A multi-source information fusion intelligence command integration disposal method and system

Abstract

The present application relates to the technical field of information processing and command and dispatch, and discloses a kind of multi-source information fusion intelligence command disposal method and system.The method is analyzed information flow characteristics by event emergency degree evaluation algorithm, and the emergency degree score is calculated by combining event type rule base and arrival rate deviation degree;The queue length and average waiting time of processing queue are obtained to calculate the queue congestion factor;According to the product relationship of emergency degree and queue congestion factor, the dynamic priority value is calculated, and when the system load increases, the priority gap of high and low emergency degree information flow is widened;Based on the dynamic priority value, the processing queue is reordered;When the queue congestion factor exceeds the threshold, the sampling frequency of low priority information flow is reduced and the processing node is expanded.The present application solves the technical problem that key intelligence is delayed in the multi-source information flow concurrent scenario, and improves the real-time intelligence disposal in the emergency scenario.

Description

Technical Field

[0001] This invention relates to the field of information processing and command and dispatch technology, and more specifically, to an intelligence command and dispatch method and system that integrates multi-source information. Background Technology

[0002] In emergency response, the command center needs to simultaneously receive and process real-time data streams from multiple information sources, such as on-site video surveillance and sensor alarms. These information streams arrive at the processing system concurrently at different rates, placing stringent demands on the timeliness of command decisions.

[0003] Currently, the scheduling and processing of multi-source information streams mainly adopts the first-come-first-served (FFS) scheduling strategy and the fixed-priority scheduling strategy. The FFS scheduling strategy processes information streams sequentially according to their arrival time, treating all information streams as equally important; the fixed-priority scheduling strategy pre-sets a fixed priority based on the type of information source, for example, setting all video streams to the same priority and all text streams to another priority.

[0004] However, the aforementioned existing technologies have the following drawbacks: First, the first-come, first-served strategy cannot distinguish the urgency of different information streams. When multiple information streams arrive concurrently, critical high-urgency intelligence is forced to queue, delaying the commander's perception of the situation and the generation of decision-making. Second, although the fixed-priority strategy can distinguish the importance of different information sources, it cannot perceive changes in the urgency of specific events. For example, the urgency of a live fire monitoring video stream is significantly higher than that of a regular patrol video stream, but the fixed-priority strategy may classify them as the same category of "video stream" and assign them the same priority. Third, traditional scheduling strategies lack awareness of system load status and cannot dynamically adjust processing strategies when resource constraints cause congestion in the processing queue, further exacerbating the problem of insufficient real-time performance.

[0005] Therefore, how to dynamically adjust the processing priority based on the urgency of the event and the system load status under the condition of concurrent arrival of multi-source information streams and limited resources, so as to ensure that critical intelligence is processed first, has become an urgent technical problem to be solved. Summary of the Invention

[0006] This invention provides an intelligence command and control method and system that integrates multi-source information, solving the technical problems in related technologies where critical intelligence is delayed in concurrent multi-source information flow scenarios and traditional scheduling strategies cannot dynamically adjust processing priorities according to the urgency of events and system load status.

[0007] This invention discloses a multi-source information fusion intelligence command and control method, comprising: acquiring real-time data streams from multiple information sources, extracting the stream identifier, data type, and arrival rate of each data stream, and generating an information stream feature set; analyzing the information stream feature set using an event urgency assessment algorithm, accessing a predefined event urgency rule base to obtain a baseline urgency value corresponding to each data type, calculating the deviation of the current arrival rate from the historical average arrival rate, calculating the urgency score of each information stream based on the baseline urgency value and the deviation, and generating an urgency vector; acquiring the queue length and average waiting time of the processing queue, calculating a queue congestion factor based on the weighted sum of the ratio of the queue length to a queue length threshold and the ratio of the average waiting time to a waiting time threshold, and generating system load status parameters; and combining the urgency vector with the system load status parameters. The data is input into the priority calculation function, which calculates the dynamic priority value of each information flow based on the product of the urgency score and the queue congestion factor. When the system load increases, the priority gap between high-urgency and low-urgency information flows is widened. Based on the dynamic priority value, the processing queue is reordered using a priority queue scheduling algorithm, placing information flows with higher priority values ​​at the front of the queue to generate a scheduled processing queue. The queue congestion factor is monitored in real time. When the queue congestion factor exceeds the congestion threshold, information flows with priority values ​​lower than the priority threshold are filtered to form a low-priority information flow subset. The sampling frequency of each information flow in the low-priority information flow subset is reduced, and a processing node expansion instruction is triggered. The information flows in the scheduled processing queue are sequentially extracted and input into the intelligence processing module for processing, and the processing results are output to the command and decision-making system.

