Distributed collaborative measurement method and system for load sensor array

Abstract

The application discloses a distributed cooperative measurement method and system of a load sensor array, relates to the fields of sensor networks and distributed measurement technology, and comprises the following steps: performing finite element simulation of a target structure, identifying key stress areas, boundary areas and load types, and establishing a structure and demand feature library; deploying a three-layer nested sensor array according to the feature library, configuring a communication protocol and a global clock, synchronizing and collecting data to establish a time series data set; and finally outputting a load measurement result according to a cooperative authentication result after cooperative processing. The application solves the technical problem that the prior art lacks a multivariable distributed measurement mechanism and intelligent sensors are insufficient in cooperation, and it is difficult to comprehensively and accurately obtain structural load information, and cannot meet the demand for accurate measurement and effective control of structural load, achieves the technical effects of comprehensive and accurate acquisition of structural load information, and meets the demand for accurate measurement and effective control of structural load.

Description

Technical Field

[0001] This invention relates to the field of sensor networks and distributed measurement technology, and in particular to a distributed collaborative measurement method and system for load sensor arrays. Background Technology

[0002] Structural load measurement is crucial for the safe operation of equipment and the stability assessment of large structures. Current technologies primarily employ traditional centralized sensors or single-variable measurement devices for load detection, which have played a role in monitoring stable conditions or simple structures. However, with increasing demands for monitoring complex structures and dynamic loads, these methods have revealed limitations: a lack of multi-variable distributed measurement capabilities and insufficient intelligent sensor collaboration mechanisms prevent the comprehensive capture of multi-dimensional load information from different regions, resulting in fragmented and inaccurate data that fails to meet the needs of precise assessment and effective control. Summary of the Invention

[0003] This application provides a distributed collaborative measurement method and system for load sensor arrays, which addresses the technical problem that existing technologies lack multivariate distributed measurement mechanisms and have insufficient intelligent sensor collaboration, making it difficult to comprehensively and accurately obtain structural load information and meet the requirements for precise measurement and effective control of structural loads.

[0004] The first aspect of this application provides a distributed collaborative measurement method for a load sensor array. The method includes: performing finite element simulation of the target structure; identifying key stress regions, boundary regions, and load types based on the finite element simulation results; establishing a structural feature library and a demand feature library; deploying a sensor array based on the structural feature library and the demand feature library, wherein the sensor array is a three-layer nested array, including a base layer node, a relay layer node, and a global layer node; configuring a communication protocol for the sensor array and completing network topology initialization; activating the sensor array to collect node data after configuring global clock synchronization, and establishing a time-series dataset; performing collaborative processing of the time-series dataset under the three-layer nested array based on the network topology; and outputting load measurement results based on the collaborative authentication results.

[0005] A second aspect of this application provides a distributed collaborative measurement system for a load sensor array. The system includes: a feature library construction module, used to perform finite element simulation of the target structure, identify key stress areas, boundary areas, and load types based on the finite element simulation results, and establish a structural feature library and a demand feature library; a sensor array deployment module, used to deploy a sensor array according to the structural feature library and the demand feature library, wherein the sensor array is a three-layer nested array, including a base layer node, a relay layer node, and a global layer node; a network topology initialization module, used to configure the communication protocol for the sensor array and complete network topology initialization; a time-series dataset construction module, used to activate the sensor array to collect node data and establish a time-series dataset after configuring global clock synchronization; and a load measurement result acquisition module, used to perform collaborative processing of the time-series dataset under the three-layer nested array according to the network topology, and output load measurement results based on the collaborative authentication results.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] This application establishes a structural feature library and a demand feature library by performing finite element simulation of the target structure. Based on this, a three-layer nested intelligent sensor array containing basic, intermediate, and global layer nodes is deployed. After network topology initialization and global clock synchronization, the array is activated to collect and establish a time-series dataset. Then, through the collaborative processing and data authentication of the three-layer nodes, multivariate distributed measurement is achieved, thereby accurately obtaining the load information of the structure. This makes the load measurement results more accurate and reliable, achieving comprehensive and accurate acquisition of structural load information and meeting the technical requirements for accurate measurement and effective control of structural loads. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in 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.

[0009] Figure 1 This is a schematic flowchart of the distributed collaborative measurement method for a load sensor array provided in the embodiments of this application.

[0010] Figure 2 This is a schematic diagram of the structure of the distributed collaborative measurement system of the load sensor array provided in the embodiments of this application.

[0011] Figure labeling: Feature library construction module 1, sensor array deployment module 2, network topology initialization module 3, time series dataset construction module 4, load measurement result acquisition module 5. Detailed Implementation

[0012] This application provides a distributed collaborative measurement method and system for load sensor arrays, which addresses the technical problem that existing technologies lack multivariate distributed measurement mechanisms and have insufficient intelligent sensor collaboration, making it difficult to comprehensively and accurately obtain structural load information and meet the requirements for precise measurement and effective control of structural loads.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0015] Example 1, as Figure 1 As shown, a distributed collaborative measurement method for a load sensor array is provided, wherein the method includes:

[0016] Step A100: Perform finite element simulation of the target structure, identify the key stress areas, boundary areas, and load types based on the finite element simulation results, and establish a structural feature library and a demand feature library.

[0017] In this embodiment, finite element simulation refers to simulation analysis of the target structure, through which results such as the key stress regions, boundary regions, and load types of the structure can be obtained. The structural feature library is a library established based on structural information such as key stress regions and boundary regions identified in the finite element simulation results. The requirement feature library is a library established based on information such as load types identified in the finite element simulation results.

[0018] Specifically, when performing finite element simulation of a target structure, it is first necessary to collect comprehensive parameters of the target structure, such as the main beam of a large bridge or the load-bearing frame of heavy machinery. This includes the material properties, geometric dimensions, and connection methods used in actual operation, such as bolted connections and welded fixation. Based on these parameters, a three-dimensional geometric model of the target structure is first constructed, and then it is discretized into several finite element elements using mesh generation technology. For critical parts with complex stresses, such as near the supports of a bridge or at the corners of machinery, a fine mesh with an element size of 5cm × 5cm can be used, while a coarser mesh with an element size of 20cm × 20cm is used in non-critical areas to balance computational accuracy and efficiency.

[0019] The next step involves constructing the mechanical calculation model: first, constraints are defined, such as the fixed hinge supports at both ends of the bridge's main beam, limiting horizontal and vertical displacement; and the fixed constraints connecting the mechanical frame to the ground, limiting three-dimensional displacement. Then, load conditions are set according to the actual working scenario of the structure, including static loads, such as the bridge's self-weight and the constant load of the machinery; and dynamic loads, such as the impact load of vehicles crossing the bridge and the periodic load of the machinery during operation, clearly defining the magnitude, direction, and location of each load. Simultaneously, material constitutive relations (such as linear elastic models and elastoplastic models) are introduced to describe the deformation of the material under stress. The iterative accuracy of the solver (e.g., convergence error 1e-6) and the time step (set to 0.01s for dynamic simulation) are set to ensure that the mechanical calculation model accurately reflects the mechanical response of the structure.

