Multi-source alarm processing method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a multi-source alarm processing method and device, electronic equipment and a storage medium, and relates to the technical field of computers. Obtaining a to-be-processed alarm, and matching and determining an early warning classification label of the to-be-processed alarm from a pre-constructed knowledge graph according to the entity information of the to-be-processed alarm; and inputting the alarm information of the to-be-processed alarm into a pre-trained multi-fusion model to determine an early warning classification label under the condition that the matching of the early warning classification label according to the entity information of the to-be-processed alarm fails. By the adoption of the technical scheme, the difficulty brought to follow-up study and judgment due to the fact that alarms come from different information collecting devices and report different original alarm information is avoided, the workload and the error probability of manual alarm checking, analysis, classification and integration are reduced, and the working efficiency is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a method, device, electronic device and storage medium for multi-source alarm processing. Background Art

[0002] Monitoring the same target through multiple different information collection devices is a commonly used technical means in this field.

[0003] For example, monitoring the use of cultivated land is an important means to implement the cultivated land protection system and ensure food security. In related technologies, cultivated land can be monitored through various information collection devices such as camera equipment (and computer equipment connected to camera equipment), satellites, and multi-spectral remote sensing equipment. When various information collection devices detect abnormal conditions, they will each report corresponding alarm information. Since the alarm information reported by different information collection devices is different and the reporting time is also different, different alarms may be reported repeatedly for the same piece of cultivated land, which is not conducive to the staff's research and judgment.

[0004] In the related art, in order to solve the above problems, manual inspection of alarms, analysis, classification and integration are usually adopted, which is inefficient and prone to errors. Summary of the invention

[0005] The purpose of the embodiments of the present disclosure is to provide a method, device, electronic device and storage medium for multi-source alarm processing.

[0006] To solve the above technical problems, the embodiments of the present disclosure are implemented through the following aspects.

[0007] According to a first aspect of an embodiment of the present disclosure, a method for multi-source alarm processing is provided, the method comprising: obtaining an alarm to be processed, the alarm to be processed being reported by any one of a plurality of different information collection devices; matching and determining an early warning classification label of the alarm to be processed from a pre-constructed knowledge graph according to entity information of the alarm to be processed; in the event that matching the early warning classification label according to the entity information of the alarm to be processed fails, inputting the alarm information of the alarm to be processed into a pre-trained multi-fusion model to determine the early warning classification label, the multi-fusion model comprising a first early warning classification model and a second early warning classification model; wherein the matching of the early warning classification label according to the entity information of the alarm to be processed fails. In the case of signing failure, the alarm information of the alarm to be processed is input into a pre-trained multi-fusion model to determine the warning classification label, including: inputting the alarm information of the alarm to be processed into the first warning classification model, and obtaining the first probabilities corresponding to the multiple warning classification labels output by the first warning classification model; inputting the alarm information of the alarm to be processed into the second warning classification model, and obtaining the second probabilities corresponding to the multiple warning classification labels output by the second warning classification model; determining the weighted probabilities corresponding to the multiple warning classification labels according to the preset weight coefficient, the first probability and the second probability; and taking the warning classification label with the highest weighted probability as the warning classification label of the alarm to be processed.

[0008] According to a second aspect of an embodiment of the present disclosure, a device for multi-source alarm processing is provided, the device comprising: an acquisition module, used to acquire an alarm to be processed, the alarm to be processed is reported by any information collection device among a plurality of different information collection devices; a determination module, used to match and determine the warning classification label of the alarm to be processed from a pre-constructed knowledge graph according to the entity information of the alarm to be processed; the determination module is further used to input the alarm information of the alarm to be processed into a pre-trained multi-fusion model to determine the warning classification label when the matching of the warning classification label according to the entity information of the alarm to be processed fails, the multi-fusion model comprising a first warning classification model and a second warning classification model; Wherein, in the case where the matching of the warning classification label according to the entity information of the to-be-processed alarm fails, inputting the alarm information of the to-be-processed alarm into a pre-trained multi-fusion model to determine the warning classification label includes: Inputting the alarm information of the to-be-processed alarm into the first early warning classification model to obtain first probabilities corresponding to a plurality of early warning classification labels output by the first early warning classification model; Inputting the alarm information of the to-be-processed alarm into the second early warning classification model to obtain second probabilities corresponding to a plurality of early warning classification labels output by the second early warning classification model; Determine the weighted probabilities corresponding to the plurality of warning classification labels respectively according to a preset weight coefficient, the first probability and the second probability; The warning classification label with the highest weighted probability is used as the warning classification label of the alarm to be processed.

[0009] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the multi-source alarm processing method described in the first aspect.

[0010] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for multi-source alarm processing as described in the first aspect is implemented.

[0011] By adopting the above technical solution, it is possible to determine the corresponding warning classification labels for the pending alarms reported by multiple different information collection devices through a pre-constructed knowledge graph or a trained multi-fusion model, thereby avoiding the difficulties in subsequent analysis and judgment caused by the alarms coming from different information collection devices reporting different original alarm information, reducing the workload and error probability of manual inspection, analysis and classification of alarms, and improving work efficiency.