[0008] Furthermore, the event urgency assessment algorithm includes the following sub-steps: accessing a predefined event urgency rule base, which stores the mapping relationship between different event types and urgency; matching the corresponding rule in the rule base according to the data type of the information flow to obtain a baseline urgency value; obtaining the arrival rate and historical average arrival rate of the information flow, and calculating the rate deviation factor; calculating the urgency score based on the product of the baseline urgency value and an adjustment factor containing the rate deviation factor; and organizing the urgency scores of all information flows into an urgency vector.

[0009] Furthermore, the event urgency assessment algorithm also includes: extracting content feature parameters from the information stream; for video streams, extracting the detection confidence of key objects such as flames, smoke, and crowds in video frames; for text streams, extracting the frequency of occurrence of urgent keywords in the text, and generating a content feature score; when calculating the urgency score, multiplying the baseline urgency value, a first adjustment factor including the rate deviation factor, and a second adjustment factor including the content feature score to obtain the corrected urgency score.

[0010] Furthermore, the generated system load status parameters also include: obtaining processor utilization and memory utilization; when calculating the queue congestion factor, the ratio of queue length to queue length threshold, the ratio of average waiting time to waiting time threshold, processor utilization, and memory utilization are weighted and summed, and the sum of each weight coefficient is equal to 1.

[0011] Furthermore, the calculation of the dynamic priority value of each information flow includes: multiplying the urgency score by a load adjustment factor that includes the queue congestion factor to obtain the dynamic priority value. When the queue congestion factor increases, the dynamic priority value of the high urgency information flow is amplified by a greater margin than the dynamic priority value of the low urgency information flow.

[0012] Furthermore, the calculation of the dynamic priority value of each information flow also includes: obtaining the waiting time of each information flow; adding the product of the urgency score and the load adjustment factor to the product of the waiting time and the time compensation coefficient to obtain the corrected dynamic priority value, so that the priority of low-priority information flows increases with the increase of waiting time.

[0013] Furthermore, it also includes: allocating processing resource quotas according to the dynamic priority value of each information flow, determining the resource quota allocated to each information flow by the proportion of the priority value of each information flow to the sum of the priority values ​​of all information flows, and generating a resource allocation scheme; the intelligence processing module allocates processor time slices, memory space and network bandwidth to each information flow according to the resource allocation scheme.

[0014] Furthermore, the allocation of processing resource quotas based on the dynamic priority value of each information flow includes: reserving a minimum resource quota for all information flows, allocating the remaining resources according to the proportion of the priority value of each information flow to the sum of the priority values ​​of all information flows, and the sum of the minimum resource quota and the proportionally allocated resource quota is used as the final resource quota for that information flow.

[0015] Furthermore, reducing the sampling frequency of each information stream in the low-priority information stream subset includes: reducing the current sampling frequency according to the sampling frequency attenuation factor; reducing the sampling rate of video frames for video streams; increasing the time interval of text batches for text streams; and reducing the data point acquisition density for sensor data streams.

[0016] Furthermore, it also includes: when the queue congestion factor drops below the recovery threshold, acquiring a subset of the information flow with the reduced sampling frequency, gradually restoring its sampling frequency to the original value, and obtaining a new sampling frequency by adding the product of the current reduced sampling frequency and the difference between the sampling frequency and the original sampling frequency to the recovery rate coefficient, thereby achieving gradual recovery.

[0017] Furthermore, it also includes: periodically repeating the steps of acquiring information flow characteristics, calculating urgency vector, calculating queue congestion factor, calculating dynamic priority value, scheduling processing queue, and monitoring queue congestion factor, and dynamically updating the scheduling strategy based on real-time changes in information flow characteristics and system load.

[0018] This invention discloses a multi-source information fusion intelligence command and control system, comprising: a data receiving module for acquiring real-time data streams from multiple information sources and extracting stream identifiers, data types, and arrival rates to generate an information stream feature set; an urgency assessment module for analyzing the information stream feature set, accessing an event urgency rule base, and calculating the urgency score of each information stream to generate an urgency vector; a load monitoring module for acquiring the queue length and average waiting time of the processing queue, and calculating the queue congestion factor to generate system load status parameters; a priority calculation module for calculating the dynamic priority value of each information stream based on the urgency vector and the system load status parameters; a scheduling module for reordering the processing queue based on the dynamic priority value to generate a scheduled processing queue; an adaptive scheduling module for real-time monitoring of the queue congestion factor, reducing the sampling frequency of low-priority information streams and triggering processing node expansion instructions when the congestion factor exceeds a threshold; and an intelligence processing module for sequentially processing the information streams in the scheduled processing queue and outputting the processing results to the command and decision-making system.

