High-position remote landslide disaster early warning method and system based on multi-source data fusion

Abstract

The application discloses a high-position remote landslide disaster early warning method and system based on multi-source data fusion, and relates to the technical field of early warning. First, a hardware risk perception unit is deployed at the data acquisition front end of an early warning server, running data streams of each sensor data channel are obtained through the hardware risk perception unit, and multi-dimensional labels of each sensor data channel are synchronously acquired, then according to the running data streams, interruption risk index values of each sensor data channel are analyzed and obtained, the multi-dimensional labels are analyzed in real time, early warning results of each sensor data channel are marked, and then based on the interruption risk index values and the early warning results of each sensor data channel, the running mode of an early warning dispatcher of each sensor data channel is adjusted, based on the adjusted running mode of the early warning dispatcher, scheduling execution adjustment of each sensor data channel is carried out, and high-position remote landslide disaster early warning is carried out after the scheduling execution adjustment, thereby improving the reliability and timeliness of high-position remote landslide disaster early warning.

Description

Technical Field

[0001] This invention relates to the field of early warning technology, specifically to a method and system for early warning of high-altitude remote landslide disasters based on multi-source data fusion. Background Technology

[0002] With the increasing demand for geological disaster monitoring and early warning, high-altitude remote landslide disaster early warning systems constructed using multi-source data fusion technology have become an important means to ensure safety and infrastructure stability. The system aims to achieve accurate early warning of high-altitude remote landslide disasters by integrating data from various sensors, such as displacement, humidity, and pressure, and conducting comprehensive analysis. Traditional multi-source data fusion early warning methods are mainly based on the analysis of single-type data or simple data overlay, hoping to discover potential risks of landslide disasters in a timely manner through these methods.

[0003] However, in practical applications, especially under complex geological conditions and variable environmental factors, this approach has many limitations. For example, during the landslide acceleration phase, a large number of sensors will send high-frequency alarm data almost simultaneously, instantly creating a hardware interruption storm. At this time, the traditional static priority scheduling used by existing technologies may cause priority inversion or queue overflow, resulting in the interruption of the most critical displacement data being overwhelmed. More seriously, if the system is performing a background silent update at this time, the update process may occupy critical hardware resources, causing non-deterministic delays in interrupt response, thus causing the entire system to freeze for several seconds at the most critical moment. This not only makes it impossible to transmit early warning information in a timely and accurate manner, but may also miss the best early warning opportunity, bringing great difficulties to subsequent disaster prevention and mitigation work and seriously threatening life safety and the stable operation of infrastructure. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a high-altitude remote landslide disaster early warning method and system based on multi-source data fusion, which can effectively solve the problems mentioned in the background technology.

[0005] To achieve the above objectives, the first aspect of the present invention is implemented through the following technical solution: a high-level remote landslide disaster early warning method based on multi-source data fusion, including deploying a hardware risk perception unit at the data acquisition front end of the early warning server, acquiring the operating data stream of each sensor data channel through the hardware risk perception unit, and simultaneously acquiring the multi-dimensional tags of each sensor data channel.

[0006] Based on the operational data stream of each sensor data channel, the interruption risk index value of each sensor data channel is obtained through analysis.

[0007] Real-time analysis of the multidimensional labels of each sensor data channel to mark the early warning results of each sensor data channel.

[0008] Based on the interruption risk index values ​​of each sensor data channel and the early warning results of each sensor data channel, the operation mode of the early warning scheduler for each sensor data channel is adjusted.

[0009] Based on the adjusted operation mode of the early warning scheduler for each sensor data channel, the scheduling and execution of each sensor data channel are adjusted, and high-level remote landslide disaster early warning is carried out after the scheduling and execution adjustment.

[0010] Furthermore, the specific process of acquiring the operational data stream of each sensor data channel through the hardware risk perception unit is as follows: the operational data stream of each sensor data channel within a preset monitoring period is acquired through the hardware risk perception unit. The operational data stream of each sensor data channel includes the total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response delay of each sensor data channel.

[0011] Furthermore, the process of synchronously acquiring multi-dimensional tags for each sensor data channel is as follows: the hardware risk perception unit continuously monitors the hardware level signal of each sensor data channel. When the hardware level signal of a certain sensor data channel meets the preset interrupt triggering condition, the hardware risk perception unit generates a corresponding interrupt event. The interrupt event includes the sensor data channel identifier, timestamp, and event type code.

[0012] While generating the corresponding interrupt event, the hardware risk perception unit queries the tag mapping table preset in the disaster early warning database based on the sensor data channel identifier, extracts the static priority value preset for the sensor data channel, and records it as the static priority value of the sensor data channel.