[0020] After constructing the mechanical calculation model, simulations are executed to solve the equilibrium equations of each finite element unit through numerical iteration, outputting data such as stress distribution, strain values, and displacements of the structure under different working conditions. This data will be directly used to identify critical stress regions, i.e., areas where stress is concentrated and exceeds 80% of the material's allowable stress; boundary regions, i.e., transitional regions where constraints are located and stress gradients change abruptly; and load types, i.e., stress is stable under static loads and stress fluctuates periodically over time under dynamic loads. This provides a quantitative basis for subsequently establishing structural feature libraries and demand feature libraries.

[0021] For example, when simulating the boom of a certain type of crane, the stress value in the middle section of the boom under rated load reaches 200 MPa, far higher than the 50 MPa at both ends, thus identifying the middle section as the critical stress area. Simultaneously, the connection between the boom and the base, due to stress concentration and a displacement change rate 40% higher than other areas, is marked as a boundary region. Based on the simulation output data, the load type is further analyzed. If the load on the boom during lifting suddenly increases from 0 to 1.2 times the rated value within 0.3 seconds, accompanied by high-frequency fluctuations, then an impact load is identified; while during the uniform lifting phase, the load fluctuation amplitude remains stable within ±5%, indicating a static load. Through this type of analysis, the type and characteristic parameters of the loads acting on the target structure are clarified.

[0022] Subsequently, the above identification results are integrated to establish a structural feature library and a demand feature library. The structural feature library includes the three-dimensional coordinates of key stress areas, stress thresholds, connection forms of boundary areas, and critical strain values; the demand feature library determines measurement requirements based on load type, such as a 1kHz sampling frequency for impact loads to capture transient changes, and a 0.1kHz sampling frequency for static loads to ensure data stability. It also includes threshold values ​​for monitoring anomalies such as overload and rapid change, for example, the overload threshold for impact loads is set to 200kN, and the rapid change threshold is set to 50kN / ms.

[0023] By performing finite element simulation of the target structure, key stress areas, boundary areas, and load types are identified, and corresponding feature libraries are established, laying the foundation for the scientific deployment of sensor arrays and achieving the effect of improving the relevance of load measurement and the effectiveness of data.

[0024] Step A200: Deploy a sensor array according to the structural feature library and the demand feature library. The sensor array is a three-layer nested array, including a base layer node, a relay layer node, and a global layer node.

[0025] Optionally, before deploying the sensor array, the core information of the structural feature library and the requirement feature library must be clearly defined. The structural feature library records the three-dimensional coordinates and mechanical characteristics of the target structure's key stress areas and boundary areas; the requirement feature library clarifies the measurement requirements based on the load type, such as static loads and impact loads. For example, impact loads require a sampling frequency of 1 kHz to capture transient changes, while static loads require a sampling frequency of 0.1 kHz to balance accuracy and energy consumption. It also includes thresholds for monitoring anomalies such as overload and rapid changes.

[0026] Based on the structural feature library, basic layer nodes are preferentially deployed in critical stress areas and boundary regions. Taking a robotic arm as an example, its joints are considered critical areas, with nodes deployed every 10cm. 2 Deploy one base layer node to ensure it can capture subtle stress changes; the deployment density in the boundary area is one node per 20cm. 2 One node covers the stress gradient transition range, enabling the collected data to directly reflect the load state of the core structural components. These nodes integrate lightweight algorithms to meet the real-time preprocessing requirements of the feature library.

[0027] The deployment of relay layer nodes is based on the regional division and basic layer node distribution in the structural feature library. If the structural feature library divides the robotic arm into 3 independent force-bearing regions, with 40-60 basic layer nodes in each region, the relay layer nodes are deployed at the geometric center of each region, ensuring that the communication distance with all basic layer nodes in that region does not exceed 8 meters. Each relay layer node is responsible for receiving and initially integrating data from 20-30 basic layer nodes, which avoids data delays caused by excessive communication distances and efficiently processes load information within the region.

[0028] The deployment of global layer nodes needs to cover the entire target structure, and the number is determined based on the distribution of the relay layers. For example, if each of the three areas of the robotic arm has one relay layer node, the global layer nodes can be deployed near the control system of the robotic arm to ensure that the communication latency with each relay layer node does not exceed 30ms, so as to quickly aggregate data from each area, meet the real-time requirements of global collaborative analysis, and adapt to the accuracy requirement of <2% for global data integration in the requirement feature library.

[0029] If sensor array deployment relies on manual experience, it can easily lead to insufficient nodes in critical areas and redundant nodes in non-critical areas, resulting in low data acquisition efficiency and missing core information. However, by deploying according to the aforementioned structural feature library and demand feature library, the basic layer, relay layer, and global layer nodes can accurately cover the required areas, and their configuration can match the load measurement requirements, enabling efficient collaboration among nodes at each layer in data acquisition, transmission, and analysis.

[0030] By deploying a three-layer nested sensor array consisting of basic, intermediate, and global layer nodes according to the structural feature library and the demand feature library, reasonable node coverage and configuration are ensured, thereby improving the targeting of load measurement and the efficiency of system coordination.

[0031] Step A300: After configuring the communication protocol for the sensor array, complete the network topology initialization.

[0032] In one embodiment of this application, when configuring the communication protocol for the sensor array, firstly, an adaptation protocol is selected based on the functional requirements of the three-layer nodes. Between the base layer nodes and the relay layer nodes, pre-processed time-series data needs to be transmitted with high real-time requirements and a latency of <20ms. A low-power wireless protocol based on IEEE 802.15.4 is adopted, with a communication rate set to 250kbps. A collision avoidance mechanism is also configured to ensure that data uploads from 30 base layer nodes can be completed within 100ms. Between the relay layer nodes and the global layer nodes, data from multiple nodes within the area needs to be aggregated. The LoRaWAN protocol is used to balance transmission distance and bandwidth, with a spreading factor of 10 and a transmission rate of 50kbps, ensuring a packet loss rate of <1% between the relay layer and global layer nodes within a 100-meter range.

[0033] After the communication protocol configuration is complete, network topology initialization begins with the base layer nodes. Base layer nodes actively send broadcast signals to detect other nearby base layer nodes and connectable relay layer nodes, filtering and establishing a primary connection table based on signal strength and communication stability. The primary connection table contains core communication connection information recorded by each base layer node, specifically including the communication addresses of 3-5 adjacent base layer nodes for local information sharing between nodes, and the communication address of one primary relay layer node, serving as the main data upload channel. To ensure transmission reliability, when the signal strength of the primary relay layer node drops below -85dBm, the primary connection table automatically switches to a pre-stored backup relay layer node address. This redundancy design ensures stable data upload from the base layer to the relay layer.

[0034] After the base layer completes the establishment of the first-level connection table, the relay layer nodes initiate the next initialization step. The relay layer nodes collect the first-level connection table information of all base layer nodes within their jurisdiction and integrate it to generate a second-level connection table. The second-level connection table not only contains the communication relationships between the relay layer and the base layer nodes it governs, but also records the addresses of the two highest-priority adjacent relay layer nodes through interactions with other relay layer nodes. These addresses are used for cross-regional data collaboration. When data anomalies occur within a region requiring correlation analysis, the relay layer can quickly forward information through the adjacent addresses in the second-level connection table, enabling time-series consistency analysis and data linkage between regions.