[0019] This invention utilizes an event urgency assessment algorithm, combined with an event type rule base, arrival rate deviation, and content feature parameters, to calculate urgency scores, achieving quantitative identification of the urgency levels of different information flows and overcoming the limitation of traditional first-come-first-served strategies in distinguishing information urgency. Through load-aware dynamic priority calculation based on queue congestion factors, urgency is combined with system load status. When system load increases, the priority gap between high and low urgency information flows is automatically widened, overcoming the limitation of fixed priority strategies in not being aware of system load status. A priority queue scheduling algorithm reorders processing queues based on dynamic priority values, placing high-priority information flows at the front of the queue to ensure critical intelligence enters the processing flow first. A congestion-triggered adaptive scheduling mechanism proactively reduces the sampling frequency of low-priority information flows and triggers processing node expansion when the queue congestion factor exceeds a threshold, achieving dynamic resource allocation. Finally, a resource allocation algorithm allocates differentiated processing resource quotas to each information flow based on priority values, granting high-priority information flows more processor time and memory space, accelerating their processing. This invention solves the technical problems of delayed processing of critical intelligence in scenarios with concurrent multi-source information flows and the inability of traditional scheduling strategies to dynamically adjust processing priorities based on the urgency of events and system load status. It achieves the technical effect of ensuring priority processing of critical intelligence under conditions of limited resources and queue congestion, and improving the real-time performance of intelligence handling in emergency scenarios. Attached Figure Description

[0020] Figure 1 This is a flowchart of the information flow feature extraction, urgency assessment, load monitoring, priority calculation, and queue scheduling process of the present invention;

[0021] Figure 2 This is a flowchart of the sub-steps of the event urgency assessment algorithm of the present invention;

[0022] Figure 3 This is a flowchart of the resource allocation, adaptive scheduling, and intelligence processing of the present invention;

[0023] Figure 4 This is a flowchart of the periodic scheduling and update process of the present invention. Detailed Implementation

[0024] In emergency response, command centers need to simultaneously receive and process real-time data streams from multiple information sources, such as on-site video surveillance and sensor alarms. These data streams arrive at the processing system concurrently at different rates. Under limited resource conditions, traditional first-come, first-served (FFS) scheduling strategies treat all data streams as equally important and process them sequentially according to their arrival order. While fixed-priority strategies can distinguish the importance of different data sources, they cannot perceive changes in the urgency of specific events. For example, a video stream from an on-site fire monitoring station may have a significantly higher urgency than a regular patrol video stream, but a fixed-priority strategy might classify them as the same type of "video stream" and assign them the same priority. When multiple data streams arrive concurrently, causing congestion in the processing queue, critical, high-urgency intelligence is forced to wait in line, delaying the commander's perception of the situation and the generation of response decisions, thus missing the optimal response opportunity. Furthermore, traditional scheduling strategies lack awareness of system load status and cannot dynamically adjust processing strategies when resources are scarce, further exacerbating the problem of insufficient real-time performance.

[0025] It should be understood that the method described in this embodiment is executed in a computer system, which includes hardware and software components such as a data receiving module, a processing queue, a computing resource pool, and a scheduling controller, and is capable of receiving multi-source information streams and performing real-time scheduling processing.

[0026] The method of this embodiment includes the following steps;

[0027] Step 100: Obtain real-time data streams from multiple information sources, extract metadata features of each data stream including stream identifier, data type, and arrival rate, and generate an information stream feature set;

[0028] Real-time data streams are acquired from multiple information sources, including on-site video surveillance systems, sensor networks, and network monitoring platform systems. For each data stream, its metadata features are extracted: a stream identifier uniquely identifies the information source; data types include category tags such as "video stream," "text stream," and "sensor data stream"; and the arrival rate represents the amount of data arriving per unit time. The extracted features are then organized into an information stream feature set. ,in Indicates the first The feature vector of an information flow For flow identification, For data types, The arrival rate.

[0029] It should be noted that the above-mentioned arrival rate is calculated within a time window. The number of data packets or bytes of data arriving in the internal statistics stream is calculated using the following formula: ,in For the amount of data to be statistically analyzed, This represents the length of the time window.