[0013] The dynamic confidence level value of the sensor data channel is obtained by synchronously matching the event type encoding.

[0014] Simultaneously, based on the event type code, a preset resource mapping table is retrieved to obtain a list of hardware resources required to process the interrupt event of that type, which serves as the resource dependency signature for the sensor data channel.

[0015] The static priority value, dynamic confidence value, and resource dependency signature of the sensor data channel are encapsulated into a multi-dimensional label corresponding to the interruption event, and denoted as the multi-dimensional label of the sensor data channel.

[0016] Furthermore, the analysis yields the interruption risk index value for each sensor data channel. The specific analysis process is as follows: based on the running data stream of each sensor data channel, the interruption risk index value for each sensor data channel is obtained. The interruption risk index value for each sensor data channel represents the quantitative result of the total number of interruption events, interruption frequency, memory utilization rate, and average interruption response delay of each sensor data channel, which together quantify the degree of blocking risk of that sensor data channel.

[0017] Furthermore, the analysis of the multidimensional tags of each sensor data channel specifically involves the following steps: analyzing the multidimensional tags of each sensor data channel, extracting the static priority value, dynamic confidence value, and resource dependency signature of each sensor data channel encapsulated in the multidimensional tags of each sensor data channel, comparing the static priority value with a preset highest priority threshold, comparing the dynamic confidence value with a preset dynamic confidence threshold, and simultaneously comparing the resource dependency signature with a preset hardware resource warning zone list.

[0018] Furthermore, the specific process of marking the warning results of each sensor data channel is as follows: when the static priority value of a certain sensor data channel is equal to the preset highest priority threshold, and the dynamic confidence value is greater than or equal to the preset dynamic confidence threshold, and the intersection of the resource dependency signature and the preset hardware resource warning zone list is an empty set, then the warning task carried by the sensor data channel is determined to be a critical warning task, and the warning result of the sensor data channel is marked as needing priority scheduling; otherwise, the warning result of the sensor data channel is marked as not needing priority scheduling.

[0019] Furthermore, the specific adjustment process for adjusting the early warning scheduler operation mode of each sensor data channel is as follows: the interruption risk index value of each sensor data channel is compared with the preset interruption risk index threshold. When the interruption risk index value of a certain sensor data channel is greater than or equal to the interruption risk safety threshold, and the early warning result of the sensor data channel is marked as needing priority scheduling, the early warning scheduler operation mode of the sensor data channel is switched from the efficiency priority mode to the survival mode; otherwise, the early warning scheduler operation mode of the sensor data channel continues to operate in the efficiency priority mode.

[0020] Furthermore, the specific process of scheduling and adjusting the data channels of each sensor is as follows: based on the early warning scheduler operation mode of the sensor data channel, the corresponding scheduling adjustment is performed.

[0021] When the early warning scheduler is in efficiency-first mode, the scheduler obtains the comprehensive scheduling score of each sensor data channel based on the static priority value and dynamic confidence value of each sensor data channel through a preset weight calculation formula, and sequentially schedules and executes the interruption event handling tasks of all sensor data channels in descending order of comprehensive scheduling score.

[0022] When the early warning scheduler switches to survival mode, the scheduler immediately stops and clears the interrupt event handling task queue of the current sensor data channel. At the same time, based on the early warning results of each sensor data channel, it reconstructs an execution queue that only contains sensor data channels whose early warning results are marked as needing priority scheduling. The early warning scheduler only extracts interrupt event handling tasks from this execution queue for scheduling and execution, and performs a suspension operation on interrupt handling requests initiated by sensor data channels whose early warning results are marked as not needing priority scheduling.

[0023] Furthermore, the process of conducting high-level remote landslide disaster early warning after scheduling and execution adjustment is as follows: after the early warning scheduler completes the scheduling and execution adjustment of each sensor data channel, it receives the multi-source monitoring data stream of each scheduled sensor data channel to obtain the high-level remote landslide disaster early warning signal.

[0024] The second aspect of the present invention provides a high-altitude remote landslide disaster early warning system based on multi-source data fusion, comprising: a data acquisition module, used to deploy a hardware risk perception unit at the data acquisition front end of the early warning server, to acquire the operating data stream of each sensor data channel through the hardware risk perception unit, and to simultaneously acquire the multi-dimensional tags of each sensor data channel.

[0025] The risk analysis module is used to analyze the interruption risk index value of each sensor data channel based on the operating data stream of each sensor data channel.

