A method and system for intelligent monitoring of targets in complex environments

Abstract

The present invention relates to a method and system for intelligent monitoring of targets in complex environments, the method comprising: screening historical monitoring data based on environmental monitoring tags to obtain event sequences corresponding to different environmental types k under the same historical monitoring data; determining characteristic event subsequences based on the event sequence for each environmental type; obtaining current event sequences corresponding to different environmental types based on current monitoring data, processing the current event sequence based on a first characteristic event subsequence and a second characteristic event subsequence; and predicting based on current monitoring data. The present invention is based on historical monitoring data supported by big data, starting from a time-based event sequence, and providing high-time resolution information about changes in complex scenarios by revealing the order and impact of events, dialectically processing monitoring data, and maintaining stable monitoring data output, which helps to more accurately monitor targets.

Description

[Technical field]

[0001] The present invention belongs to the technical field of target monitoring, and in particular relates to a method and system for intelligent target monitoring in a complex environment. [Background technology]

[0002] In terms of target monitoring methods, they can be divided into two categories: traditional methods and deep learning methods. Traditional methods mainly rely on manually designed feature extractors and classifiers to complete target positioning and identification. However, with the rise of deep learning technology, target monitoring methods based on deep neural networks have become mainstream. Representative network structures include the R-CNN series, YOLO, SSD, etc. These methods can automatically learn and extract features in images, thereby having higher accuracy and robustness in identifying and locating objects. The application range of target monitoring is very wide. In the security field, target monitoring can be used for surveillance cameras to automatically identify abnormal behaviors and events. In intelligent transportation systems, target monitoring can be used for vehicle monitoring, pedestrian flow statistics and other functions. In the medical field, in addition, target monitoring is also widely used in face recognition, autonomous driving, smart home and other fields. Target monitoring is also often used to obtain monitoring data for the target, so as to control the target action in real time;

[0003] In practical applications, on the one hand, the performance of target monitoring is affected by many factors, such as lighting conditions, object occlusion, scale changes, etc. In order to improve the accuracy and robustness of target monitoring, researchers are constantly exploring new methods and technologies. For example, use data enhancement technology to increase the diversity of training data; use multimodal fusion technology to comprehensively utilize information from different modalities; or use adaptive scale adjustment strategies to process objects of different scales. On the other hand, the complexity of target monitoring is also very high. With the development of sensor technology, more and more types of data can be provided for monitoring or real-time monitoring. The increase in processing power has also brought new challenges to the processing of monitoring data. At the same time, the amount of data that needs to be processed in real time has also shown an explosive increase trend; this has brought great challenges to target monitoring; automatic recognition and monitoring of targets in complex scenes is the current research hotspot and difficulty. How to face some complex environments and situations, such as rain, haze, occlusion, illumination changes, large differences in target scales, etc., to maintain the stability of monitoring is a problem to be solved;

[0004] Event sequences can provide high-temporal-resolution information about scene changes, which helps to monitor and track moving targets more accurately. By analyzing event sequences, moving targets that are different from background changes can be identified, thereby improving the accuracy of monitoring. Event sequence analysis focuses on analyzing the order in which events occur and their impact, and can understand the dynamic process in complex systems; most existing methods rely on large-parameter neural networks to improve the efficiency of intelligent monitoring, which means large amount of calculation and low reasoning speed, which is contrary to the high temporal resolution of event streams. In order to solve the problems in the prior art, the present invention proposes a method and system for intelligent monitoring of targets in complex environments, based on historical monitoring data supported by big data, starting from a time-based event sequence, and providing high-temporal-resolution information about changes in complex scenes by revealing the order and impact of events, dialectically processing the monitoring data, and maintaining a stable monitoring data output, which helps to monitor targets more accurately. [Summary of the invention]

[0005] In order to solve the above problems in the prior art, the present invention proposes a method and system for intelligent monitoring of targets in a complex environment, the method comprising:

[0006] Step S1: Obtain historical monitoring data for the target; filter the historical monitoring data based on the environmental monitoring tag to obtain event sequences sq corresponding to different environmental types k under the same historical monitoring data k , the event sequence is the events corresponding to the historical monitoring data that meet the environmental monitoring tags, which are sorted in chronological order;