[0035] The secondary connection table of the relay layer is ultimately uploaded to the global layer nodes, which then use it to construct a global topology graph. The global topology graph provides a panoramic view of the entire sensor network's connectivity, including the communication addresses of all base and relay layer nodes, as well as marking the real-time status of each node, such as online duration and signal stability. To ensure topology accuracy, the global topology graph is updated every 5 seconds. If a node fails to communicate three times consecutively, it is temporarily removed from the graph to prevent invalid data transmission from consuming resources. Through the global topology graph, the global layer nodes can clearly understand the connection logic of the entire network, providing a complete network structure basis for the collaborative processing of time-series datasets by the subsequent three layers of nodes.

[0036] By configuring an appropriate communication protocol for the three-layer nested sensor array and dynamically initializing the network topology, efficient connection and data transmission between nodes are ensured, thereby improving the system's communication reliability and data transmission real-time performance.

[0037] Step A400: After configuring global clock synchronization, activate the sensor array to collect node data and establish a time-series dataset.

[0038] Specifically, when configuring global clock synchronization, the global layer node is used as the time reference source. Time synchronization signals are sent to each relay layer node through a preset high-precision synchronization protocol, namely the precise time protocol based on IEEE 1588. After receiving the signal, the relay layer node forwards it to the base layer node under its jurisdiction, forming a hierarchical synchronization link of global layer-relay layer-base layer.

[0039] During the synchronization process, the global layer node sends a timestamp every 10 milliseconds. After receiving the timestamp, the relay layer node calibrates its local clock to ensure that the time deviation from the global reference does not exceed 5 microseconds. Then, it forwards the synchronization signal to the base layer node at the same frequency. The base layer node adjusts the crystal oscillator frequency by comparing the received timestamp with the local clock to control the time deviation within ±10 microseconds.

[0040] Meanwhile, the system will monitor the synchronization status of each node in real time. If a node has a synchronization deviation of more than 20 microseconds for three consecutive times, the global layer node will trigger a forced synchronization mechanism, resend the high-precision time base and record the synchronization anomaly log to ensure that the acquisition time of all nodes in the basic layer, relay layer and global layer is consistent, providing a reliable time base for the construction of subsequent time series datasets and the time series consistency analysis of the relay layer.

[0041] Next, the sensor array is activated to collect node data to establish a time-series dataset. A front-end event sensing unit is configured for each sensor node. During data collection, event sensing is performed based on this unit to establish node sensing results. The sensing results are shared through a first-level neighborhood set to update the adaptive acquisition window. Then, the acquisition is continued using this window. The specific steps are explained in detail in A410-A440.

[0042] By configuring a global clock synchronization to ensure time consistency among nodes, and combining the node data acquisition after activating the sensor array, a time-series dataset is established to provide basic data for time alignment in subsequent three-layer collaborative processing.

[0043] Step A500: Perform time-series dataset collaborative processing under the three-layer nested array according to the network topology, and output the load measurement results based on the collaborative authentication results.

[0044] Specifically, the collaborative processing of time-series datasets under a three-layer nested array includes: basic layer preprocessing to establish the first abnormal load identifier; relay layer analysis to establish the second abnormal load identifier; mapping structure based on network topology and structural feature library; global layer collaborative analysis; and data collaborative authentication by combining multiple information and outputting load measurement results. The specific steps are explained in detail in A510-A550.

[0045] Furthermore, step A500 in the method provided in this application embodiment includes:

[0046] A510: Perform local node preprocessing of the time series dataset at the base layer node to establish the first abnormal load identifier.

[0047] A520: Upload the time-series dataset and the first abnormal load identifier to the relay layer node, perform regional load analysis, and establish a second abnormal load identifier.

[0048] A530: A graph structure is established based on a network topology and structural feature library. The nodes of the graph structure are sensor nodes, and the edges of the graph structure represent the information channels or structural coupling between any two nodes.

[0049] A540: Couple the graph structure to the global layer node, perform global collaborative analysis, and establish global collaborative analysis results.

[0050] A550: Performs data collaborative authentication based on global collaborative analysis results, first abnormal load identifier, second abnormal load identifier, and time series dataset, and outputs load measurement results.

[0051] In this embodiment, the graph structure is established based on the network topology and structural feature library, where the nodes are sensor nodes and the edges represent the information channels or structural coupling between any two nodes.

[0052] Specifically, firstly, the base layer nodes preprocess the time series dataset using a lightweight algorithm, configure abnormal indicators such as overload, rapid change, and non-stationary fluctuations, and identify the abnormalities in the preprocessing results based on the indicators to establish the first abnormal load identifier. The specific steps are explained in detail in A511-A513.

[0053] Next, the relay layer node obtains the communication relationship with the base layer node and calls the time series dataset to establish a local perception subgraph. Based on this graph and the first abnormal load identifier, the timing consistency of nodes in the region is analyzed to identify synchronization, propagation, and isolation anomalies, and then a second abnormal load identifier is established. The specific steps are explained in detail in A521-A523.

[0054] Then, when building the graph structure based on the network topology and structural feature library, all nodes in the sensor array, including the base layer, relay layer, and global layer nodes, are first used as vertices of the graph structure. Each vertex is associated with attributes such as node number, layer level, and monitoring area coordinates. Next, the information channels between nodes are determined by combining the network topology: the communication relationships between each node are extracted from the network topology, such as the communication link between the base layer node and the relay layer node that governs it, and the adjacent communication paths between relay layer nodes. If two nodes have stable data transmission, i.e., communication success rate ≥90% and latency ≤50ms, then they are connected by a communication edge in the graph. The weight of the edge is set according to the communication quality, such as the weight of an edge with high signal strength is 0.8-1.0, and the weight of a weaker edge is 0.3-0.7.

[0055] Simultaneously, structural coupling relationships are supplemented based on the structural feature library: structural connection information of the sensor node locations is obtained from the library, such as whether two nodes are located on the same force transmission path, which can be determined through stress distribution in finite element simulation; whether they belong to adjacent critical stress areas or boundary areas; if there is a physical structural force transmission correlation, a structural edge is used to connect them, and the edge weight is set according to the structural coupling strength, such as a weight of 0.9 for directly rigidly connected nodes and 0.4-0.8 for indirectly transmitted nodes. In the final graph structure, each edge may individually represent an information channel or structural coupling, or it may combine both attributes to fully present the communication connections and structural correlations between sensor nodes.

[0056] Next, the graph structure is coupled to the global layer nodes to perform global collaborative analysis to establish global collaborative analysis results. This includes parsing the graph structure to construct a dual verification graph containing structural subgraphs constructed through structural connection relationships and similar subgraphs constructed based on node feature similarity. Based on this graph, collaborative analysis is performed starting from the central node. The global collaborative analysis results are output based on all dual verification graphs. The specific steps are explained in detail in A541-A542.

[0057] Finally, during data collaborative authentication, the global collaborative analysis results are first cross-referenced with the first and second abnormal load identifiers to verify whether there is a logical correlation between anomalies identified at different levels. For example, does the overload anomaly marked at the base layer manifest as a propagation anomaly in the regional time-series consistency analysis at the relay layer, and does it correspond to structural jump anomalies or similar isolated anomalies identified at the global layer? Subsequently, the time-series dataset is called to verify the original data such as timestamps and load values ​​corresponding to each anomaly identifier, confirming the time synchronization (based on the global clock synchronization results) and data continuity of the anomaly occurrence, such as whether the load changes before and after the anomaly conform to the mechanical laws in the structural feature library. For anomalies with contradictions, such as a base layer anomaly not supported by the relay layer or global analysis, a secondary verification is performed by combining detailed features in the time-series data, such as sampling frequency and fluctuation trends, to eliminate misjudgments caused by single-node errors or local interference. Finally, all information that has passed collaborative authentication is integrated to form load measurement results that include normal load data, anomaly load types confirmed at multiple levels, and details of their occurrence, ensuring the accuracy and reliability of the output results.