[0030] Furthermore, the length of the time window The value ranges from 1 second to 10 seconds, with the specific value determined based on the data type of the information stream: for high-frequency data streams such as video streams and sensor data streams, The value is set to 1 to 3 seconds to ensure timely capture of rate changes; for low-frequency data streams such as text streams and alarm messages, The value is set between 5 and 10 seconds to avoid misjudgments due to instantaneous fluctuations. The specified time window length... The system is pre-configured based on the historical arrival characteristics of each information source during system initialization and remains fixed during operation.

[0031] Step 200: Analyze the information flow feature set using the event urgency assessment algorithm, combine it with the predefined event urgency rule base, calculate the urgency score of each information flow, and generate an urgency vector;

[0032] Information flow feature set The information is input into the event urgency assessment algorithm, which includes the following sub-steps:

[0033] Step 201: Access the predefined event urgency rule base, which stores the mapping relationship between different event types and urgency levels. For example, "fire monitoring" corresponds to high urgency and "routine patrol" corresponds to low urgency.

[0034] Step 202: Based on the data type of the information flow Match the corresponding rule in the rule base to obtain the baseline urgency value. ;

[0035] Step 203: Obtain the arrival rate of the information stream and historical average arrival rate Calculate the rate deviation factor ;

[0036] Step 204: Calculate the urgency score based on the baseline urgency value and the rate deviation factor. The calculation formula is as follows: ,in This is the rate influence coefficient;

[0037] Step 205: Organize the urgency scores of all information streams into an urgency vector. ,in Indicates the first The urgency score of each information stream.

[0038] The aforementioned event urgency assessment algorithm can further incorporate a content feature extraction step between steps 203 and 204: for video streams, extract the detection confidence of key objects such as flames, smoke, and crowds in the video frames; for text streams, extract the frequency of occurrence of keywords such as "emergency," "help," and "casualties" in the text to generate content feature scores. In step 204, content features are used as adjustment factors, and the corrected calculation formula is as follows: ,in This represents the content impact coefficient.

[0039] Furthermore, the rate influence coefficient The value ranges from 0.1 to 0.5, and the specific value is determined based on the rate fluctuation characteristics of the information source: for information sources with large rate fluctuations (such as video surveillance). Use a smaller value, 0.1 to 0.2, to reduce the impact of rate deviation on urgency; for information sources with small rate fluctuations but strong indicativeness of sudden rate increases (such as sensor alarms). The influence of amplification rate deviation is considered by taking a larger value, 0.3 to 0.5. This is the reference urgency value. The value ranges from 1 to 10, where 1 represents the lowest urgency and 10 represents the highest urgency. This value is predefined by the event urgency rule base based on the data type and event type of the information flow.

[0040] Furthermore, the content feature score The value ranges from 0 to 1, where 0 indicates no urgent content was detected and 1 indicates highly urgent content was detected. The content influence coefficient... The value ranges from 0.2 to 0.8, with the specific value determined based on the reliability of the content features: for video target detection with high accuracy, Take a larger value, 0.5 to 0.8; for text keyword matching with relatively low accuracy, Take the smaller value, 0.2 to 0.4.

[0041] Step 300: Obtain the current processing queue status data, including queue length and average waiting time, calculate the queue congestion factor, and generate system load status parameters;

[0042] Access the status monitoring interface of the processing queue to obtain the number of information streams waiting to be processed in the current queue as the queue length. Get the waiting time of all waiting information streams in the queue and calculate the average as the average waiting time. Using the queue congestion factor calculation formula ,in This is the queue length threshold. As the waiting time threshold, and The weighting coefficients are satisfied. Generate system load status parameters. .

[0043] Furthermore, the weighting coefficients and The value of is determined based on the performance bottleneck characteristics of the system: for systems where queue length has a more significant impact on performance, set . and For systems where latency is a critical performance indicator, set... and For systems where both are equally important, set .

[0044] The queue length threshold The value is determined based on the system design capacity and is set as the maximum allowed length of the processing queue, typically ranging from 50 to 200 information streams.

[0045] The waiting time threshold The value is determined based on the timeliness requirements of the business. For high-timeliness scenarios, the value is 1 to 3 seconds, and for regular scenarios, the value is 5 to 10 seconds.

[0046] It should be noted that the range of values ​​for the above-mentioned queue congestion factor is as follows: ,when This indicates that the system load is normal. This indicates that the system is experiencing congestion.

[0047] In this embodiment, to more comprehensively reflect the system load status, the system load status parameters also include computing resource utilization parameters. This involves obtaining the processor CPU utilization rate. and memory utilization The corrected formula for calculating the queue congestion factor is as follows: ,in and The four weighting coefficients are the weighting factors for resource utilization, and they satisfy the following conditions: .