[0026] The tag analysis module is used to analyze the multidimensional tags of each sensor data channel in real time and mark the early warning results of each sensor data channel.

[0027] The mode adjustment module is used to adjust the operation mode of the early warning scheduler for each sensor data channel based on the interruption risk index value of each sensor data channel and the early warning results of each sensor data channel.

[0028] The early warning analysis module is used to adjust the scheduling and execution of each sensor data channel based on the adjusted operation mode of the early warning scheduler, and to conduct high-level remote landslide disaster early warning after the scheduling and execution adjustment.

[0029] The present invention has the following beneficial effects:

[0030] (1) This invention uses an independent hardware risk perception unit to monitor and quantify the system’s underlying interrupt traffic and resource status in real time. Based on this quantitative indicator, the scheduler is driven to dynamically reconstruct the global operation mode and implement preventive isolation for key hardware resources. This transforms the system from passively responding to faults to actively ensuring survival. This transformation shifts the system performance protection target from optimizing the average performance index in a statistical sense to ensuring that the response delay of the core early warning task is still within the preset safety threshold under severe and complex scenarios such as business peaks and internal updates.

[0031] (2) This invention proposes a three-dimensional tagging system of static priority, dynamic confidence and resource dependency signature, which gives each interruption event context information of different dimensions. The dynamic confidence dimension is dynamically assigned according to the signal quality of the data source or preset rules, so that the scheduler can intelligently identify and reduce the scheduling order of tasks that have high static priority but questionable real-time data quality in a timely manner. Thus, when system resources are tight, it effectively avoids the ineffective occupation and blockage of low-quality data streams on limited processing resources and key hardware channels.

[0032] (3) This invention opens up an exclusive and deterministic processing path for the core early warning data stream through the scheduling strategy in survival mode. This mechanism ensures that even under the worst pressure scenario where the system is simultaneously subjected to external business interruption storms and internal silent updates, the core data supply chain supporting the final early warning decision can still maintain continuity, integrity and low latency. This transforms the theoretical potential of the upper-layer early warning algorithm into a stable and predictable actual output, significantly enhancing the system's functional resilience and overall service reliability under extreme pressure.

[0033] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0035] Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Please see Figure 1As shown, the first aspect of the present invention provides a technical solution: a high-level remote landslide disaster early warning method based on multi-source data fusion, including deploying a hardware risk perception unit at the data acquisition front end of the early warning server, acquiring the operating data stream of each sensor data channel through the hardware risk perception unit, and simultaneously acquiring the multi-dimensional tags of each sensor data channel.

[0038] It should be added that the early warning server is the core device responsible for managing and coordinating the data processing and workflow execution of the entire high-altitude remote landslide disaster early warning system to realize the disaster early warning function; the hardware risk perception unit is deployed at the data acquisition front end of the early warning server, used to collect the operating data streams of each sensor data channel and simultaneously acquire multi-dimensional tags, providing a data foundation for subsequent early warning analysis; the sensor data channel connects the sensors and the early warning server, used to transmit sensor monitoring data, providing a pathway for information sources for high-altitude remote landslide disaster early warning analysis.

[0039] Specifically, the hardware risk perception unit collects the operating data streams of each sensor data channel. The specific process is as follows: the hardware risk perception unit collects the operating data streams of each sensor data channel within a preset monitoring period. The operating data streams of each sensor data channel include the total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response delay of each sensor data channel.

[0040] It should be noted that the total number of interruption events for each sensor data channel within the regulatory period reflects the data activity level and potential load pressure of that channel during the regulatory period. This number is obtained directly by accumulating interruption events that meet the triggering conditions through the event counter in the hardware risk perception unit. The interruption frequency of each sensor data channel represents the density of data arrivals per unit time, reflecting the requirements for real-time processing. This is calculated by dividing the total number of interruption events by the duration of the regulatory period. The memory utilization rate of each sensor data channel characterizes the memory resources occupied when processing data from that channel. This is obtained by querying the real-time usage of the memory buffer dynamically allocated by the operating system kernel for the interrupt service routine of that channel and calculating the ratio of this to the pre-allocated total buffer size. The average interrupt response latency of each sensor data channel is used to directly measure the real-time performance of responding to interruption events of that channel. Its value is obtained by calculating the difference between the high-precision timestamp recorded by the hardware risk perception unit when the interruption event is generated and the timestamp recorded by the scheduler when the event begins to be processed. This single latency is then averaged over all interruption events within a regulatory period.