[0058] By employing a three-layer nested array for hierarchical collaborative processing—namely, basic layer preprocessing, intermediate layer regional analysis, and global layer collaborative analysis—as well as multi-information collaborative authentication of global collaborative results, two-level anomaly identification, and time-series datasets, the technical effect of improving the accuracy and reliability of load measurement results is achieved.

[0059] Furthermore, step A510 in the method provided in this application embodiment includes:

[0060] A511: Perform time-series dataset preprocessing using a lightweight algorithm integrated into the base layer nodes and establish the preprocessing results.

[0061] A512: Configure abnormal indicators associated with the base layer nodes, including overload indicators, rapid change indicators, and non-stationary fluctuation indicators.

[0062] A513: Based on the abnormal indicators, perform corresponding preprocessing result anomaly identification and establish the first abnormal load identifier.

[0063] Optionally, when the base layer node performs local preprocessing on the time series dataset, it calls the integrated simplified Kalman filter algorithm to initialize the raw load data of 100 sampling points per second and set an initial state vector, which includes the current load estimate and error covariance. The initial load estimate is the average of the first 5 sampling points, and the error covariance is set to 0.1 to reflect the initial uncertainty.

[0064] After entering the prediction phase, the algorithm predicts the estimated load value and error range for the current moment based on the load value of the previous moment and the preset state transition matrix, taking into account structural dynamic characteristics such as the upper limit of the load change rate of 5 kN / ms. For example, based on 100 kN at moment t-1, the predicted load at moment t is 102 kN ± 0.5 kN. Then, in the update phase, the current actual sampled value, such as 103 kN at moment t, is compared with the predicted value. The predicted value is corrected using Kalman gain. The algorithm calculates the load value based on the prediction error and the measurement noise variance, setting the measurement noise variance to 0.8 to filter high-frequency interference, thus obtaining the filtered load value. This removes instantaneous high-frequency noise and better reflects the overall trend.

[0065] Next, the algorithm applies a sliding window smoothing technique to the filtered sequence, with a window size of 5 sampling points. The mean of each window is calculated as the smoothed value at that moment, making the data curve smoother. Finally, key feature values ​​are extracted: by traversing the smoothed sequence, peak values ​​exceeding three adjacent points and valley values ​​falling below three adjacent points are identified, and the slope between consecutive peaks is calculated. The final result includes the filtered value, smoothed value, peak value, valley value, and slope, preserving the core trend while compressing the data volume to reduce space usage.

[0066] Subsequently, based on the stress characteristics of the target structure and the monitoring standards in the demand feature library, specific anomaly indicators are configured for the basic layer nodes. Among them, the overload indicator is set to 1.2 times the allowable load of the material in that area. For example, if the allowable load of the material is 160kN, the overload indicator is set to 192kN. The rapid change indicator is set to 50kN / ms. If the load change per unit time exceeds this value, it is judged as abnormal. The non-stationary fluctuation indicator is set to the load fluctuation amplitude of three consecutive sampling points exceeding ±10%. For example, if the baseline value is 100kN, the fluctuation exceeds the range of 90kN-110kN.

[0067] Next, the preprocessing results are compared with the above-mentioned abnormal indicators one by one to identify anomalies: if the load reaches 200kN at a certain moment in the preprocessing results, exceeding the overload index of 192kN, it is marked as an overload anomaly; if the load increases from 180kN to 195kN within 0.02 seconds, the calculated rapid change is 750kN / s, exceeding 50kN / ms, and it is marked as a rapid change anomaly; if the loads at three consecutive sampling points are 100kN, 112kN, and 98kN, and their fluctuation range is ±12% > ±10%, it is marked as a non-stationary fluctuation anomaly. These anomaly types and their occurrence times are integrated to establish the first abnormal load identifier.

[0068] By using lightweight algorithm preprocessing, targeted anomaly indicator configuration, and anomaly identification in the base layer nodes, a first anomaly payload identifier is established, achieving the effect of reducing data transmission volume and improving the timeliness and accuracy of anomaly identification.

[0069] Furthermore, step A520 in the method provided in this application embodiment includes:

[0070] A521: Obtain the communication relationship between the relay layer node and the base layer node, and call the time series dataset according to the communication relationship to establish a local sensing subgraph.

[0071] A522: Based on the local sensing subgraph and the first abnormal load identifier, perform temporal consistency analysis on the nodes within the region and establish temporal consistency anomalies, which include synchronization anomalies, propagation anomalies, and isolation anomalies.

[0072] A523: Establish a second abnormal load identifier based on the aforementioned timing consistency anomaly.

[0073] Specifically, relay layer nodes first acquire their communication relationships with base layer nodes. This relationship is based on the secondary connection table constructed during network topology initialization in step A300, clearly defining the range of base layer nodes managed by each relay layer node (e.g., a relay layer node connects to 20-30 base layer nodes). Simultaneously, the communication signal strength between nodes is recorded (e.g., a signal strength ≥ -70dBm is required for normal communication) and the data transmission frequency is 30 times per second. Based on this communication relationship, the relay layer node accesses the time-series dataset of the corresponding base layer node. These datasets have been preprocessed by the base layer and include timestamps accurate to microseconds and payload values. Next, these base layer nodes are treated as subgraph nodes, with communication relationships as edges. Edges with higher signal strength are assigned higher weights, such as 0.8-1.0, constructing a locally perceptual subgraph that visually presents the connection and data transmission status of nodes within the region.

[0074] Next, based on the local perception subgraph and the first abnormal load identifier uploaded by the base layer, the relay layer node performs a timing consistency analysis within the region. Synchronization anomalies are identified according to the global clock synchronization standard. If the timestamp deviation between a base layer node and more than 80% of the nodes in the region exceeds a certain threshold, it is necessary to combine the load propagation characteristics in the structural feature library. For example, according to finite element simulation results, the theoretical time for a load to propagate from node A to the adjacent node B is 20 milliseconds. If the actual data shows a propagation delay of 100 milliseconds, and this occurs more than three times consecutively, it is marked as a propagation anomaly. Isolated anomalies refer to the first anomaly identifier of a base layer node. For example, an overload anomaly differs significantly from three or more adjacent nodes and cannot be explained by the communication link. Other anomalies include normal signals from surrounding nodes but no anomalies, such as a node reporting overload three times consecutively while the loads of adjacent nodes remain stable within the normal range. In such cases, it is determined to be an isolated anomaly.

[0075] Finally, based on the identified synchronization anomalies, propagation anomalies, and isolation anomalies, the relay layer nodes are integrated to form a second anomaly load identifier. This identifier includes the anomaly type, such as synchronization anomaly; the node number involved, such as node 5 in the base layer; the anomaly severity, such as a time deviation of 50 microseconds; and the associated first anomaly identifier information, such as the node previously having an overload anomaly, thus providing a more comprehensive anomaly description of the load status within the region.