[0012] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0013] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0015] Figure 1 A schematic diagram showing a flow chart of a method for multi-source alarm processing provided by an embodiment of the present disclosure; Figure 2 A schematic diagram showing the structure of the knowledge graph provided by an embodiment of the present disclosure; Figure 3 Another flowchart of the method for multi-source alarm processing provided by an embodiment of the present disclosure is shown; Figure 4Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown; Figure 5 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown; Figure 6 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown; Figure 7 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown; Figure 8 A block diagram of a device for multi-source alarm processing provided by an embodiment of the present disclosure is shown; Fig. 9 A block diagram showing another apparatus for multi-source alarm processing provided by an embodiment of the present disclosure; Fig.10 A block diagram showing another apparatus for multi-source alarm processing provided by an embodiment of the present disclosure; Fig.11 A schematic diagram of the hardware structure of an electronic device for executing the multi-source alarm processing method provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0016] In order to enable those skilled in the art to better understand the technical solutions in the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present disclosure.

[0017] It should be noted that although the present disclosure uses farmland protection as an example for explanation, the method for multi-source alarm processing provided by the embodiments of the present disclosure can also be applied to other scenarios. For example, in the field of security, the method of the embodiments of the present disclosure can be used to process alarms reported by a variety of different security devices. The present disclosure does not limit this.

[0018] Figure 1 A flowchart of a method for multi-source alarm processing provided by an embodiment of the present disclosure is shown as follows: Figure 1 As shown, the method may include the following steps: In step S101, an alarm to be processed is obtained.

[0019] The alarm to be processed is reported by any information collection device among a plurality of different information collection devices.

[0020] The multiple different information collection devices may be different types of information collection devices. Taking farmland monitoring as an example, the multiple different information collection devices may be at least two of camera devices, satellites, and multispectral remote sensing devices. Even if they are the same type of information collection devices, they may report different alarm information of pending alarms due to different manufacturers, models, and versions. Multiple information collection devices of the same type that report different alarm information of pending alarms may also be multiple different information collection devices.

[0021] The camera device can collect pictures and videos of cultivated land, and can upload the collected pictures and videos of cultivated land to a computer connected to the camera device to analyze the pictures and videos, and report alarm information when certain preset conditions are met. Of course, the function of analyzing pictures and videos and reporting alarm information can also be integrated into the camera device, and this application does not limit this.

[0022] The satellite can also collect pictures and videos of cultivated land, and can also upload the collected pictures and videos of cultivated land to a computer connected to the satellite to analyze the pictures and videos, and report alarm information under certain preset conditions. Of course, the function of analyzing pictures and videos and reporting alarm information can also be integrated into the satellite, and this application does not limit this.

[0023] The multispectral remote sensing device can use sensors with more than two spectral channels to synchronously image the cultivated land, or it can form a video with multiple synchronous images collected in a continuous time. The collected pictures and videos of the cultivated land can be uploaded to a computer connected to the multispectral remote sensing device to analyze the pictures and videos, and report alarm information under certain preset conditions. Of course, the function of analyzing pictures and videos and reporting alarm information can also be integrated into the multispectral remote sensing device, and this application does not limit this.

[0024] The alarm to be processed may include entity information of the alarm, and the entity information may be, for example, the alarm type reported on different information collection devices. The entity information of the alarm reported on different information collection devices may be the same or different.

[0025] For example: when there is illegal construction on cultivated land, the entity information of the alarm reported on the camera device may be "simple house identification alarm", "tower crane operation monitoring alarm", "construction site dust net monitoring alarm", "mechanical construction monitoring alarm", "illegal construction and occupation monitoring alarm" or "suspected construction site identification alarm", the entity information of the alarm reported on the satellite may be "house construction", "factory construction", "park construction", "bridge" or "hardened ground", and the entity information of the alarm reported on the multispectral remote sensing device may be "non-agricultural-cultivated land becomes building" or "non-agricultural-cultivated land becomes road".

[0026] In some embodiments, the pending alarm may include alarm source information, which is used to characterize the information collection device corresponding to the alarm. The pending alarm may also include alarm time information, which is used to characterize the time when the alarm occurs. The pending alarm may also include alarm location information. In some possible implementations, the alarm location information may be, for example, the latitude and longitude information corresponding to the cultivated land to be monitored.

[0027] In some embodiments, the alarm to be processed may also include information such as alarm confidence and alarm description. The alarm confidence represents the credibility of the alarm, and the alarm description may further describe the detailed information of the alarm. For example, in the case where the alarm is a "tower crane operation monitoring alarm", the alarm description may be that the number of monitored tower cranes is 2.

[0028] In step S102, according to the entity information of the alarm to be processed, the warning classification label of the alarm to be processed is matched and determined from the pre-constructed knowledge graph.

[0029] In order to determine the warning classification labels of the alarms to be processed, a knowledge graph for processing multi-source alarms can be pre-built.

[0030] In some embodiments, the structure of the knowledge graph may include alarm source nodes corresponding to multiple different information collection devices and a multi-source data warning node. There is a corresponding path between each alarm source node and the multi-source data warning node, and each path includes the correspondence between the entity information of the alarm corresponding to the information collection device and the warning classification label.

[0031] In another embodiment, the knowledge graph may also include a map spot node associated with a multi-source data warning node. The data of the map spot node may include multiple map spots corresponding to different land units. The map spots may be based on a preset ratio (e.g., 1:10000) as a working base map, and the land units (plots) with basically the same landforms, land use types, and basically the same soil and water loss types are classified into one category, and then the land units (plots) are drawn on the topographic map to form map spots. Each map spot corresponds to a corresponding longitude and latitude range, so that the alarm location information of the multi-source alarm can be associated with the map spot, which is convenient for determining the target map spot corresponding to the alarm to be processed. One or more alarms corresponding to a map spot can also be determined.

[0032] In another embodiment, the knowledge graph may further include at least one of the following: administrative divisions associated with the multi-source data warning nodes, and persons associated with the map spots. The persons associated with the map spots may be tenants of the corresponding land units, or may be collective organizations corresponding to the land units.