[0048] Furthermore, the specific values ​​of the four weighting coefficients are determined based on the type of resource bottleneck in the system: for compute-intensive systems, the values ​​are set as follows: For memory-intensive systems, set For I / O intensive systems, set For resource-balanced systems, set The processor utilization and memory utilization The values ​​range from 0 to 1.

[0049] Step 400: Input the urgency vector and system load status parameters into the priority calculation function to generate dynamic priority values ​​for each information flow;

[0050] The urgency vector generated in step 200 and the system load status parameters generated in step 300 The input is fed into the priority calculation function. This function comprehensively considers the urgency of the event and the system load status. When the system load is normal, the priority is mainly determined by the urgency. When the system load is too high, the priority gap between high-urgency and low-urgency information flows is further widened to ensure that critical intelligence is processed first. The priority calculation function is as follows: ,in For the first The dynamic priority value of each information stream. For load sensitivity coefficient, This represents the queue congestion factor. It also generates a dynamic priority vector. .

[0051] Furthermore, the load sensitivity coefficient The value ranges from 0.5 to 2.0, and the specific value is determined based on the system's response requirements to load changes: when the system needs to respond quickly to load changes and strongly distinguish between high and low priorities, Choose a larger value, 1.5 to 2.0; when system load fluctuations are small or a gentle adjustment of priority is required, Take a smaller value between 0.5 and 1.0. The load sensitivity coefficient. The system is pre-configured based on historical load fluctuation data and business scenario characteristics during system initialization.

[0052] It should be noted that the design principle of the above priority calculation function is: when When smaller, With a value close to 1, the priority of each information flow is mainly determined by its urgency. Decision; when When the priority of high-urgency information flow increases, This is further amplified, increasing the priority gap with low-urgency information flows, thus enabling load-aware priority adjustment.

[0053] In this embodiment, to avoid the "starvation" phenomenon caused by low-priority information streams not being processed for extended periods, a waiting time compensation mechanism is introduced into the priority calculation function. This involves obtaining the elapsed waiting time for each information stream. The revised priority calculation formula is as follows: ,in This is the time compensation coefficient, which varies with the waiting time. As the number of processes increases, the priority of information flows gradually rises, ensuring that low-priority information flows are eventually processed.

[0054] Step 500: Based on the dynamic priority value, the processing queue is reordered using the priority queue scheduling algorithm to generate the scheduled processing queue;

[0055] Dynamic priority vector The input is fed into a priority queue scheduling algorithm. This algorithm takes each information stream and its corresponding priority value as input and outputs a sorted processing queue. The algorithm sorts the information streams in the processing queue according to their priority values ​​from highest to lowest, with higher priority streams appearing earlier in the queue. For newly arriving information streams, they are inserted into the appropriate position in the queue based on their priority value. This generates the scheduled processing queue. ,in Indicates the number of elements in the queue. Information flow at each location, satisfying .

[0056] The aforementioned priority queue scheduling algorithm is implemented using a heap data structure, which stores the information stream and its priority value in a max heap, with the top element of the heap being the information stream with the highest current priority.

[0057] Step 600: Based on the priority value of each information stream, calculate the processing resource quota for each stream using a resource allocation algorithm, and generate a resource allocation scheme;

[0058] Obtain the total available computing resources of the system This includes the number of processor cores, memory capacity, etc. A resource allocation algorithm is used to allocate processing resource quotas based on the priority value of each information flow. The input of this algorithm is the total amount of available system resources. and priority values ​​of each information stream The output is the resource quota for each information flow. The formula for calculating the resource quota is: ,in For the first Resource quotas are allocated to each information flow. A resource allocation scheme is generated. .

[0059] The aforementioned resource allocation algorithm distributes resource quotas according to priority ratios, ensuring that high-priority information flows receive more processing resources. Resource quotas It can be represented as the specific values ​​of resources such as processor time slices, memory space, and network bandwidth allocated to the information stream.

[0060] In this embodiment of the application, in order to ensure the basic processing needs of low-priority information streams, the resource allocation algorithm sets a minimum resource quota threshold. The revised resource quota calculation method is as follows: First, reserve the minimum quota for all information flows. Remaining resources Resource allocation is based on priority ratio, and the final resource quota is: .