[0076] By acquiring communication relationships to establish a local perception subgraph, analyzing temporal consistency to identify synchronization, propagation, and isolated anomalies, and establishing a second anomalous load identifier, the comprehensiveness and accuracy of load anomaly analysis within the region are improved, providing reliable regional-level data support for global collaborative processing.

[0077] Furthermore, step A540 in the method provided in this application embodiment includes:

[0078] A541: Perform structural analysis on the graph structure to construct a node-centered dual verification graph. The dual verification graph includes a structural subgraph and a similar subgraph. The structural subgraph is constructed through structural connection relationships, and the similar subgraph is constructed based on node feature similarity.

[0079] A542: Based on the aforementioned dual verification graph, perform collaborative analysis starting from the central node, and output the global collaborative analysis results based on all dual verification graphs.

[0080] Specifically, after receiving the graph structure, the global layer nodes first perform structural analysis. During the analysis, the attribute information of all nodes in the graph structure is extracted, such as their level, monitoring area, load measurement range, and edge association type, such as information channel or structural coupling. Combined with the target structural connection relationships in the structural feature library, such as rigid connections between critical and boundary areas and force flow transmission paths, edges related to the physical connections of the structure are selected to construct a structural subgraph. For example, sensor nodes belonging to the same critical stress area and rigidly connected by bolts are connected by edges, with the edge weight set according to the connection strength: 0.9 for rigid connections and 0.5 for flexible connections.

[0081] Simultaneously, the feature vectors of each node are calculated, including sampling frequency, historical load mean, anomaly identification threshold, etc. The feature similarity of the nodes is calculated by the cosine similarity algorithm. Nodes with similarity ≥ 0.7 are connected by edges to construct a similar subgraph, with the weight of the edge being the similarity value. Finally, a dual verification graph centered on each node is formed.

[0082] After completing the construction of the dual verification graph, a collaborative analysis is performed based on the graph, starting from the central node. By analyzing the dual verification graph of each central node, the collaborative analysis results of all nodes are summarized to form a global collaborative analysis result. The specific steps are explained in detail in A542-1-A542-5.

[0083] By constructing a dual verification graph through graph structure analysis, and combining the collaborative analysis starting from the central node with the summarized results, the accuracy and comprehensiveness of the global collaborative analysis are improved, providing a reliable global basis for the certification of load measurement results.

[0084] Furthermore, step A542 in the method provided in this application embodiment includes:

[0085] A542-1: Starting from the central node, determine the multi-order neighbors based on the structural subgraph and the similar subgraph respectively, and establish the structural neighborhood set and the similar neighborhood set.

[0086] A542-2: Utilize the aforementioned structural neighborhood set to perform force flow consistency analysis between nodes and establish structural jump anomalies.

[0087] A542-3: Use the aforementioned similar neighborhood set to perform group deviation analysis and establish similar island anomalies.

[0088] A542-4: Configure load consistency score based on the structural jump anomaly, and configure similarity deviation score based on similar island anomaly.

[0089] A542-5: Establish joint anomaly confidence based on load consistency score and similarity deviation score to output global collaborative analysis results.

[0090] In this embodiment, force flow consistency analysis utilizes a structural neighborhood set to perform force flow consistency analysis between nodes, thereby establishing an analysis process for structural jump anomalies. Group deviation analysis is an analysis process for establishing similar island anomalies.

[0091] Specifically, starting from the central node, in the structural subgraph, multi-level neighbors are determined based on structural connection relationships, including nodes directly connected to the central node through structural edges, as well as nodes indirectly connected through these directly connected nodes. These nodes are integrated to form a structural neighborhood set. At the same time, in the similarity subgraph, based on node feature similarity, such as sampling features and monitoring area attributes, directly similar nodes with similar features to the central node are determined, as well as indirectly similar nodes associated through these directly similar nodes. These are integrated to form a similar neighborhood set.

[0092] Next, when performing force flow consistency analysis between nodes using the structural neighborhood set, the load transfer patterns of the target structure recorded in the structural feature library are referenced, such as the continuity and attenuation characteristics of force flow transfer in different regions, to compare the load transfer status of each node within the neighborhood. If there is a significant discontinuity between the load transfer status of a node and that of its neighboring nodes, such as when the load change of a node is not reasonably correlated with the load change of its neighboring nodes on a normal force flow path, it is determined to be an abnormal structural jump.

[0093] When performing group deviation analysis using similar neighborhood sets, focus on the common load characteristics of nodes within the similar neighborhood, such as load fluctuation range and trend. If the load characteristics of a node differ significantly from those of most nodes in its neighborhood, such as the load fluctuation of this node significantly exceeding the normal fluctuation range of other nodes under similar working conditions, and this difference cannot be explained by differences in node characteristics, then it is identified as a similar island anomaly.

[0094] Then, when configuring the load consistency score, the core basis should be the ratio of the deviation value of the structural jump anomaly to the corresponding threshold, and the principle of the larger the deviation, the lower the score should be strictly followed. First, a judgment threshold is set for the structural jump anomaly. For example, based on the force flow transmission law in the structural feature library, the threshold for a certain area is set to 12kN. That is, when the deviation between the actual load and the theoretical value exceeds 12kN, it is judged as a structural jump anomaly. The score calculation formula is: Load consistency score = 1 - (deviation value / threshold), and the result must be limited to the range of 0-1. That is, when the deviation value ≥ the threshold, the score is 0; when the deviation value ≤ 0, the score is 1. For example, if the deviation value is 6kN and does not reach the threshold, then the score = 1 - (6 / 12) = 0.5; when the deviation value is 12kN and just reaches the threshold, the score = 1 - (12 / 12) = 0; when the deviation value is 15kN and exceeds the threshold, the score is also 0. If the global synchronization threshold of 10 microseconds is met, it is judged as a synchronization anomaly. The logic of the score decreasing as the deviation increases is reflected in the propagation anomaly.

[0095] When configuring the similarity deviation score, the ratio of the deviation value of the similarity island anomaly to the corresponding threshold is calculated, following the rule that the larger the deviation, the lower the score. First, a judgment threshold is set for the similarity island anomaly. For example, based on the similarity of node features within the similar neighborhood, a threshold of 20kN is set for a certain region. That is, when the deviation of a node load from the mean of the similar neighborhood exceeds 20kN, it is judged as a similar island anomaly. The score calculation formula is: Similarity Deviation Score = 1 - (Deviation Value / Threshold), and the result is limited to the range of 0-1 (when the deviation value ≥ the threshold, the score is 0; when the deviation value ≤ 0, the score is 1). For example, if the deviation value is 10kN (not reaching the threshold), the score = 1 - (10 / 20) = 0.5; if the deviation value is 20kN (just reaching the threshold), the score = 1 - (20 / 20) = 0; if the deviation value is 30kN (exceeding the threshold), the score is 0, clearly reflecting the rule that the score decreases as the deviation increases.