[0041] Specifically, the multi-dimensional tags of each sensor data channel are acquired synchronously. The specific process is as follows: the hardware level signal of each sensor data channel is continuously monitored by the hardware risk perception unit. When the hardware level signal of a certain sensor data channel meets the preset interrupt triggering condition, the hardware risk perception unit generates a corresponding interrupt event. The interrupt event includes the sensor data channel identifier, timestamp, and event type code.

[0042] It should be added that the preset interrupt triggering condition is as follows: the hardware risk perception unit internally presets a set of level determination parameters for each sensor data channel. This set of level determination parameters includes a high-level effective threshold, a low-level invalid threshold, and a minimum stable duration. During continuous monitoring, the hardware risk perception unit determines that the hardware level signal of a sensor data channel meets the interrupt triggering condition if and only if it detects that the hardware level signal of a certain sensor data channel rises from below the low-level invalid threshold to above the high-level effective threshold, and then continuously maintains a state above the high-level effective threshold to reach or exceed the minimum stable duration.

[0043] While generating the corresponding interrupt event, the hardware risk perception unit queries the tag mapping table preset in the disaster early warning database based on the sensor data channel identifier, extracts the static priority value preset for the sensor data channel, and records it as the static priority value of the sensor data channel.

[0044] The dynamic confidence level value of the sensor data channel is obtained by synchronously matching the event type encoding.

[0045] It should be added that the event type code is matched with the dynamic confidence value corresponding to each event type code stored in the disaster early warning database, and the dynamic confidence value corresponding to the event type code is counted and recorded as the dynamic confidence value of the sensor data channel. The dynamic confidence value represents the real-time reliability of the sensor data corresponding to the interruption event at the current moment.

[0046] Simultaneously, based on the event type code, a preset resource mapping table is retrieved to obtain a list of hardware resources required to process the interrupt event of that type, which serves as the resource dependency signature for the sensor data channel.

[0047] The static priority value, dynamic confidence value, and resource dependency signature of the sensor data channel are encapsulated into a multi-dimensional label corresponding to the interruption event, and denoted as the multi-dimensional label of the sensor data channel.

[0048] In this embodiment, by encapsulating each interrupt event with a three-dimensional label containing business logic, data quality, and resource requirements in real time, the scheduler can make refined decisions based on multi-dimensional information, rather than relying solely on the original interrupt priority. This effectively solves the problem of high-priority but low-quality data crowding out resources due to temporary sensor failures or signal interference. Without labeling and encapsulation, the scheduler lacks sufficient contextual information for intelligent judgment.

[0049] Based on the operational data stream of each sensor data channel, the interruption risk index value of each sensor data channel is obtained through analysis.

[0050] Specifically, the interruption risk index value of each sensor data channel is obtained through analysis. The specific analysis process is as follows: based on the running data stream of each sensor data channel, the interruption risk index value of each sensor data channel is obtained. The interruption risk index value of each sensor data channel represents the quantitative result of the total number of interruption events, interruption frequency, memory utilization rate and average interruption response delay of each sensor data channel, which together quantify the degree of blocking risk of the sensor data channel.

[0051] It should be added that the total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response delay of each sensor data channel are compared with the preset maximum values ​​of the total number of interrupt events, interrupt frequency, memory utilization, and interrupt response delay, respectively. This process is then normalized, converting the original data into dimensionless values ​​between 0 and 1, resulting in the normalized total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response delay of each sensor data channel.

[0052] In this embodiment, the interruption risk index value of each sensor data channel can be obtained through the following analysis method, with the specific analysis conditions as follows:

[0053] ;

[0054] In the formula, This represents the interruption risk index value of the i-th sensor data channel. This represents the normalized total number of interrupt events for the i-th sensor data channel. This represents the weighting factor corresponding to the set total number of interrupt events. This represents the normalized interrupt frequency of the i-th sensor data channel. This represents the weighting factor corresponding to the set interrupt frequency. This represents the normalized memory utilization rate of the i-th sensor data channel. This represents the weighting factor corresponding to the set memory utilization rate. This represents the normalized average interrupt response delay of the i-th sensor data channel. This represents the weighting factor corresponding to the set average interruption response delay, where i represents the number of each sensor data channel, i=1, 2, 3, ..., n, and n represents the total number of sensor data channels.

[0055] It should be added that, in this embodiment, the preset weighting factors corresponding to the total number of interruption events, interruption frequency, memory utilization, and average interruption response delay are obtained from the disaster early warning database.