[0033] Figure 2 A schematic diagram showing the structure of the knowledge graph provided by the embodiment of the present disclosure is shown as follows: Figure 2As shown, the knowledge graph includes camera equipment alarm nodes, satellite alarm nodes, multi-spectral remote sensing alarm nodes, multi-source data warning nodes, and warning classification label nodes.

[0034] Among them, the path of the camera device alarm node and the multi-source data warning node includes the corresponding relationship between the camera device alarm and the warning classification label. The path of the satellite alarm node and the multi-source data warning node includes the corresponding relationship between the satellite alarm and the warning classification label. The path of the multi-spectral remote sensing alarm node and the multi-source data warning node includes the corresponding relationship between the multi-spectral remote sensing alarm and the warning classification label.

[0035] Also connected to the multi-source data warning node may be a warning classification label node including multiple preset warning classification labels, an administrative division node including each preset administrative division, a map node including multiple map spots, and a related person node for the related person corresponding to each map spot.

[0036] In some possible implementations, the correspondence between the warning classification labels and the entity information of the alarms collected and reported by different information collection devices is shown in Table 1 below.

[0037] Table 1

[0038] As shown in Table 1 above, the first and second columns in Table 1 are the correspondence between the entity information of the alarms collected and reported by the camera equipment contained in the paths of the camera equipment alarm node and the multi-source data warning node and the warning classification labels. The first and third columns in Table 1 are the correspondence between the entity information of the alarms collected and reported by the satellite contained in the paths of the satellite alarm node and the multi-source data warning node and the warning classification labels. The first and fourth columns in Table 1 are the correspondence between the entity information of the alarms collected and reported by the multi-spectral remote sensing equipment contained in the paths of the multi-spectral remote sensing alarm node and the multi-source data warning node and the warning classification labels.

[0039] In some embodiments, the warning classification label of the alarm to be processed can be determined based on the corresponding relationship between the entity information of the alarm to be processed and the information collection device corresponding to the alarm to be processed in the path of the knowledge graph.

[0040] For example, if the alarm to be processed is an alarm collected and reported by a camera device, and the entity information is "mechanical construction monitoring alarm", then the "mechanical construction monitoring alarm" can be matched according to the correspondence between the first column and the second column in Table 1 to determine that the warning classification label corresponding to the "mechanical construction monitoring alarm" is "construction".

[0041] In some embodiments, after determining the corresponding warning classification label, it is also possible to further determine the map spot corresponding to the "mechanical construction monitoring alarm", the related people of the map spot, and the administrative division corresponding to the "mechanical construction monitoring alarm" according to the structure of the knowledge graph and the alarm location information of the alarm to be processed.

[0042] In step S103, when matching the warning classification label according to the entity information of the alarm to be processed fails, the alarm information of the alarm to be processed is input into a pre-trained multi-fusion model to determine the warning classification label.

[0043] In some embodiments, the multi-fusion model may include a first early warning classification model and a second early warning classification model.

[0044] In some possible implementations, the first warning classification model may be a random forest model, and the second warning classification model may be a BERT (Bidirectional Encoder Representations from Transformers) model. Of course, those skilled in the art may also set different warning classification models as needed, and this application does not limit this.

[0045] As an integrated learning method, the random forest model improves classification performance by building multiple decision trees and combining their prediction results. Although the random forest performs well in processing feature classification, it still has the following shortcomings in classifying multi-source alarm data: Little feature information: Multi-source alarm data has few effective features for model training, and the model selects a part of the features for training each time, resulting in a single model feature information and a lack of robustness.

[0046] Poor model interpretability: Although a single decision tree is easy to interpret, the overall decision-making process of random forest as an integrated model may not be transparent enough.

[0047] Overfitting risk: If the number of trees is too large or the depth of the trees is too large, the random forest may overfit the training data, affecting its ability to generalize to new data.

[0048] BERT is a pre-trained language model based on the Transformer architecture, which performs well in processing text data. However, the BERT model also has the following shortcomings when classifying multi-source alarm data: Data dependency: The performance of the BERT model is highly dependent on the quality and diversity of pre-training data. If the distribution of pre-training data is inconsistent with the actual alarm data, performance may degrade.

[0049] Sample distribution problem: Due to the imbalance of sample data for each classification label, the model ultimately predicts inaccurate results for small sample data with fewer labels.

[0050] In this embodiment, the alarm information of the alarm to be processed can be input into the trained random forest model and BERT model respectively, and the output of the random forest model and the BERT model can be fused to obtain the final warning classification label. The use of the above-mentioned multi-fusion model approach can integrate the advantages of different models, improve the accuracy of the model, and enhance the generalization ability. The multi-model fusion algorithm can adapt to new data patterns and changes, and is easy to expand to handle more alarm features and types. Multi-model fusion can also reduce the risk of overfitting that may exist in a single model through a diversified model combination.

[0051] Figure 3 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown as follows: Figure 3 As shown, the warning classification label of the alarm to be processed can be determined from the pre-built knowledge graph according to the entity information of the alarm to be processed. If the match is successful, the warning classification label corresponding to the alarm to be processed is obtained. If the match is unsuccessful, the alarm information of the alarm to be processed (such as entity information, alarm source information, alarm time information, alarm location information, etc.) can be encoded and mapped, respectively input into the random forest model and the BERT model, and then the output results of the random forest model and the BERT model are fused, and the warning classification label corresponding to the alarm to be processed is determined according to the fused results.

[0052] Figure 4 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown as follows: Figure 4 As shown, step S103 may specifically include the following steps.