[0061] Furthermore, the minimum resource quota threshold The calculation method is as follows ,in To reserve a proportional coefficient, This represents the total number of current information flows. The reserved ratio coefficient... The value ranges from 0.1 to 0.3, and this coefficient is determined based on the minimum processing requirements of low-priority information streams: when the minimum processing requirements of an information stream are high, Take a larger value, 0.2 to 0.3; when the minimum processing requirement is low, Take the smaller value between 0.1 and 0.15. The minimum resource quota. Constraints must be met That is, the total reserved resources must be less than the total available resources of the system.

[0062] Step 700: Monitor the queue congestion factor in real time. When the congestion factor exceeds the congestion threshold, obtain a subset of low-priority information streams, reduce their sampling frequency, and trigger the processing node expansion instruction.

[0063] Establish a real-time monitoring mechanism for queue congestion factors, and periodically execute step 300 to calculate the current congestion factor. . judge Does it exceed the preset congestion threshold? .when When this happens, select items from the current processing queue whose priority values ​​are lower than the priority threshold. Information flow, generating a low-priority information flow subset. For subsets For each information stream in the process, reduce its sampling frequency, that is, reduce the amount of data read from that information stream per unit time. The sampling frequency adjustment formula is: ,in This is the sampling frequency attenuation factor. Simultaneously, it generates a processing node expansion instruction, which triggers the distributed processing cluster to start new processing nodes or wake up dormant nodes, increasing the system's processing capacity.

[0064] The congestion threshold The value ranges from 1.0 to 1.5. When the queue congestion factor exceeds this threshold, a sampling frequency reduction mechanism is triggered. The priority threshold... The calculation method is as follows ,in This is the average priority of all current information streams. This is a threshold coefficient, ranging from 0.5 to 0.8. Information streams with priority values ​​lower than this threshold are classified into a low-priority subset.

[0065] It should be noted that the above-mentioned methods for reducing the sampling frequency include: for video streams, reducing the sampling rate of video frames, for example, from 30 frames per second to 10 frames per second; for text streams, increasing the time interval between text batches; and for sensor data streams, reducing the data point acquisition density.

[0066] In this embodiment, to avoid system instability caused by frequent node expansion triggers, a cooldown time mechanism is also set up for processing node expansion command triggers. The timestamp of the last triggered expansion command is recorded. The current timestamp is Only when and New extended instructions are only triggered at certain times, among which This is the cooling time threshold.

[0067] Furthermore, the cooling time threshold The value ranges from 30 seconds to 120 seconds, with the specific value determined based on the startup time of the processing node and the system stabilization time: when the processing node starts up quickly, Use the smaller value, 30 to 60 seconds; when the processing node starts up slowly or requires a longer stabilization time. Choose a larger value, between 90 and 120 seconds. This setting ensures that newly started processing nodes have enough time to take effect, avoiding repeated triggering of extended instructions within a short period.

[0068] In this embodiment of the application, when the queue congestion factor Drop to recovery threshold (in When the sampling frequency falls below a certain threshold, the sampling frequency recovery mechanism is triggered. A subset of the information stream whose sampling frequency was previously reduced is retrieved, and its sampling frequency is gradually restored to its original value. The restoration formula is as follows: ,in The recovery rate coefficient is used to avoid sudden changes in system load through gradual recovery.

[0069] Furthermore, the recovery threshold The value is the congestion threshold. 0.6 to 0.8 times, that is ,in The value ranges from 0.6 to 0.8. This setting establishes a buffer between the congestion threshold and the recovery threshold to prevent the system from frequently switching sampling frequency adjustment strategies near critical states. The recovery rate coefficient... The value ranges from 0.2 to 0.3, and this coefficient controls the proportion of the gap that is recovered in each update. A value of 0.2 indicates that the frequency difference is recovered 20% each time. A value of 0.3 indicates a 30% frequency difference in recovery time. A smaller value... This value makes the recovery process smoother and avoids sudden increases in system load.

[0070] Step 800: Extract the information streams from the scheduled processing queue in sequence, input them into the intelligence processing module for processing, and output the processing results to the command and decision-making system;

[0071] The processed queue after scheduling generated in step 500 In the process, information streams are retrieved sequentially according to the queue order. These retrieved information streams are then input into the intelligence processing module, which performs corresponding processing operations based on the data type of the information stream: target detection and scene recognition for video streams, key information extraction and sentiment analysis for text streams, and anomaly detection and trend analysis for sensor data streams. After processing, structured intelligence results are generated, including information such as event type, location, time, and key parameters. The intelligence results are then output to the command and decision-making system for commanders to conduct situational assessments and make appropriate decisions.