[0056] It should be explained that these weighting factors are used to adjust the importance of the data in the operational data stream of each sensor data channel in the process of analyzing and obtaining the interruption risk index value. For example, the disaster early warning database sets a mapping relationship between the operational data characteristics of each sensor data channel and the weighting factors. Through the pre-set mapping relationship, the weighting factors corresponding to the real-time operational data stream of each sensor data channel can be matched. By matching the operational data stream of each sensor data channel with the pre-set mapping relationship, the weighting factors corresponding to the total number of interruption events, interruption frequency, memory utilization, and average interruption response delay time can be obtained.

[0057] In this implementation plan, the total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response latency of each sensor data channel are correlated and not independent. For example, a sharp increase in the total number of interrupt events will directly lead to an increase in interrupt frequency. High-frequency interrupts will exacerbate the CPU scheduling burden and memory allocation requests, potentially increasing memory utilization and triggering memory resource contention. Conversely, an increase in memory utilization or near-saturation will weaken the system's real-time processing capabilities, leading to an accumulation of interrupt response queues and further extending the average interrupt response latency. Conversely, a significant increase in interrupt response latency means a backlog of pending events, which statistically manifests as a cumulative increase in the total number of interrupt events in subsequent cycles, forming a negative cycle. Comprehensive analysis yields the interruption risk index value for each sensor data channel, which can quantitatively assess the real-time processing health and congestion trend of the data channel. This helps to implement proactive controls such as priority scheduling and resource isolation before the system performance deteriorates significantly, ensuring the certainty of the core early warning data path and ultimately improving the overall reliability of the high-altitude remote landslide disaster early warning system.

[0058] In this embodiment, discrete operating data streams are fused into an intuitive risk quantification indicator, enabling early and quantitative perception of extreme operating conditions such as interruption storms. This allows the system to identify specific channels where load pressure is accumulating, rather than just perceiving the overall load. Without this quantitative analysis, the system cannot distinguish between normal load fluctuations and true precursors to blockage risks, and may miss the best time for intervention.

[0059] Real-time analysis of the multidimensional labels of each sensor data channel to mark the early warning results of each sensor data channel.

[0060] Specifically, the multidimensional tags of each sensor data channel are analyzed. The specific analysis process is as follows: the multidimensional tags of each sensor data channel are analyzed, and the static priority value, dynamic confidence value, and resource dependency signature of each sensor data channel encapsulated in the multidimensional tags of each sensor data channel are extracted. The static priority value is compared with a preset highest priority threshold, the dynamic confidence value is compared with a preset dynamic confidence threshold, and the resource dependency signature is compared with a preset hardware resource warning zone list.

[0061] It should be added that the three-step comparison rule constitutes an accurate critical task filter, requiring that the task is not only important and the data reliable, but also that its execution path cannot conflict with critical hardware resources that may be occupied. This design avoids scheduling a task that is destined to be blocked due to resource conflicts in an emergency, and ensures the executability of the selected task. Without such a strict comparison, the marked critical task may be unable to execute in survival mode due to resource lock-in, resulting in control failure.

[0062] Specifically, the warning results of each sensor data channel are marked. The specific process is as follows: when the static priority value of a sensor data channel is equal to the preset highest priority threshold, and the dynamic confidence value is greater than or equal to the preset dynamic confidence threshold, and the intersection of the resource dependency signature and the preset hardware resource warning zone list is an empty set, then the warning task carried by the sensor data channel is determined to be a critical warning task, and the warning result of the sensor data channel is marked as needing priority scheduling; otherwise, the warning result of the sensor data channel is marked as not needing priority scheduling.

[0063] It should be noted that this labeling process transforms abstract 3D labels into explicit and operable scheduling instructions indicating whether or not priority scheduling is required. This provides clear input for subsequent mode adjustment and queue reconstruction, enabling the scheduler to execute complex decision results efficiently. Without this explicit labeling, the scheduler would struggle to process complex 3D information in real time within extreme timeframes, leading to decision delays.

[0064] In this embodiment, this step intelligently classifies massive interruption events based on static configuration and dynamic data quality, accurately identifying key early warning tasks. This ensures that the system's limited emergency resources can be accurately delivered to the data streams that need the most protection. If this marking is not performed, all interruptions will be treated equally. When resources are scarce, key deformation data may be overwhelmed by a large amount of secondary environmental data, causing the early warning to fail.

[0065] Based on the interruption risk index values ​​of each sensor data channel and the early warning results of each sensor data channel, the operation mode of the early warning scheduler for each sensor data channel is adjusted.