[0007] Step S2: For each environment type, based on the event sequence sq k =(ac k,n )=(ac k,1 , ac k,2

[0008] ,…,ac k,N.k ) determining a characteristic event subsequence, wherein the characteristic event subsequence includes a first characteristic event subsequence and a second characteristic event subsequence; wherein: the first characteristic event subsequence is a typical occurrence event subsequence; the second characteristic event subsequence is a typical non-occurrence event subsequence; ac k,n is the nth event under environment type k, n = 1 ~ Nk; Nk is the number of event elements corresponding to the environmental monitoring tag of environment type k;

[0009] Step S3: Perform real-time monitoring to obtain current monitoring data, obtain current event sequences corresponding to different environment types based on the current monitoring data, and process the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence;

[0010] The processing of the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence specifically includes the following steps:

[0011] Step S3A1: Determine the interference weights win of the first characteristic event subsequence and the second characteristic event subsequence of each environment type k ;

[0012] Step S3A2: Obtain an unprocessed environment type k;

[0013] Step S3A3: Determine whether there is a second characteristic event subsequence in the current event sequence. If a subsequence matching the second characteristic event subsequence of environment type k appears in the current event sequence, it is called the target subsequence. Then, the count value ct of the target subsequence is set to j Increase win k ; Repeat this step until there is no subsequence matching its second characteristic event subsequence;

[0014] Step S3A4: if there is still unprocessed environment type k, return to step S3A2; otherwise, proceed to the next step;

[0015] Step S3A5: Calculate the weighted count value ctt=∑ k ct j ;

[0016] Step S3A6: intercepting a target subsequence whose weighted count value ctt is greater than the count value threshold; if there is a subsequence at the head of the target subsequence that constitutes the post-order part of the first characteristic event subsequence of any environment type k in the current event sequence, or there is a subsequence at the tail of the target subsequence that constitutes the pre-order part of the first characteristic event subsequence of any environment type k in the current event sequence, then the target subsequence is regarded as a pseudo target subsequence; deleting the non-pseudo target subsequence from the current event sequence, and deleting the current monitoring data portion corresponding to the deleted portion in the current event sequence;

[0017] Step S4: Make predictions based on current monitoring data.

[0018] Furthermore, the prediction model is an artificial intelligence model.

[0019] Furthermore, the prediction model is an LSTM model.

[0020] Furthermore, the LSTM model is pre-trained.

[0021] Furthermore, the monitoring data includes one or more types.

[0022] Furthermore, the count value threshold is a preset value.

[0023] Furthermore, the count value threshold is set to 1.

[0024] Furthermore, the count value ct of the target subsequence j The initial value of is 0.

[0025] A target intelligent monitoring system based on a complex environment, wherein the target intelligent monitoring system based on a complex environment is used to implement the above-mentioned target intelligent monitoring method based on a complex environment.

[0026] A target intelligent monitoring server based on a complex environment, wherein the target intelligent monitoring server based on a complex environment is used to implement the above-mentioned target intelligent monitoring method based on a complex environment.

[0027] The beneficial effects of the present invention include:

[0028] (1) Based on historical monitoring data supported by big data, starting from the time-based event sequence, by revealing the order and impact of events, it provides high-temporal resolution information about changes in complex scenes, performs dialectical processing on monitoring data, maintains stable monitoring data output, and helps to monitor targets more accurately; further: reverse dialectics are provided through the second characteristic event subsequence, and forward dialectics are achieved through the first characteristic event subsequence, ultimately eliminating the fuzzy or even contradictory data in the current monitoring data caused by the complex environment, improving the clarity of the monitoring data to support high-resolution monitoring; it helps to distinguish actual moving targets from false movements caused by environmental changes, thereby reducing false monitoring and missed monitoring, and has strong robustness;

[0029] (2) Defining environmental event labels supports event description of monitoring data from multiple dimensions, and has strong scalability of monitoring data types. At the same time, there is a many-to-many causal relationship between screening rules and events, which supports the discovery of complex environmental influencing factors and also supports cross-influences brought about by non-environmental types, with strong scalability.