[0072] It should be noted that the processing procedure of the aforementioned intelligence processing module is based on the resource quota allocated in step 600. Execution. High-priority information flows receive more resource quotas and enjoy more processor time and memory space, resulting in faster processing speeds.

[0073] In this embodiment of the application, the following steps are also included:

[0074] Step 900: Periodically repeat steps 100 to 700, and dynamically update the scheduling strategy based on the real-time changes in information flow characteristics and system load.

[0075] Establish a periodic scheduling and update mechanism at fixed time intervals. Steps 100 to 700 are repeated periodically. Within each period, the information flow feature set is reacquired, the urgency vector is recalculated, the system load status is reassessed, dynamic priority values ​​are regenerated, processing queues are rescheduled, resource quotas are reallocated, and queue congestion is re-detected. Through periodic updates, the scheduling strategy adapts to real-time changes in information flow characteristics and system load, dynamically adjusting the processing priority and resource allocation for each information flow.

[0076] It should be noted that the time interval for the above periodic updates It is determined based on the timeliness requirements of the emergency. For high-timeliness events, Set to a shorter time, such as 1 to 5 seconds; for regular events, It can be set to a longer time, such as 10 to 30 seconds.

[0077] This implementation uses an event urgency assessment algorithm (step 200) to extract the data type and arrival rate characteristics of information flows, and calculates urgency scores based on an event urgency rule base. This allows the system to quantify the urgency of different information flows, rather than treating all information flows as equally important. A priority calculation function (step 400) combines urgency with system load status to generate dynamic priority values. When the system load is normal, priority is primarily determined by urgency. When the system load is too high, the priority gap between high and low urgency information flows is further widened, ensuring that critical intelligence receives more significant priority processing when resources are scarce. A priority queue scheduling algorithm (step 500) reorders the processing queues according to the dynamic priority values, placing high-priority information flows at the front of the queue to ensure that critical intelligence enters the processing flow first. These techniques overcome the inability of traditional first-come, first-served strategies to distinguish the urgency of information, solving the problem of delayed processing of critical intelligence.

[0078] Furthermore, this implementation uses a real-time monitoring mechanism for queue congestion factors (steps 300 and 700) to quantify the system load status based on queue length and average waiting time. When the congestion factor exceeds the congestion threshold, an adaptive scheduling strategy is proactively triggered: the sampling frequency of low-priority information flows is reduced to decrease their data volume, releasing processing resources for high-priority information flows; simultaneously, processing node expansion instructions are triggered to increase system processing capacity. Through a resource allocation algorithm (step 600), differentiated processing resource quotas are allocated to each information flow based on its priority value, with high-priority information flows receiving more processor time and memory space, accelerating their processing. These techniques overcome the limitations of traditional fixed scheduling strategies, which cannot perceive system load or dynamically adjust processing strategies, thus solving the problem of delayed critical intelligence under limited resource conditions.

[0079] In summary, this implementation method, through a combination of techniques including emergency quantification assessment, load-aware dynamic priority calculation, congestion-triggered adaptive scheduling, and differentiated resource quota allocation, enables the multi-source information flow scheduling system to identify key intelligence, dynamically adjust priorities, and adaptively allocate resources. This solves the technical problem of delayed processing of key intelligence and missed optimal handling opportunities in emergency scenarios.

Claims

1. A method for intelligence command and control based on multi-source information fusion, characterized in that, Includes the following steps: Acquire real-time data streams from multiple information sources, extract the stream identifier, data type, and arrival rate of each data stream, and generate an information stream feature set; The information flow feature set is analyzed using an event urgency assessment algorithm. A predefined event urgency rule base is accessed to obtain the baseline urgency value corresponding to each data type. The deviation of the current arrival rate from the historical average arrival rate is calculated. Based on the baseline urgency value and the deviation, the urgency score of each information flow is calculated, and an urgency vector is generated. Obtain the queue length and average waiting time of the processing queue, calculate the queue congestion factor based on the weighted sum of the ratio of the queue length to the queue length threshold and the ratio of the average waiting time to the waiting time threshold, and generate system load status parameters; The urgency vector and the system load status parameters are input into the priority calculation function. The dynamic priority value of each information flow is calculated based on the product relationship between the urgency score and the queue congestion factor. When the system load increases, the priority gap between high-urgency information flows and low-urgency information flows is widened. Based on the dynamic priority value, the processing queue is reordered using a priority queue scheduling algorithm, and the information stream with higher priority value is moved to the front of the queue to generate a scheduled processing queue. The queue congestion factor is monitored in real time. When the queue congestion factor exceeds the congestion threshold, information streams with priority values ​​lower than the priority threshold are filtered to form a low-priority information stream subset. The sampling frequency of each information stream in the low-priority information stream subset is reduced, and the processing node expansion instruction is triggered. The information streams in the scheduled processing queue are sequentially extracted and input into the intelligence processing module for processing, and the processing results are output to the command and decision-making system.