[0066] Specifically, the operation mode of the early warning scheduler for each sensor data channel is adjusted. The specific adjustment process is as follows: the interruption risk index value of each sensor data channel is compared with the preset interruption risk index threshold. When the interruption risk index value of a certain sensor data channel is greater than or equal to the interruption risk safety threshold, and the early warning result of the sensor data channel is marked as needing priority scheduling, the operation mode of the early warning scheduler for that sensor data channel is switched from the efficiency priority mode to the survival mode. Otherwise, the operation mode of the early warning scheduler for that sensor data channel continues to operate in the efficiency priority mode.

[0067] It should be added that the characteristic of this adjustment process lies in the triggering conditions and logic: the highest level of survival mode is triggered only when both high risk and critical tasks are detected simultaneously. This prevents overreaction when there is only high risk but no critical tasks, such as when the system self-test generates a large number of interruptions, or when there are critical tasks but the system load is still acceptable. This maintains the system's operating efficiency under non-extreme conditions. If the adjustment logic is only based on a single risk threshold, it will lead to frequent and unnecessary mode switching, introducing additional system overhead and instability.

[0068] Based on the adjusted operation mode of the early warning scheduler for each sensor data channel, the scheduling and execution of each sensor data channel are adjusted, and high-level remote landslide disaster early warning is carried out after the scheduling and execution adjustment.

[0069] Specifically, the scheduling and adjustment of each sensor data channel are carried out. The specific process is as follows: based on the early warning scheduler operation mode of the sensor data channel, the corresponding scheduling adjustment is performed.

[0070] When the early warning scheduler is in efficiency-first mode, the scheduler obtains the comprehensive scheduling score of each sensor data channel based on the static priority value and dynamic confidence value of each sensor data channel through a preset weight calculation formula, and sequentially schedules and executes the interruption event handling tasks of all sensor data channels in descending order of comprehensive scheduling score.

[0071] In this embodiment, under the efficiency-first mode, scheduling is performed through comprehensive scoring, which takes into account both the importance of the task and the real-time quality of the data, and achieves a balanced optimization of system throughput and processing quality, providing high-performance assurance for normal operation.

[0072] It should be added that the weight calculation formula is as follows:

[0073] ;

[0074] In the formula, This represents the comprehensive scheduling score for the i-th sensor data channel. This represents the static priority value of the i-th sensor data channel. This represents the weight coefficient corresponding to the set static priority value. This represents the dynamic confidence level value of the i-th sensor data channel. This represents the weighting coefficient corresponding to the set dynamic confidence level value, where i represents the number of each sensor data channel, i=1, 2, 3, ..., n, and n represents the total number of sensor data channels.

[0075] It should be added that, in this embodiment, the weight coefficients corresponding to the preset static priority values ​​and dynamic confidence values ​​are obtained from the disaster early warning database. These weight coefficients are used to adjust the importance of the static priority values ​​and dynamic confidence values ​​of each sensor data channel in the process of analyzing and obtaining the comprehensive scheduling score.

[0076] When the early warning scheduler switches to survival mode, the scheduler immediately stops and clears the interrupt event handling task queue of the current sensor data channel. At the same time, based on the early warning results of each sensor data channel, it reconstructs an execution queue that only contains sensor data channels whose early warning results are marked as needing priority scheduling. The early warning scheduler only extracts interrupt event handling tasks from this execution queue for scheduling and execution, and performs a suspension operation on interrupt handling requests initiated by sensor data channels whose early warning results are marked as not needing priority scheduling.

[0077] It needs to be explained that the scheduling and execution adjustment in survival mode is a key action to ensure determinism. Clearing non-critical queues and reconstructing queues essentially creates a green channel for critical data flow at the software level, and physically isolates interference sources by suspending non-critical requests. This ensures that the response time of core early warning tasks is no longer affected by the workload of other tasks. This step directly eliminates resource competition between core tasks and non-critical tasks. If this queue reconstruction and suspension operation is not performed, non-critical tasks will still consume scheduling cycles and memory bandwidth, and the response time of core tasks will still not be guaranteed in a storm.

[0078] Specifically, after scheduling and adjustment, a high-level remote landslide disaster early warning is issued. The specific process is as follows: after the early warning scheduler completes the scheduling and adjustment of the data channels of each sensor, it receives the multi-source monitoring data stream of each sensor data channel that has been scheduled, and obtains the high-level remote landslide disaster early warning signal.

[0079] It should be added that the early warning server uses conventional multi-source data fusion early warning algorithms in the field, such as feature-weighted or decision-level fusion algorithms, to fuse the multi-source monitoring data streams from the scheduled sensor data channels to obtain high-level remote landslide disaster early warning signals.

