. Specifically, it is expressed as: ； ； ； In the formula, It is the azimuth angle. It is the pitch angle.

5. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 4, characterized in that: The steps for extracting the terrain occlusion coefficient of each sensor node include: Terrain images are acquired through visual monitoring cameras mounted on each sensor node; Extract the terrain structure contours from the terrain image, classify the terrain structure contours according to preset analysis conditions, and filter out potential occlusion areas. Connectivity analysis is performed on the potential occlusion regions to extract continuous occlusion regions, and the pixel area of ​​each occlusion region is calculated. The pixel areas of each occluded region are summed to obtain the total number of occluded pixels, and the terrain occlusion coefficient is calculated by combining it with the total number of pixels in the image.

6. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 5, characterized in that: The steps for calculating the error correction coefficients for each sensor node include: The height influence factor and distance attenuation factor are calculated based on the three-dimensional relative coordinates of each sensor node. Based on the terrain occlusion coefficient, an error correction model is constructed to calculate the error correction coefficient. The error correction model is expressed as follows: ； In the formula, , The preset weighting coefficients, This is the terrain occlusion coefficient. As a high-impact factor, For distance attenuation factor, This is the error correction factor; The original wind field monitoring data is compensated using the error correction coefficient to obtain compensated wind field monitoring data.

7. The method for compensating for errors in wind field monitoring data by multi-source sensor fusion as described in claim 6, characterized in that: The steps for continuously detecting drift in the wind field monitoring data of each sensor node include: Acquire the historical monitoring sequence of the target sensor node within a set sliding window period; The mean of the historical monitoring sequence is calculated, and the mean data of the neighboring sensor nodes is introduced as a reference benchmark. Determine whether the data change rate of the target sensor node exceeds the preset abnormal threshold. If it exceeds the threshold and the duration reaches the alarm duration, then the sensor node is determined to have drifted abnormally.

8. The method for compensating for errors in wind field monitoring data based on multi-source sensor fusion as described in claim 7, characterized in that: The steps for replacing the error correction coefficients of abnormal nodes include: Centered on the sensor node that has experienced drift anomalies, multiple neighboring sensor nodes with normal operating status are selected within a preset communication radius. Obtain the three-dimensional relative coordinates, terrain occlusion coefficient, and currently effective error correction coefficient of each of the neighboring sensor nodes; Calculate the straight-line distance between each of the neighboring sensor nodes and the abnormal node, and assign the contribution weight of each neighboring node according to the reciprocal of the distance; Compare the terrain occlusion coefficients among the neighboring sensor nodes, remove neighboring nodes whose terrain occlusion coefficients differ from those of abnormal nodes by more than a preset proportion, and then readjust the contribution weights. The error correction coefficients of the remaining neighboring sensor nodes are multiplied by their corresponding contribution weights and summed to generate alternative correction coefficients for the abnormal node. The replacement correction coefficient is used to overwrite the original coefficient of the abnormal node, and the wind field monitoring data of the abnormal node is re-acquired for secondary verification.

9. The method for compensating for errors in wind field monitoring data by multi-source sensor fusion as described in claim 8, characterized in that: The steps of the secondary verification include: Extract the real-time monitoring value of the abnormal node after applying the substitution correction coefficient, and calculate the residual between it and the monitoring mean of the neighboring sensor node group; Determine whether the residual has converged to a preset error allowable range; If the residuals do not converge, retrieve the historical health data features of the points corresponding to the abnormal nodes and adjust the gain of the substitution correction coefficients. If the residual still fails to converge after a preset number of fine-tuning operations, the abnormal node will be marked as a hardware fault.

10. A wind field monitoring data error compensation system based on multi-source sensor fusion, employing the wind field monitoring data error compensation method based on multi-source sensor fusion as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to collect wind field monitoring data and communication signal strength data; The terrain occlusion coefficient extraction module is used to determine the spatial location information of each sensor node based on the communication signal strength data, and to extract the terrain occlusion coefficient of each sensor node. The calculation module is used to calculate the error correction coefficient of each sensor node based on the spatial location information and the terrain occlusion coefficient. The compensation module is used to calculate the wind field monitoring data of each sensor node based on the error correction coefficient and output the compensated wind field monitoring data.

Images