[0053] In step S1031, the alarm information of the alarm to be processed is input into the first warning classification model to obtain first probabilities corresponding to a plurality of warning classification labels output by the first warning classification model.

[0054] For example, the first warning classification model may be a random forest model, and the alarm information of the alarm to be processed is input into the random forest model to obtain first probabilities corresponding to a plurality of warning classification labels output by the random forest model.

[0055] In step S1032, the alarm information of the alarm to be processed is input into the second warning classification model to obtain second probabilities corresponding to a plurality of warning classification labels output by the second warning classification model.

[0056] For example, the second warning classification model may be a BERT model, and the alarm information of the alarm to be processed is input into the BERT model to obtain second probabilities corresponding to a plurality of warning classification labels output by the BERT model.

[0057] In step S1033, weighted probabilities corresponding to a plurality of warning classification labels are determined according to the preset weight coefficient, the first probability and the second probability.

[0058] In some possible implementations, the weighted probabilities corresponding to the multiple warning classification labels can be determined by the following formula 1.

[0059] (Formula 1) in, is the weighted probability corresponding to the i-th warning classification label, i=1~N, N is the total number of warning classification labels, is the preset weight coefficient corresponding to the first warning classification model. When the first warning classification model is a random forest model, It can be 0.21, is the preset weight coefficient corresponding to the second warning classification model. When the second warning classification model is a BERT model, It can be 0.79. is the first probability corresponding to the i-th warning classification label, is the second probability corresponding to the i-th warning classification label.

[0060] In step S1034, the warning classification label with the highest weighted probability is used as the warning classification label of the alarm to be processed.

[0061] It is understandable that the number of different warning classification models in the multi-fusion model can be greater than 2, and the types of warning classification models are not limited to random forest models and BERT models. The types of warning classification models can also be adaptively adjusted, and more warning classification models can be added for fusion. When performing fusion, it is only necessary to adjust the corresponding preset weight coefficients according to the number of warning classification models and the performance of the models. This application does not limit this. By adopting the above-mentioned technical solution, the advantages of different models can be integrated through a multi-fusion model, the accuracy of the model can be improved, and the generalization ability can be enhanced. The multi-model fusion algorithm can adapt to new data patterns and changes, and is easy to expand to handle more alarm features and types. Multi-model fusion can also reduce the risk of overfitting that may exist in a single model through a diversified model combination.

[0062] In some embodiments, the input information of the multi-fusion model may further include alarm information of related alarms of the alarm to be processed. For example, the pattern corresponding to the alarm to be processed can be determined through the alarm position information of the alarm to be processed, and the alarm corresponding to the same pattern as the alarm to be processed, the alarm source information is other information collection equipment, and the difference in alarm time information is less than a preset threshold is taken as a related alarm. In this way, other alarms with the same pattern and similar time can be input into the multi-fusion model together with the alarm to be processed, so as to further improve the accuracy of the multi-fusion model.

[0063] By adopting the above technical solution, it is possible to determine the corresponding warning classification labels for the pending alarms reported by multiple different information collection devices through a pre-constructed knowledge graph or a trained multi-fusion model, thereby avoiding the difficulties in subsequent analysis and judgment caused by the alarms coming from different information collection devices reporting different original alarm information, reducing the workload and error probability of manual inspection, analysis and classification of alarms, and improving work efficiency.

[0064] Figure 5 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown as follows: Figure 5 As shown, step S102 may include the following steps.

[0065] In step S1021, the entity information and alarm source information of the alarm to be processed are obtained.

[0066] In some embodiments, entity information may be extracted from the alert to be processed through a universal information extraction UIE model.

[0067] For example, the alarm information of the pending alarm is "alarm source: camera alarm; alarm time: 2024-03-01 11:12:40; alarm type: yellow land exposed monitoring; alarm location: longitude XXX, latitude YYY", and the entity information of the pending alarm can be extracted from the alarm information of the pending alarm as "yellow land exposed monitoring" through the general information extraction UIE model. Similarly, the alarm source information of the pending alarm can be extracted from the alarm information of the pending alarm as "camera device alarm" through the general information extraction UIE model.

[0068] In step S1022, the target path of the alarm to be processed in the knowledge graph is determined according to the alarm source information, and at least one target correspondence relationship corresponding to the target path is obtained.

[0069] For example, according to the alarm source information of the alarm to be processed being "camera device alarm", it can be determined that the target path of the alarm to be processed in the knowledge graph is the path of the camera device alarm node and the multi-source data warning node. The target correspondence may include multiple correspondences between camera device alarms and warning classification labels shown in the first and second columns of Table 1.

[0070] In step S1023, the warning classification label of the alarm to be processed is determined by matching the entity information of the alarm to be processed with the entity information of the alarm in at least one target correspondence relationship.

[0071] The entity information of different camera device alarms in the target correspondence can be matched with the entity information of the alarm to be processed by using text similarity matching. The warning classification label corresponding to the entity information of the camera device alarm with the highest matching degree in Table 1 is used as the warning classification label of the alarm to be processed.

[0072] In some embodiments, step S1023 may specifically include the following steps: Step 10: Calculate the cosine similarity between the entity information of the alarm and the entity information of the alarm to be processed in each target correspondence relationship.

[0073] That is, each camera device alarm and the entity information of the alarm to be processed included in Table 1 are vectorized by the BERT vectorization model, and the cosine similarity of each camera device alarm and the entity information of the alarm to be processed included in Table 1 is calculated. For example, the cosine similarity between "loess exposed monitoring alarm" and "loess exposed monitoring", the cosine similarity between "farmland artificial lake monitoring alarm" and "loess exposed monitoring", the cosine similarity between "farmland tree planting monitoring alarm" and "loess exposed monitoring", ... ..., the cosine similarity between "garbage dump monitoring alarm" and "loess exposed monitoring".