[0080] In this embodiment, the ultimate goal of all the above steps is to ensure that the multi-source data fusion early warning algorithm in this high-altitude remote landslide disaster early warning system can obtain continuous, complete, and timely high-quality input data, providing a reliable and deterministic data supply platform for the upper-level early warning algorithm, thereby fundamentally improving the credibility and timeliness of the entire early warning system. Without the guarantee of a deterministic response in the preceding steps, even the most advanced fusion algorithm cannot generate the correct early warning signal when the data source is interrupted or delayed.

[0081] Please see Figure 2 As shown, the second aspect of the present invention provides a high-altitude remote landslide disaster early warning system based on multi-source data fusion, comprising: a data acquisition module, used to deploy a hardware risk perception unit at the data acquisition front end of the early warning server, to acquire the operating data stream of each sensor data channel through the hardware risk perception unit, and to simultaneously acquire the multi-dimensional tags of each sensor data channel.

[0082] The risk analysis module is used to analyze the interruption risk index value of each sensor data channel based on the operating data stream of each sensor data channel.

[0083] The tag analysis module is used to analyze the multidimensional tags of each sensor data channel in real time and mark the early warning results of each sensor data channel.

[0084] The mode adjustment module is used to adjust the operation mode of the early warning scheduler for each sensor data channel based on the interruption risk index value of each sensor data channel and the early warning results of each sensor data channel.

[0085] The early warning analysis module is used to adjust the scheduling and execution of each sensor data channel based on the adjusted operation mode of the early warning scheduler, and to conduct high-level remote landslide disaster early warning after the scheduling and execution adjustment.

[0086] It should be noted that the high-level remote landslide disaster early warning method and system based on multi-source data fusion also includes a disaster early warning database, which stores preset interruption trigger conditions, tag mapping tables, resource mapping tables, dynamic confidence scores corresponding to each event type code, the maximum baseline value of the total number of interruption events, the maximum baseline value of interruption frequency, the maximum baseline value of memory utilization, the maximum baseline value of interruption response delay, the weight factor corresponding to the total number of interruption events, the weight factor corresponding to interruption frequency, the weight factor corresponding to memory utilization, the weight factor corresponding to the average interruption response delay, the highest priority threshold, the dynamic confidence threshold, the hardware resource warning zone list, the interruption risk indicator threshold, the weight coefficient corresponding to the static priority value, and the weight coefficient corresponding to the dynamic confidence value, obtained through analysis of historical data.

[0087] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0088] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A high-altitude remote landslide disaster early warning method based on multi-source data fusion, characterized in that, include: A hardware risk perception unit is deployed at the data acquisition front end of the early warning server. The hardware risk perception unit collects the operating data stream of each sensor data channel and simultaneously acquires the multi-dimensional tags of each sensor data channel. Based on the operational data stream of each sensor data channel, the interruption risk index value of each sensor data channel is obtained through analysis; Real-time analysis of multidimensional labels for each sensor data channel to mark the early warning results for each sensor data channel; Based on the interruption risk index value of each sensor data channel and the early warning results of each sensor data channel, adjust the operation mode of the early warning scheduler for each sensor data channel. Based on the adjusted operation mode of the early warning scheduler for each sensor data channel, the scheduling and execution of each sensor data channel are adjusted, and high-level remote landslide disaster early warning is carried out after the scheduling and execution adjustment. The specific process of synchronously acquiring multidimensional tags from each sensor data channel is as follows: The hardware risk perception unit continuously monitors the hardware level signals of each sensor data channel. When the hardware level signal of a certain sensor data channel meets the preset interrupt triggering conditions, the hardware risk perception unit generates a corresponding interrupt event. The interrupt event includes the sensor data channel identifier, timestamp, and event type code. While generating the corresponding interrupt event, the hardware risk perception unit queries the tag mapping table preset in the disaster early warning database based on the sensor data channel identifier, extracts the static priority value preset for the sensor data channel, and records it as the static priority value of the sensor data channel. The dynamic confidence value of the sensor data channel is obtained by synchronously matching the event type encoding. Simultaneously, based on the event type code, a preset resource mapping table is retrieved to obtain a list of hardware resources required to process the interrupt event of that type, which serves as the resource dependency signature for the sensor data channel; The static priority value, dynamic confidence value, and resource dependency signature of the sensor data channel are encapsulated into a multi-dimensional label corresponding to the interruption event, and denoted as the multi-dimensional label of the sensor data channel.

2. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 1, characterized in that: The specific process of acquiring the operational data streams of each sensor data channel through the hardware risk perception unit is as follows: The hardware risk perception unit collects the operating data streams of each sensor data channel within a preset regulatory period. The operating data streams of each sensor data channel include the total number of interrupt events, interrupt frequency, memory utilization, and average interrupt response delay of each sensor data channel.

3. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 2, characterized in that: The analysis yielded interruption risk index values ​​for each sensor data channel. The specific analysis process is as follows: Based on the running data stream of each sensor data channel, the interruption risk index value of each sensor data channel is obtained. The interruption risk index value of each sensor data channel represents the quantitative result of the total number of interruption events, interruption frequency, memory utilization rate and average interruption response delay of each sensor data channel, which together quantify the degree of blocking risk of the sensor data channel.

4. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 1, characterized in that: The analysis of the multidimensional labels of each sensor data channel is specifically as follows: Analyze the multidimensional tags of each sensor data channel, extract the static priority value, dynamic confidence value and resource dependency signature of each sensor data channel encapsulated in the multidimensional tags of each sensor data channel, compare the static priority value with the preset highest priority threshold, compare the dynamic confidence value with the preset dynamic confidence threshold, and compare the resource dependency signature with the preset hardware resource warning zone list.

5. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 4, characterized in that: The process of marking the early warning results for each sensor data channel is as follows: When the static priority value of a sensor data channel is equal to the preset highest priority threshold, the dynamic confidence value is greater than or equal to the preset dynamic confidence threshold, and the intersection of the resource dependency signature and the preset hardware resource warning zone list is an empty set, the early warning task carried by the sensor data channel is determined to be a critical early warning task, and the early warning result of the sensor data channel is marked as needing priority scheduling; otherwise, the early warning result of the sensor data channel is marked as not needing priority scheduling.

6. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 1, characterized in that: The specific adjustment process for the early warning scheduler's operating mode, which adjusts the data channels of each sensor, is as follows: The interruption risk index value of each sensor data channel is compared with the preset interruption risk index threshold. When the interruption risk index value of a sensor data channel is greater than or equal to the interruption risk safety threshold, and the early warning result of the sensor data channel is marked as needing priority scheduling, the early warning scheduler operation mode of the sensor data channel is switched from efficiency priority mode to survival mode. Otherwise, the early warning scheduler operation mode of the sensor data channel continues to use efficiency priority mode.

7. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 6, characterized in that: The specific process of scheduling and adjusting the data channels of each sensor is as follows: Based on the operating mode of the early warning scheduler for the sensor data channel, corresponding scheduling adjustments are executed; When the early warning scheduler is in efficiency-first mode, the scheduler obtains the comprehensive scheduling score of each sensor data channel based on the static priority value and dynamic confidence value of each sensor data channel through a preset weight calculation formula, and schedules and executes the interrupt event handling tasks of all sensor data channels in order of comprehensive scheduling score from high to low. When the early warning scheduler switches to survival mode, the scheduler immediately stops and clears the interrupt event handling task queue of the current sensor data channel. At the same time, based on the early warning results of each sensor data channel, it reconstructs an execution queue that only contains sensor data channels whose early warning results are marked as needing priority scheduling. The early warning scheduler only extracts interrupt event handling tasks from this execution queue for scheduling and execution, and performs a suspension operation on interrupt handling requests initiated by sensor data channels whose early warning results are marked as not needing priority scheduling.

8. The high-altitude remote landslide disaster early warning method based on multi-source data fusion according to claim 1, characterized in that: The specific process for issuing high-level remote landslide disaster early warning after scheduling and adjustment is as follows: After the early warning dispatcher completes the scheduling and adjustment of the data channels of each sensor, it receives the multi-source monitoring data streams of each sensor data channel after scheduling, and obtains the high-level remote landslide disaster early warning signal.

9. A high-altitude remote landslide disaster early warning system based on multi-source data fusion, characterized in that: include: The data acquisition module is used to deploy a hardware risk perception unit at the data acquisition front end of the early warning server. The hardware risk perception unit collects the operating data stream of each sensor data channel and simultaneously acquires the multi-dimensional tags of each sensor data channel. The risk analysis module is used to analyze and obtain the interruption risk index value of each sensor data channel based on the operating data stream of each sensor data channel; The tag analysis module is used to analyze the multidimensional tags of each sensor data channel in real time and mark the early warning results of each sensor data channel; The mode adjustment module is used to adjust the operation mode of the early warning scheduler of each sensor data channel based on the interruption risk index value of each sensor data channel and the early warning results of each sensor data channel. The early warning analysis module is used to adjust the scheduling and execution of each sensor data channel based on the adjusted operation mode of the early warning scheduler, and to conduct high-level remote landslide disaster early warning after the scheduling and execution adjustment.

Images