[0096] Finally, when establishing the joint anomaly confidence score, a weighted summation method is used to integrate the load consistency score and similarity deviation score. The weights can be set according to the priority of structural importance and feature similarity, for example, each set to 0.5. The specific steps are as follows: For example, first obtain the load consistency score (0.3) and similarity deviation score (0.2) of a certain node, then calculate the joint anomaly confidence score using the formula: Joint Anomaly Confidence Score = Load Consistency Score × Weight + Similarity Deviation Score × Weight, resulting in 0.3 × 0.5 + 0.2 × 0.5 = 0.25. The result is then compared with a preset significant anomaly threshold (e.g., 0.3). If it is lower than the threshold (e.g., 0.25 < 0.3), it is judged as non-significant anomaly; if it is higher than or equal to the threshold (e.g., a node score of 0.4 > 0.3), it is judged as significant anomaly. Finally, the judgment results of all nodes are integrated to form a global collaborative analysis result.

[0097] By establishing sets of multi-level neighbors, identifying anomalies in two dimensions, configuring quantification scores, and constructing joint anomaly confidence, the comprehensiveness and accuracy of global collaborative analysis are improved, providing a reliable global basis for judgment of load measurement results.

[0098] Furthermore, step A400 in the method provided in this application embodiment includes:

[0099] A410: Configure a front-end event sensing unit on each sensor node of the sensor array.

[0100] A420: When performing node data collection, event perception is performed based on the aforementioned pre-event perception unit to establish node perception results.

[0101] A430: Configure the first-level neighborhood set of sensor nodes, share node perception results based on the first-level neighborhood set, and update the adaptive acquisition window.

[0102] A440: Utilize the adaptive acquisition window to perform continued data acquisition in order to establish a time-series dataset.

[0103] In this embodiment, the pre-event sensing unit is a unit configured in each sensor node of the sensor array. It is used to sense events during node data acquisition, establish node sensing results, and provide a basis for subsequent sharing of sensing results based on a first-level neighborhood set and updating the adaptive acquisition window. The adaptive acquisition window is used to continue data acquisition, and it dynamically adjusts parameters such as the sampling frequency according to the sensed event situation.

[0104] In one embodiment, before activating the sensor array for node data acquisition, a pre-event sensing unit is configured in each sensor node (including base layer and relay layer nodes). This unit integrates a simple signal discrimination module that can monitor the abrupt changes in load signals in real time, such as the 50kN / ms rate of change in rapid change indicators and amplitude abrupt changes in non-stationary fluctuation indicators. It can quickly respond to potential load events without relying on complex calculations. In the prior art, sensors often use a fixed frequency, such as 100 times per second, to collect data. This can easily generate redundant data when the load is stable, or lose key information due to excessively large sampling intervals during transient changes. The pre-event sensing unit can capture event signs in advance, providing a basis for dynamically adjusting the acquisition strategy.

[0105] During node data acquisition, the pre-event sensing unit starts working first, continuously monitoring the dynamic signals of the local load in real time. It calculates the instantaneous load value, the change per unit time, and the stability of signal fluctuations, such as whether the fluctuation amplitude of 10 consecutive sampling points exceeds ±5%. When any indicator exceeds a preset threshold, such as a rapid change indicator exceeding 50 kN / ms, a load exceeding the 200 kN overload threshold, or a fluctuation amplitude exceeding ±5% for three consecutive sampling points, a sensing response is immediately triggered. The precise time of the event is recorded, along with a timestamp after global clock synchronization, the current load value, the specific type of indicator exceeding the threshold, and details of the change, such as an increase from 290 kN to 360 kN within 0.05 seconds. This forms the node sensing result and is uniformly marked as a potential abnormal event. This design eliminates the need for complex data processing, using only lightweight algorithms for initial judgment. It can quickly generate sensing results, providing immediate basis for subsequent information sharing of the first-level neighborhood set and adjustment of the adaptive acquisition window. This ensures rapid capture of load anomalies while avoiding response delays caused by complex analysis.

[0106] Subsequently, a primary neighborhood set is configured for each sensor node. This is based on the primary connection table constructed during network topology initialization in step A300, determining 3-5 neighboring nodes for each node, such as the neighboring basic layer nodes of a basic layer node. Each node shares its own node sensing results through this set. For example, when a basic layer node senses a rapidly changing load, it sends the result to the other four nodes in its primary neighborhood. Upon receiving the result, the neighboring nodes compare it with their own sensing results. If two or more nodes also sense a similar signal, it is determined to be a significant event within the region, requiring increased acquisition accuracy. If only a single node senses the signal, it is determined to be local interference, and normal acquisition is maintained. Based on this shared analysis, the node automatically updates its adaptive acquisition window: under normal conditions, the window is set to 100ms / sample; during significant events within the region, it is adjusted to 10ms / sample; and during local interference, it remains at 100ms but with extended monitoring duration, ensuring the capture of critical data while reducing invalid acquisition.

[0107] Finally, using the updated adaptive acquisition window, each node continues to perform data acquisition: the base layer nodes record the real-time values ​​of the load at the adjusted frequency, such as generating 100 data points per second at 10ms / time, and the relay layer nodes synchronously acquire the aggregated signals in the area. All acquired data are accompanied by a timestamp after global clock synchronization, and finally integrated into a time series dataset containing time series, signal strength, and event markers.

[0108] If sensor data acquisition uses a fixed window, it is easy to miss transient loads or have data redundancy in stable conditions, resulting in time series datasets that either lack key information or have an excessively high proportion of invalid data. However, by utilizing proactive event perception, neighborhood sharing, and adaptive window adjustment, the acquisition strategy can dynamically match the characteristics of load changes, ensuring that the time series data contains complete details of abnormal events while avoiding the accumulation of redundant information.

[0109] By configuring a front-end event sensing unit for the sensor node, combining the sensing results of the first-level neighborhood set to update the adaptive acquisition window, and then continuously acquiring data to build a time-series dataset, the targeting and efficiency of data acquisition are improved, and a high-quality time-series dataset is constructed.

[0110] Furthermore, step A500 in the method provided in this application embodiment includes:

[0111] A610: Based on the load measurement results, identify abnormal loads and establish an abnormal load identification focus.

[0112] A620: Call the historical load database, perform attention verification based on the historical load database and the abnormal load identifier, and update the abnormal load identifier.

[0113] Optionally, when identifying abnormal loads based on the load measurement results after collaborative authentication, it is necessary to combine the load change characteristics in the measurement data, such as numerical magnitude, rate of change, and fluctuation stability, and compare them with the normal load range of that area in the structural feature library, such as the normal fluctuation range under static conditions and the peak upper limit under dynamic conditions, to identify load states that exceed the normal range. For example, when the load in a critical area continuously exceeds the allowable range of the material, or undergoes drastic changes in a short period of time, or the fluctuation amplitude is far greater than the normal level during stable operation, these states are marked as abnormal. Furthermore, they are prioritized according to the potential impact of the abnormalities (such as whether they involve core load-bearing components or whether they may cause structural damage), forming a list of abnormal loads requiring special attention.

[0114] When accessing the historical load database, this database must contain historical load records of the target structure under different operating conditions, past anomaly cases, and their handling results, such as the frequency, duration, and final impact of similar anomalies. Each anomaly in the anomaly load identification watchlist is compared with historical data: if the characteristics of the current anomaly (such as the area of ​​occurrence and load change pattern) highly match a certain type of anomaly that has occurred multiple times in the past, and past cases show that its impact on the structure is limited, then it can be confirmed as a common anomaly; if the current anomaly has never appeared in the historical record, or its characteristics differ significantly from known anomalies, then it is determined to be a novel anomaly. The anomaly load identification is updated based on the comparison results, supplementing the historical correlation information or novel markers for the anomaly. Examples of verification and updating for different anomaly types are shown in Table 1. By using historical data matching analysis, the attributes of the current anomaly can be clarified, avoiding misjudging interference as serious anomalies, while also identifying novel anomalies to enhance attention.