[0074] Step 11: Use the warning classification label in the target corresponding relationship with the highest cosine similarity as the warning classification label of the alarm to be processed.

[0075] The warning classification label in the target correspondence relationship with the highest cosine similarity exceeding the first similarity threshold (e.g., 0.8) can be used as the warning classification label of the alarm to be processed. For example, the cosine similarity between "loess exposed monitoring alarm" and "loess exposed monitoring" is the highest, and the cosine similarity exceeds the first similarity threshold, then the "abandoned farmland" corresponding to the "loess exposed monitoring alarm" in Table 1 can be used as the warning classification label of the alarm to be processed.

[0076] If the cosine similarities corresponding to all target correspondences are less than the second similarity threshold (for example, 0.6), it is determined that the matching of the warning classification label of the pending alarm from the knowledge graph has failed.

[0077] The adoption of the above technical solution can avoid the difficulties in subsequent analysis and judgment caused by the alarms coming from different information collection devices reporting different original alarm information, reduce the workload and error probability of manual inspection, analysis and classification of alarms, and improve work efficiency.

[0078] Figure 6 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown as follows: Figure 6 As shown, the method may further include the following steps.

[0079] In step S104, the data of the knowledge graph is updated according to the correspondence between the entity information of the alarm to be processed and the warning classification label of the alarm to be processed.

[0080] In some embodiments, when the warning classification label of the pending alarm is successfully determined, and the entity information of the pending alarm is different from the entity information in the target correspondence with the highest cosine similarity, the entity information of the pending alarm can also be updated to the knowledge graph, that is, the entity information of the pending alarm is added to the corresponding path. When it is determined that the warning classification label corresponding to "loess bare monitoring" is "abandoned farmland", and it is different from the original "loess bare monitoring alarm" in the correspondence, the correspondence between "loess bare monitoring" and "abandoned farmland" can be updated to the knowledge graph, so that the knowledge graph can be updated adaptively.

[0081] In other embodiments, it is also possible to respond to the user's annotation of new warning classification labels and train the multi-fusion model based on the annotated training data to improve the generalization ability of the multi-fusion model. It can be understood that the above-mentioned multi-fusion model can output warning classification labels that do not exist in the knowledge graph, that is, the correspondence between the entity information of the alarm to be processed and the warning classification label of the alarm to be processed is a non-existent correspondence. Then, the correspondence in the knowledge graph and the corresponding warning classification label data of the warning classification label node in the knowledge graph can be updated according to the correspondence between the entity information of the alarm to be processed and the warning classification label of the alarm to be processed.

[0082] By adopting the above technical solution, the knowledge graph can automatically adapt to changes in alarm information and / or warning classification labels, further improving the adaptability of the knowledge graph.

[0083] In some embodiments, the knowledge graph can be generated through the following steps.

[0084] In step 20, a plurality of historical alarm information collected by a plurality of information collection devices is obtained, and each information collection device corresponds to at least one historical alarm information.

[0085] In step 21, alarm time information and alarm location information are extracted from a plurality of historical alarm information.

[0086] For example, the alarm time information and the alarm location information in a plurality of historical alarm information may be extracted through the UIE model.

[0087] In step 23, the image patch associated with each historical alarm information is determined according to the alarm location information.

[0088] Each map spot corresponds to a physical land unit, which can be determined by the latitude and longitude range. The alarm location information can be the location where the alarm occurs represented by the longitude and latitude. When the alarm location information is within the longitude and latitude range of a certain map spot, the association relationship between the historical alarm information and the map spot can be determined.

[0089] In step 24, relevant historical alarm information is screened from a plurality of historical alarm information, the relevant historical alarm information is associated with the same image spot, and the maximum difference of alarm time information is less than a preset threshold.

[0090] For example, the entity information of a historical alarm information reported by a camera device is "loess exposure monitoring alarm", the corresponding alarm time information is t1, and the alarm location information is "longitude XXX1, latitude YYY1". The entity information of another historical alarm information reported by a multispectral remote sensing device is "non-grain-cultivated land becomes abandoned", the corresponding alarm time information is t2, and the alarm location information is "longitude XXX2, latitude YYY2".

[0091] If the following conditions are met at the same time, the two historical alarm information can be determined as related alarm information: All historical alarm information is associated with the same image patch; The difference between t1 and t2 is smaller than a preset threshold.

[0092] It can be understood that when there are multiple historical alarm information, "the time difference between t1 and t2 is less than the preset threshold" can be adjusted to "the maximum difference between multiple alarm time information is less than the preset threshold". When the above two conditions are met, multiple historical alarm information can be determined as related alarm information.

[0093] In step 25, the historical warning classification label corresponding to the relevant historical warning information is determined, and a corresponding relationship is generated according to the historical warning classification label and the entity information in the relevant historical warning information.

[0094] When the knowledge graph is initially generated, the warning classification label of the associated historical alarm information can be determined in response to manual annotation. For example, it can be determined that the warning classification label corresponding to the two associated historical alarm information of "loess bare monitoring alarm" and "non-grain-cultivated land becoming abandoned" is "cultivated land abandonment", thereby generating the corresponding relationship in the knowledge graph as shown in the first row of Table 1 excluding the header.

[0095] By adopting the above-mentioned technical solution, it is possible to determine the historical warning classification labels corresponding to the relevant historical alarm information based on the correlation relationships among multiple historical alarm information, and generate corresponding relationships in the corresponding knowledge graph, which facilitates the subsequent determination of corresponding warning classification labels for the pending alarms reported by multiple different information collection devices through the knowledge graph, thereby improving work efficiency.