2. The method according to claim 1, characterized in that, The event urgency assessment algorithm includes the following sub-steps: Access a predefined event urgency rule base, which stores the mapping relationship between different event types and urgency levels; Based on the data type of the information flow, match the corresponding rule in the rule base to obtain the baseline urgency value; Obtain the arrival rate and historical average arrival rate of the information stream, and calculate the rate deviation factor; The urgency score is calculated based on the product of the baseline urgency value and an adjustment factor that includes the rate deviation factor. Organize the urgency scores of all information streams into an urgency vector.

3. The method according to claim 2, characterized in that, The event urgency assessment algorithm also includes: Extract content feature parameters from the information stream. For video streams, extract the detection confidence of key objects such as flames, smoke, and crowds in video frames. For text streams, extract the frequency of occurrence of urgent keywords in the text and generate content feature scores. When calculating the urgency score, the baseline urgency value, a first adjustment factor including the rate deviation factor, and a second adjustment factor including the content feature score are multiplied together to obtain the corrected urgency score.

4. The method according to claim 1, characterized in that, The generated system load status parameters also include: Get processor utilization and memory utilization; When calculating the queue congestion factor, the ratio of the queue length to the queue length threshold, the ratio of the average waiting time to the waiting time threshold, the processor utilization, and the memory utilization are weighted and summed, and the sum of each weight coefficient is equal to 1.

5. The method according to claim 1, characterized in that, The calculation of the dynamic priority value of each information stream includes: The urgency score is multiplied by the load adjustment factor, which includes the queue congestion factor, to obtain the dynamic priority value. When the queue congestion factor increases, the dynamic priority value of the high urgency information flow is amplified by a greater margin than that of the low urgency information flow.

6. The method according to claim 5, characterized in that, The calculation of the dynamic priority value of each information stream also includes: Get the waiting time for each information stream; The product of the urgency score and the load adjustment factor is added to the product of the waiting time and the time compensation coefficient to obtain the corrected dynamic priority value, so that the priority of low-priority information flow increases with the increase of waiting time.

7. The method according to claim 1, characterized in that, Also includes: Resource allocation is performed based on the dynamic priority value of each information flow. The resource allocation for each information flow is determined by the proportion of its priority value to the sum of the priority values ​​of all information flows, and a resource allocation scheme is generated. The intelligence processing module allocates processor time slices, memory space, and network bandwidth to each information stream according to the resource allocation scheme.

8. The method according to claim 7, characterized in that, The allocation of processing resource quotas based on the dynamic priority values ​​of each information flow includes: A minimum resource quota is reserved for all information flows. The remaining resources are allocated according to the proportion of the priority value of each information flow to the sum of the priority values ​​of all information flows. The sum of the minimum resource quota and the proportionally allocated resource quota is the final resource quota for that information flow.

9. The method according to claim 1, characterized in that, The reduction of the sampling frequency of each information stream in the low-priority information stream subset includes: The current sampling frequency is reduced based on the sampling frequency attenuation factor. For video streams, the sampling rate of video frames is reduced; for text streams, the time interval between text batches is increased; and for sensor data streams, the data point acquisition density is reduced.

10. A multi-source information fusion intelligence command and control system, used to execute the method described in any one of claims 1 to 9, characterized in that, include: The data receiving module is used to acquire real-time data streams from multiple information sources and extract stream identifiers, data types, and arrival rates to generate information stream feature sets. The urgency assessment module is used to analyze the information flow feature set, access the event urgency rule base, calculate the urgency score of each information flow, and generate an urgency vector. The load monitoring module is used to obtain the queue length and average waiting time of the processing queue, calculate the queue congestion factor, and generate system load status parameters. The priority calculation module is used to calculate the dynamic priority value of each information flow based on the urgency vector and the system load status parameters. The scheduling module is used to reorder the processing queues based on the dynamic priority values ​​to generate a scheduled processing queue. An adaptive scheduling module is used to monitor the queue congestion factor in real time. When the congestion factor exceeds the congestion threshold, the sampling frequency of low-priority information streams is reduced and a processing node expansion instruction is triggered. The intelligence processing module is used to process the information flow in the scheduled processing queue in sequence and output the processing results to the command and decision-making system.

Images