[0115] By marking anomalies based on measurement results, establishing a focus, and calling historical databases to verify and update the markers, the accuracy and relevance of anomaly load marking are improved, misjudgments are reduced, and new anomalies are identified in a timely manner.

[0116] Table 1: Exception Type Validation and Update Status

[0117] Exception types Current measurement features Historical database matching features Verification results Updated logo Overload abnormality 200kN, lasting 10s In a certain year, the temperature in the same region was 198 kN and 205 kN, lasting for 8-12 seconds. Historical reenactment type Overload anomaly (historical recurrence, averaging 2 times per year) Rapid change anomaly 75kN / ms, instantaneous The historical maximum was 60 kN / ms, with no similar instantaneous characteristics. New anomaly Rapid change anomaly (first occurrence, requires close monitoring) Non-stationary fluctuation anomaly Fluctuation of ±12% over 5 consecutive sampling points There were three fluctuations of ±10%-15% in a certain year, all of which were disturbances. Interference type Non-stationary fluctuation anomaly (caused by interference, no warning issued at this time)

[0118] Furthermore, step A620 in the method provided in this application embodiment includes:

[0119] A621: Identify the warning level for updated abnormal load identifiers and configure visual warning anomalies.

[0120] A622: Based on the visual warning anomaly and warning level identification results, perform early warning issuance management.

[0121] In one embodiment, when processing updated abnormal load identifiers, the warning level needs to be identified by combining the characteristics of the abnormality. These characteristics include the type of abnormality (e.g., overload, rapid change), the area where it occurs (e.g., critical stress areas or boundary areas), the duration, and the potential impact on the structure. By comprehensively evaluating these characteristics, abnormalities are classified into different warning levels, such as low level reflecting minor abnormalities, medium level indicating the need for attention, and high level indicating a possible serious risk. Simultaneously, visual warning abnormalities are configured for these different levels, presenting abnormal information in an intuitive way, such as using different colors to indicate different levels and using graphics to show the location and extent of the abnormality, allowing relevant personnel to quickly understand the abnormal situation.

[0122] Subsequently, based on the configured visual warning anomalies and the identified warning levels, early warning issuance management is implemented. For low-level warnings, they may only be recorded internally and maintenance personnel will be reminded to check them periodically; medium-level warnings require timely notification of relevant personnel to arrange for inspection; high-level warnings require immediate triggering of the emergency response mechanism, notification of decision-makers, and initiation of emergency handling procedures to ensure that relevant personnel can take corresponding measures according to the severity of the warning.

[0123] By identifying and visually configuring the warning levels of abnormal load identifiers, and then implementing hierarchical warning management based on the results, the effect of making the warning information clear and intuitive and the response measures precise and effective has been achieved.

[0124] In summary, the distributed cooperative measurement method for load sensor arrays provided in this application has the following technical effects:

[0125] This application establishes a structural feature library and a demand feature library by performing finite element simulation of the target structure. Based on this, a three-layer nested sensor array is deployed. After network topology initialization and global clock synchronization, the nodes are activated to collect data. Then, the three-layer nodes collaboratively process the time-series dataset and perform data collaborative authentication to output load measurement results. This makes the load measurement more accurate and reliable, achieving comprehensive and accurate acquisition of structural load information and meeting the technical requirements for accurate measurement and effective control of structural loads.

[0126] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a distributed cooperative measurement system for a load sensor array, the system comprising:

[0127] Feature library construction module 1 is used to perform finite element simulation of the target structure, identify the key stress area, boundary area and load type based on the finite element simulation results, and establish a structural feature library and a requirement feature library.

[0128] Sensor array deployment module 2 is used to deploy a sensor array according to the structural feature library and the demand feature library. The sensor array is a three-layer nested array, including a base layer node, a relay layer node, and a global layer node.

[0129] Network topology initialization module 3 is used to configure the communication protocol of the sensor array and then complete the network topology initialization.

[0130] The time series dataset construction module 4 is used to activate the sensor array to collect node data and establish a time series dataset after configuring global clock synchronization.

[0131] The load measurement result acquisition module 5 is used to perform time-series dataset collaborative processing under the three-layer nested array according to the network topology, and output load measurement results according to the collaborative authentication results.

[0132] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0133] Local node preprocessing of the time-series dataset is performed at the base layer node to establish a first abnormal load identifier; the time-series dataset and the first abnormal load identifier are uploaded to the relay layer node, and regional load analysis is performed to establish a second abnormal load identifier; a graph structure is established based on the network topology and structural feature library, where the nodes of the graph structure are sensor nodes, and the edges of the graph structure represent the information channel or structural coupling between any two nodes; the graph structure is coupled to the global layer node, and global collaborative analysis is performed to establish a global collaborative analysis result; data collaborative authentication is performed based on the global collaborative analysis result, the first abnormal load identifier, the second abnormal load identifier, and the time-series dataset, and the load measurement result is output.

[0134] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0135] A lightweight algorithm integrated into the base layer node is used to perform time series dataset preprocessing to establish preprocessing results; anomaly indicators associated with the base layer node are configured, including overload indicators, rapid change indicators, and non-stationary fluctuation indicators; anomaly identification of the corresponding preprocessing results is performed based on the anomaly indicators to establish the first anomalous load identifier.

[0136] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0137] Obtain the communication relationship between relay layer nodes and base layer nodes, call the time series dataset according to the communication relationship, and establish a local sensing subgraph; perform time series consistency analysis on nodes in the region based on the local sensing subgraph and the first abnormal load identifier, and establish time series consistency anomalies, including synchronization anomalies, propagation anomalies, and isolation anomalies; establish a second abnormal load identifier based on the time series consistency anomalies.

[0138] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0139] The graph structure is parsed to construct a node-centric dual verification graph, which includes a structural subgraph and a similar subgraph. The structural subgraph is constructed through structural connections, and the similar subgraph is constructed based on node feature similarity. Based on the dual verification graph, a collaborative analysis is performed starting from the central node, and the global collaborative analysis results are output based on all dual verification graphs.

[0140] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0141] Starting from the central node, multi-order neighbors based on the structural subgraph and similar subgraph are determined respectively, and structural neighborhood sets and similar neighborhood sets are established. The structural neighborhood sets are used to perform force flow consistency analysis between nodes to establish structural jump anomalies. The similar neighborhood sets are used to perform group deviation analysis to establish similar island anomalies. Load consistency scores are configured according to the structural jump anomalies, and similarity deviation scores are configured according to the similar island anomalies. A joint anomaly confidence score is established based on the load consistency score and the similarity deviation score to output the global collaborative analysis results.

[0142] Furthermore, the time-series dataset construction module 4 is used to perform the following steps:

[0143] A pre-event sensing unit is configured in each sensor node of the sensor array; when performing node data acquisition, event sensing is performed based on the pre-event sensing unit to establish node sensing results; a first-level neighborhood set of the sensor node is configured, node sensing results are shared according to the first-level neighborhood set, and the adaptive acquisition window is updated; continued data acquisition is performed using the adaptive acquisition window to establish a time-series dataset.