[0096] After determining a unified warning classification label for each pending alarm, multiple pending alarms under the same map within a preset time period (for example, within a week) can be classified and merged to facilitate processing by the corresponding staff, thereby further improving work efficiency.

[0097] Figure 7 Another flowchart of the method for multi-source alarm processing provided by the embodiment of the present disclosure is shown as follows: Figure 7 As shown, the method may further include the following steps.

[0098] In step S105, after obtaining the warning classification labels respectively corresponding to the multiple alarms to be processed, the fusion feature vector respectively corresponding to each alarm to be processed is obtained.

[0099] Among them, the fused feature vector is obtained by fusing at least two of the first feature vector, the second feature vector and the third feature vector, the first feature vector is a text feature vector extracted from the alarm information of the alarm to be processed, the second feature vector is a label feature vector extracted from the warning classification label of the alarm to be processed, and the third feature vector is an image feature vector extracted from the image information (which can be a picture and\or video) of the alarm to be processed.

[0100] In the related art, classification and merging are usually performed only based on the entity information of the warning information, without fully utilizing the text features and image features in the warning information, which affects the accuracy and robustness of the classification and merging.

[0101] In some possible implementations, a first feature vector can be extracted from the alarm information of the alarm to be processed through a BERT model, a second feature vector can be extracted from the warning classification label of the alarm to be processed through a BERT model, and a third feature vector can be extracted from the image information (which can be picture information or video information) of the alarm to be processed through CNN (Convolutional Neural Networks). The first feature vector, the second feature vector, and the third feature vector can be feature vectors of the same dimension, for example, they can all be M-dimensional vectors.

[0102] In a possible implementation, the fused feature vector may be obtained by the following formula 2.

[0103] (Formula 2) in, is the value of the i-th element in the fused feature vector, i=1~M, M is the dimension of the feature vector, is the value of the i-th element in the first eigenvector, is the value of the i-th element in the second eigenvector, is the value of the i-th element in the third eigenvector, The function is to find the average value.

[0104] In step S106, the multiple alarms to be processed are clustered according to the fused feature vectors respectively corresponding to the multiple alarms to be processed, so as to obtain at least one cluster corresponding to the multiple alarms to be processed.

[0105] In some embodiments, the multiple alarms to be processed may be clustered using a density clustering algorithm based on fused feature vectors corresponding to the multiple alarms to be processed, using the cosine similarity between the fused feature vectors as the distance between samples.

[0106] In some possible implementations, the density clustering algorithm may be a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm.

[0107] In step S107, multiple to-be-processed alarms are grouped and merged according to the clusters.

[0108] By adopting the above-mentioned technical scheme, the alarm data under the same map can be grouped and merged, reducing multiple warnings for the same illegal behavior. Operators no longer need to repeatedly analyze and judge the same type of alarm events in the same plot, thereby saving a lot of time and energy, and further improving the processing efficiency of multi-source alarms. At the same time, the alarm information, warning classification information and various feature information extracted from the image information are integrated during grouping and merging, further improving the accuracy of grouping and merging.

[0109] Figure 8 A block diagram of a multi-source alarm processing device provided by an embodiment of the present disclosure is shown. Figure 8 As shown, the device 100 for multi-source alarm processing includes: The acquisition module 110 is used to acquire an alarm to be processed, where the alarm to be processed is reported by any information collection device among a plurality of different information collection devices.

[0110] The determination module 120 is used to match and determine the warning classification label of the alarm to be processed from the pre-built knowledge graph according to the entity information of the alarm to be processed.

[0111] The determination module 120 is further configured to input the alarm information of the alarm to be processed into a pre-trained multi-fusion model to determine the alarm classification label when the alarm classification label fails to be matched according to the entity information of the alarm to be processed, wherein the multi-fusion model includes a first alarm classification model and a second alarm classification model; The determination module 120 is further specifically configured to: Inputting the alarm information of the alarm to be processed into the first early warning classification model to obtain first probabilities corresponding to a plurality of early warning classification labels output by the first early warning classification model; Inputting the alarm information of the alarm to be processed into the second early warning classification model to obtain second probabilities corresponding to a plurality of early warning classification labels output by the second early warning classification model; Determine the weighted probabilities corresponding to the multiple warning classification labels according to the preset weight coefficient, the first probability and the second probability; The warning classification label with the highest weighted probability is used as the warning classification label of the alarm to be processed.

[0112] Optionally, the knowledge graph includes multiple alarm source nodes corresponding to different information collection devices and a multi-source data warning node, each alarm source node and the multi-source data warning node include a corresponding path, each path includes the corresponding relationship between the entity information of the alarm corresponding to the information collection device and the warning classification label, and the determination module 120 is further used to: Get the entity information and alarm source information of the alarm to be processed; Determine the target path of the alarm to be processed in the knowledge graph according to the alarm source information, and obtain at least one target corresponding relationship corresponding to the target path; The early warning classification label of the alarm to be processed is determined by matching the entity information of the alarm to be processed with the entity information of the alarm in at least one target corresponding relationship.

[0113] Optionally, the determination module 120 is further configured to: Entity information is extracted from pending alarms through the universal information extraction UIE model.

[0114] Optionally, the determination module 120 is further configured to: Calculate the cosine similarity between the entity information of the alarm and the entity information of the alarm to be processed in each target correspondence; The warning classification label in the target correspondence with the highest cosine similarity is used as the warning classification label of the alarm to be processed.