[0144] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0145] Based on the load measurement results, abnormal loads are identified and an abnormal load identification focus is established; the historical load database is called, and the focus is verified according to the historical load database and the abnormal load identification focus, and the abnormal load identification is updated.

[0146] Furthermore, the load measurement result acquisition module 5 is used to perform the following steps:

[0147] The updated abnormal load identifiers are used to identify the warning level and a visual warning anomaly is configured; warning issuance management is carried out based on the visual warning anomaly and the warning level identification results.

[0148] The distributed collaborative measurement system for load sensor arrays provided in this embodiment of the invention can execute the distributed collaborative measurement method for load sensor arrays provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0149] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0150] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A distributed collaborative measurement method for a load sensor array, characterized in that, The method includes: Perform finite element simulation of the target structure, identify key stress areas, boundary areas, and load types based on the finite element simulation results, and establish a structural feature library and a demand feature library. The sensor array is deployed according to the structural feature library and the demand feature library. The sensor array is a three-layer nested array, including a basic layer node, a relay layer node, and a global layer node. The global layer node can be deployed near the control system. After configuring the communication protocol for the sensor array, the network topology initialization is completed; After configuring global clock synchronization, activate the sensor array to collect node data and establish a time-series dataset; Based on the network topology, perform collaborative processing of time-series datasets under a three-layer nested array, and output load measurement results based on the collaborative authentication results; The step of performing collaborative processing of time-series datasets under a three-layer nested array based on the network topology includes: Local node preprocessing of the time series dataset is performed at the base layer node to establish the first abnormal load identifier; Upload the time-series dataset and the first abnormal load identifier to the relay layer node, perform load analysis within the region, and establish a second abnormal load identifier; A graph structure is established based on a network topology and structural feature library. The nodes of the graph structure are sensor nodes, and the edges of the graph structure represent the information channels or structural coupling between any two nodes. Couple the graph structure to global layer nodes, perform global collaborative analysis, and establish global collaborative analysis results; Based on the global collaborative analysis results, the first abnormal load identifier, the second abnormal load identifier, and the time series dataset, data collaborative authentication is performed, and the load measurement results are output. The process of coupling the graph structure to global layer nodes, performing global collaborative analysis, and establishing global collaborative analysis results includes: The graph structure is parsed to construct a node-centered dual verification graph. The dual verification graph includes a structural subgraph and a similarity subgraph. The structural subgraph is constructed through structural connection relationships, and the similarity subgraph is constructed based on node feature similarity. Based on the dual verification graph, a collaborative analysis is performed starting from the central node, and the global collaborative analysis results are output based on all the dual verification graphs. The collaborative analysis based on the dual verification graph, starting from the central node, includes: Starting from the central node, multi-level neighbors based on the structural subgraph and similar subgraph are determined respectively, and a structural neighborhood set and a similar neighborhood set are established. Specifically, starting from the central node, in the structural subgraph, multi-level neighbors are determined according to the architecture connection relationship, including nodes directly connected to the central node through structural edges and nodes indirectly connected to directly connected nodes, and integrated to form a structural neighborhood set. In the similar subgraph, based on the node feature similarity, directly similar nodes with similar features to the central node and indirectly similar nodes associated with directly similar nodes are determined, and integrated to form a similar neighborhood set. The force flow consistency analysis between nodes is performed using the structural neighborhood set to establish structural jump anomalies; Using the aforementioned similar neighborhood set, a group deviation analysis is performed to establish similar isolated anomalies; Load consistency scores are configured based on the structural jump anomalies, and similarity deviation scores are configured based on similar island anomalies. A joint anomaly confidence score is established based on load consistency score and similarity deviation score to output global collaborative analysis results.

2. The distributed collaborative measurement method for a load sensor array as described in claim 1, characterized in that, The step of performing local node preprocessing of the time-series dataset at the base layer node to establish a first anomalous load identifier includes: A lightweight algorithm integrated into the base layer nodes is used to perform time series dataset preprocessing and establish the preprocessing results; Configure abnormal indicators associated with the base layer nodes, including overload indicators, rapid change indicators, and non-stationary fluctuation indicators; Based on the abnormal indicators, the corresponding preprocessing results are identified as abnormal, and the first abnormal load identifier is established.

3. The distributed collaborative measurement method for a load sensor array as described in claim 1, characterized in that, The step of uploading the time-series dataset and the first anomalous load identifier to the relay layer node, performing regional load analysis, and establishing a second anomalous load identifier includes: Obtain the communication relationship between the relay layer node and the base layer node, and call the time series dataset according to the communication relationship to establish a local sensing subgraph; Based on the local sensing subgraph and the first abnormal load identifier, a temporal consistency analysis of the nodes in the region is performed to establish temporal consistency anomalies, which include synchronization anomalies, propagation anomalies, and isolation anomalies. A second abnormal load identifier is established based on the aforementioned timing consistency anomaly.

4. The distributed collaborative measurement method for a load sensor array as described in claim 1, characterized in that, The activated sensor array performs node data acquisition to establish a time-series dataset, including: Configure a front-end event sensing unit in each sensor node of the sensor array; When performing node data collection, event perception is performed based on the aforementioned pre-event perception unit to establish node perception results; Configure a first-level neighborhood set for sensor nodes, share node perception results based on the first-level neighborhood set, and update the adaptive acquisition window. The first-level neighborhood set is a first-level connection table constructed based on network topology initialization, used to share its own node perception results. Continue data acquisition using the adaptive acquisition window to build a time-series dataset.

5. The distributed collaborative measurement method for a load sensor array as described in claim 1, characterized in that, The step of outputting load measurement results based on collaborative authentication results includes: Based on the load measurement results, abnormal loads are identified, and an abnormal load identification focus is established. The historical load database is invoked, and the abnormal load identifier is used for attention verification based on the historical load database and the abnormal load identifier. The abnormal load identifier is then updated.

6. The distributed collaborative measurement method for a load sensor array as described in claim 5, characterized in that, The updated abnormal load identifier includes: The updated abnormal load identifiers are used to identify the warning level, and a visual warning of the abnormality is configured. Early warning issuance management is carried out based on the results of the visualization of early warning anomalies and early warning level identification.

7. A distributed cooperative measurement system for a load sensor array, characterized in that, A distributed collaborative measurement method for implementing the load sensor array according to any one of claims 1-6, the system comprising: The feature library construction module is used to perform finite element simulation of the target structure, identify key stress areas, boundary areas, and load types based on the finite element simulation results, and establish a structural feature library and a requirement feature library. The sensor array deployment module is used to deploy a sensor array according to the structural feature library and the demand feature library. The sensor array is a three-layer nested array, including a base layer node, a relay layer node, and a global layer node. The network topology initialization module is used to configure the communication protocol for the sensor array and then complete the network topology initialization. The time-series dataset construction module is used to activate the sensor array to collect node data and build a time-series dataset after configuring global clock synchronization. The load measurement result acquisition module is used to perform collaborative processing of time-series datasets under a three-layer nested array according to the network topology, and output load measurement results based on the collaborative authentication results.

Images