[0115] The device 100 provided in the embodiment of the present application can execute each method in the foregoing method embodiment and realize the functions and beneficial effects of each method in the foregoing method embodiment, which will not be described in detail here.

[0116] Fig. 9 A block diagram of another multi-source alarm processing device provided by an embodiment of the present disclosure is shown. Fig. 9 As shown, the device 100 for multi-source alarm processing further includes: The graph generation and updating module 130 is used to update the data of the knowledge graph according to the correspondence between the entity information of the alarm to be processed and the warning classification label of the alarm to be processed.

[0117] Optionally, the knowledge graph also includes graph spots associated with multi-source data warning nodes, and the graph generation and updating module 130 is further used to: Acquire multiple historical alarm information collected by multiple information collection devices, each information collection device corresponds to at least one historical alarm information; Extract alarm time information and alarm location information from multiple historical alarm information; Determine the image patch associated with each historical alarm information according to the alarm location information; Filter relevant historical alarm information from multiple historical alarm information, where the relevant historical alarm information is associated with the same image spot, and the maximum difference of alarm time information is less than a preset threshold; Determine the historical warning classification label corresponding to the relevant historical warning information, and generate a corresponding relationship according to the historical warning classification label and the entity information in the relevant historical warning information.

[0118] The device 100 provided in the embodiment of the present application can execute each method in the foregoing method embodiment and realize the functions and beneficial effects of each method in the foregoing method embodiment, which will not be repeated here.

[0119] Fig.10 A block diagram of another multi-source alarm processing device provided by an embodiment of the present disclosure is shown. Fig.10 As shown, the apparatus 100 for multi-source alarm processing further includes a merging module 140, which is used to: After obtaining the warning classification labels corresponding to the plurality of to-be-processed alarms, respectively, obtaining a fused feature vector corresponding to each to-be-processed alarm, where the fused feature vector is obtained by fusing at least two of the first feature vector, the second feature vector, and the third feature vector; Clustering the multiple alarms to be processed according to the fused feature vectors respectively corresponding to the multiple alarms to be processed, so as to obtain at least one cluster corresponding to the multiple alarms to be processed; Group and merge multiple pending alarms according to clusters; Among them, the first feature vector is a text feature vector extracted from the alarm information of the alarm to be processed, the second feature vector is a label feature vector extracted from the warning classification label of the alarm to be processed, and the third feature vector is an image feature vector extracted from the image information of the alarm to be processed.

[0120] Optionally, the merging module 140 is further configured to: According to the fused feature vectors respectively corresponding to the multiple alarms to be processed, the cosine similarity between the fused feature vectors is used as the distance between samples, and the multiple alarms to be processed are clustered by a density clustering algorithm.

[0121] The device 100 provided in the embodiment of the present application can execute each method in the foregoing method embodiment and realize the functions and beneficial effects of each method in the foregoing method embodiment, which will not be repeated here.

[0122] Fig.11 FIG. 1 shows a schematic diagram of the structure of a server or a database provided in an embodiment of the present disclosure. The server or the database may be an electronic device, such as Fig.11 As shown, at the hardware level, the electronic device includes at least one processor, and optionally includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage, etc. Of course, the electronic device may also include hardware required for other services.

[0123] The processor, network interface and memory can be interconnected through an internal bus, which can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one bidirectional arrow is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0124] The memory stores the program. Specifically, the program may include a program code, and the program code includes at least one computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0125] At least one processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a device for locating the target user at the logical level. At least one processor executes the program stored in the memory and specifically performs: the method disclosed in the embodiment of the first aspect of the present disclosure and realizes the functions and beneficial effects of each method described in the method embodiment above, which will not be repeated here.

[0126] The method disclosed in the embodiment shown in the first aspect of the present disclosure can be applied to at least one processor or implemented by at least one processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware or software instructions in at least one processor. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present disclosure can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present disclosure can be directly embodied as a hardware decoding processor for execution, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0127] The electronic device can also execute the methods described in the foregoing method embodiments, and realize the functions and beneficial effects of the methods described in the foregoing method embodiments, which will not be repeated here.

[0128] Of course, in addition to software implementation, the electronic device disclosed in the present invention does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0129] The embodiments of the present disclosure also propose a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by at least one processor, they implement the method disclosed in the embodiments shown in the present disclosure and realize the functions and beneficial effects of the various methods described in the previous method embodiments, which will not be repeated here.

[0130] The computer-readable storage medium includes a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0131] Furthermore, the embodiments of the present disclosure also provide a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the following process is implemented: the method disclosed in the embodiments shown in the present disclosure and the functions and beneficial effects of the various methods described in the previous method embodiments are implemented, which will not be repeated here.

[0132] In short, the above is only a preferred embodiment of the present disclosure and does not limit the protection scope of the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

[0133] The systems, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0134] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0135] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0136] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

Claims

1. A method for multi-source alarm processing, characterized in that: The method comprises: Obtaining an alarm to be processed, where the alarm to be processed is reported by any information collection device among a plurality of different information collection devices; According to the entity information of the alarm to be processed, matching and determining the warning classification label of the alarm to be processed from the pre-built knowledge graph; In the case where the matching of the warning classification label according to the entity information of the to-be-processed alarm fails, inputting the alarm information of the to-be-processed alarm into a pre-trained multi-fusion model to determine the warning classification label, wherein the multi-fusion model includes a first warning classification model and a second warning classification model; Wherein, in the case where the matching of the warning classification label according to the entity information of the to-be-processed alarm fails, inputting the alarm information of the to-be-processed alarm into a pre-trained multi-fusion model to determine the warning classification label includes: Inputting the alarm information of the to-be-processed alarm into the first early warning classification model to obtain first probabilities corresponding to a plurality of early warning classification labels output by the first early warning classification model; Inputting the alarm information of the to-be-processed alarm into the second early warning classification model to obtain second probabilities corresponding to a plurality of early warning classification labels output by the second early warning classification model; Determine the weighted probabilities corresponding to the plurality of warning classification labels respectively according to a preset weight coefficient, the first probability and the second probability; The warning classification label with the highest weighted probability is used as the warning classification label of the alarm to be processed.

2. The method according to claim 1, characterized in that The knowledge graph includes alarm source nodes corresponding to the multiple different information collection devices and a multi-source data warning node, each of the alarm source nodes and the multi-source data warning node includes a corresponding path, and each path includes the corresponding relationship between the entity information of the alarm corresponding to the information collection device and the warning classification label; The step of matching and determining the warning classification label of the alarm to be processed from a pre-built knowledge graph according to the entity information of the alarm to be processed includes: Obtaining entity information and alarm source information of the alarm to be processed; Determine the target path of the to-be-processed alarm in the knowledge graph according to the alarm source information, and obtain at least one target corresponding relationship corresponding to the target path; The early warning classification label of the alarm to be processed is determined by matching the entity information of the alarm to be processed with the entity information of the alarm in at least one of the target corresponding relationships.

3. The method according to claim 2, characterized in that The obtaining of entity information of the to-be-processed alarm includes: Extracting the entity information from the to-be-processed alarm through a universal information extraction UIE model; The determining of the warning classification label of the alarm to be processed by matching the entity information of the alarm to be processed with the entity information of the alarm in at least one of the target correspondences respectively includes: Calculating the cosine similarity between the entity information of the alarm in each of the target correspondences and the entity information of the protection alarm to be processed; The warning classification label in the target corresponding relationship with the highest cosine similarity is used as the warning classification label of the alarm to be processed.

4. The method according to claim 1, characterized in that: The method further comprises: The data of the knowledge graph is updated according to the correspondence between the entity information of the pending alarm and the warning classification label of the pending alarm.

5. The method according to claim 2, characterized in that: The knowledge graph also includes the spots associated with the multi-source data warning nodes; the corresponding relationship is generated by the following method: Acquire multiple historical alarm information collected by the multiple information collection devices, each information collection device corresponds to at least one historical alarm information; Extracting alarm time information and alarm location information from the plurality of historical alarm information; Determine a map spot associated with each historical alarm information according to the alarm location information; Filter relevant historical alarm information from the multiple historical alarm information, where the relevant historical alarm information is associated with the same image spot and the maximum difference of alarm time information is less than a preset threshold; The historical warning classification label corresponding to the relevant historical warning information is determined, and the corresponding relationship is generated according to the historical warning classification label and the entity information in the relevant historical warning information.

6. The method according to claim 5, characterized in that The knowledge graph also includes at least one of the following: The administrative divisions associated with the multi-source data warning nodes; The persons related to the said spot.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: After obtaining the warning classification labels respectively corresponding to the multiple alarms to be processed, obtaining a fused feature vector respectively corresponding to each of the alarms to be processed, wherein the fused feature vector is obtained by fusing at least two of the first feature vector, the second feature vector and the third feature vector; Clustering the multiple alarms to be processed according to the fused feature vectors respectively corresponding to the multiple alarms to be processed, so as to obtain at least one cluster corresponding to the multiple alarms to be processed; Grouping and merging the multiple to-be-processed alarms according to the clustering clusters; Among them, the first feature vector is a text feature vector extracted from the alarm information of the alarm to be processed, the second feature vector is a label feature vector extracted from the warning classification label of the alarm to be processed, and the third feature vector is an image feature vector extracted from the image information of the alarm to be processed.

8. The method according to claim 7, characterized in that Clustering the multiple alarms to be processed according to the fused feature vectors respectively corresponding to the multiple alarms to be processed includes: According to the fused feature vectors respectively corresponding to the multiple alarms to be processed, the cosine similarity between the fused feature vectors is used as the distance between samples, and the multiple alarms to be processed are clustered by a density clustering algorithm.

9. A device for multi-source alarm processing, characterized in that: The device comprises: An acquisition module, used for acquiring an alarm to be processed, wherein the alarm to be processed is reported by any information collection device among a plurality of different information collection devices; A determination module, configured to match and determine the warning classification label of the alarm to be processed from a pre-built knowledge graph according to the entity information of the alarm to be processed; The determination module is further configured to input the alarm information of the alarm to be processed into a pre-trained multi-fusion model to determine the alarm classification label when the alarm classification label fails to be matched according to the entity information of the alarm to be processed, wherein the multi-fusion model includes a first alarm classification model and a second alarm classification model; Wherein, in the case where the matching of the warning classification label according to the entity information of the to-be-processed alarm fails, inputting the alarm information of the to-be-processed alarm into a pre-trained multi-fusion model to determine the warning classification label includes: Inputting the alarm information of the to-be-processed alarm into the first early warning classification model to obtain first probabilities corresponding to a plurality of early warning classification labels output by the first early warning classification model; Inputting the alarm information of the to-be-processed alarm into the second early warning classification model to obtain second probabilities corresponding to a plurality of early warning classification labels output by the second early warning classification model; Determine the weighted probabilities corresponding to the plurality of warning classification labels respectively according to a preset weight coefficient, the first probability and the second probability; The warning classification label with the highest weighted probability is used as the warning classification label of the alarm to be processed.

10. A server, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the multi-source alarm processing method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for multi-source alarm processing according to any one of claims 1 to 8 is implemented.

Images