Brief Description of the Drawings

[0030] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present application, but do not constitute an improper limitation of the present invention. In the drawings:

[0031] Figure 1 This is a schematic diagram of the target intelligent monitoring method in a complex environment provided by the present invention. [Specific implementation method]

[0032] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, wherein the illustrative embodiments and descriptions are only used to explain the present invention but are not intended to limit the present invention.

[0033] The present invention proposes a method and system for intelligent monitoring of targets in complex environments. Figure 1 As shown, the method comprises the following steps:

[0034] Step S1: Obtain historical monitoring data for the target; filter the historical monitoring data based on the environmental monitoring tag to obtain event sequences sq corresponding to different environmental types k under the same historical monitoring data k , the event sequence is the events corresponding to the historical monitoring data that meet the environmental monitoring tags, which are sorted in chronological order;

[0035] Preferably: the target is a moving target or a non-moving target;

[0036] Preferably: the target is an aerial target or a ground target;

[0037] The step S1 specifically includes the following steps:

[0038] Step S11: Acquire historical monitoring data for the target; the historical monitoring data is monitoring data that is sequentially sorted based on time points or time intervals, and includes multiple types of monitoring data at the same time point or time interval;

[0039] Preferably: the historical monitoring data is simulation data or actual monitoring data;

[0040] Preferably: the historical detection data is big data;

[0041] Preferably: the monitoring data at the same time point (close to the time point) or time interval are aligned according to the time point or time interval to form monitoring data arranged in chronological order;

[0042] Step S12: Obtain historical monitoring data of an unprocessed time point (taking a time point as an example, the time interval is similar) in order of time points, and filter it using the environmental monitoring tag. If the monitoring data meets an environmental monitoring tag, then put the event corresponding to the environmental monitoring tag into the event sequence sq corresponding to the environment type k. k Tail; Each environmental monitoring tag contains a set of monitoring data screening rules {r k,u}→{ac k,n}, wherein: each screening rule is used to screen one or more of the monitoring data; ac k,n is the nth event under environment type k, n = 1 ~ Nk; Nk is the number of event elements in environment type k; r k,u is the u-th screening rule in environment type k; u=1~Uk is the number of rules in the environmental monitoring tag of environment type k; the screening rule includes restrictions on the value range, type, occurrence regularity, etc. of the monitoring data;

[0043] Preferably: there are or do not exist constraints between the screening rules; one or more data screening rules correspond to one or more events;

[0044] Preferably: the monitoring data includes one or more types;

[0045] Preferably: the constraint condition is one or more of a priori conditions, time sequence conditions, etc.; each constraint condition is for one data type or multiple data types;

[0046] Preferably: the causal relationship between the screening rules and events is one-to-many, many-to-one or many-to-many, supporting the discovery of complex environmental influencing factors; event ac k,n It can exist in multiple different environment types at the same time. As long as the monitoring data can still maintain the typical occurrence or typical non-occurrence characteristics in the environment type, the event definition can be performed; subsequently, condition 2 can be used consistently to discover true subsequences;

[0047] Step S13: determine whether there is any unprocessed monitoring data, if so, obtain the historical monitoring data at the next time point, and return to step S12;

[0048] Preferably: the initial value of the event sequence corresponding to each environment type is empty;

[0049] Step S2: For each environment type, based on the event sequence sq k =(ac k,n )=(ac k,1 , ac k,2 ,…,ac k,N.k ) determine a characteristic event subsequence, wherein the characteristic event subsequence includes a first characteristic event subsequence and a second characteristic event subsequence; wherein: the first characteristic event subsequence is a typical occurrence event subsequence; the second characteristic event subsequence is a typical non-occurrence event subsequence; wherein: ac k,n is the nth event under environment type k, n = 1 ~ Nk; Nk is the number of event elements in the environmental monitoring tag of environment type k;

[0050] Step S2 specifically includes the following steps:

[0051] Step S21: obtaining an unprocessed environment type and its event sequence; the environment type includes one or more of a temperature environment type, a weather type, a climate type, a space condition type, a lighting type, an object occlusion type, a scale change type, etc.;

[0052] Step S22: If there is an event subsequence that is truly included in the event sequence and whose probability of occurrence is greater than the typical occurrence threshold, it is taken as the first characteristic event subsequence; if there is an event subsequence that is truly included in the event sequence and whose probability of occurrence is less than the typical non-occurrence threshold, it is taken as the second characteristic event subsequence;

[0053] Preferably: manually processing the typical occurrence event subsequences and the typical non-occurrence event subsequences to improve accuracy;

[0054] Preferably: for each environment type k, no deduplication operation is performed on the first characteristic event subsequence or the second characteristic event subsequence; that is, there are two subsequences in the first characteristic event subsequence or the second characteristic event subsequence of each environment type k that have a true inclusion relationship; of course, the first characteristic event subsequence and the second characteristic event subsequence need to be decontradictory, and there are no two identical subsequences between the two;

[0055] Preferably, the first characteristic event subsequence and the second characteristic event subsequence need to satisfy the condition that the number of repetitions of any event is slightly greater than the average number of event repetitions in the event sequence (for example, the adjustment coefficient is set to 1.5 to express slightly greater than); the average number of event repetitions is equal to the sum of the number of repetitions of each event divided by Nk;

[0056] Step S23: If there are still unprocessed environment types, return to step S21;

[0057] Step S3: Perform real-time monitoring to obtain current monitoring data, obtain current event sequences corresponding to different environment types based on the current monitoring data, and process the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence;

[0058] The real-time monitoring is performed to obtain current monitoring data, and the current event sequence is obtained based on the current monitoring data, specifically: the current monitoring data is processed in the manner of step S1 to obtain the current event sequence of the current monitoring data corresponding to different environment types;

[0059] The processing of the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence specifically includes the following steps:

[0060] Step S3A1: Determine the interference weights win of the first characteristic event subsequence and the second characteristic event subsequence of each environment type k ; Specifically: the interference coefficient is set based on the degree of influence of each environment type on the current environment; when the degree of influence is large, the typicality is strong, and the interference weight is set large, indicating that the error correction credibility is high; the sum of the interference weights is 1;

[0061] Alternatively: an interference weight is set based on the interference of each environment type on the current environment. If the accuracy of the monitoring data decreases greatly when environment type k appears, a higher interference weight is set for it. Otherwise, a lower interference weight is set.

[0062] Step S3A2: Obtain an unprocessed environment type k;

[0063] Step S3A3: Determine whether there is a second characteristic event subsequence in the current event sequence. If a subsequence matching the second characteristic event subsequence of environment type k appears in the current event sequence, it is called the target subsequence. Then, the count value ct of the target subsequence is set to j Increase win k ; Repeat this step until there is no subsequence matching its second characteristic event subsequence;

[0064] Preferred: The count value ct of the target subsequence j The initial value of is 0;

[0065] Step S3A4: if there is still unprocessed environment type k, return to step S3A2; otherwise, proceed to the next step;

[0066] Step S3A5: Calculate the weighted count value ctt=∑ k ct j ;

[0067] Step S3A6: obtaining a target subsequence whose weighted count value ctt is greater than a count value threshold; deleting the target subsequence from the current event sequence, and deleting the current monitoring data portion corresponding to the deleted portion in the current event sequence;

[0068] Step S3A5: Calculate the weighted count value ctt=∑ k ct j ;

[0069] Step S3A6: intercepting a target subsequence whose weighted count value ctt is greater than the count value threshold; if there is a subsequence at the head of the target subsequence that constitutes the post-order part of the first characteristic event subsequence of any environment type k in the current event sequence (and the proportion of the post-order part in the first characteristic event subsequence exceeds a preset value, for example: 50%), or there is a subsequence at the tail of the target subsequence that constitutes the pre-order part of the first characteristic event subsequence of any environment type k in the current event sequence (and the proportion of the pre-order part in the first characteristic event subsequence exceeds a preset value, for example: 50%), then the target subsequence is regarded as a pseudo-target subsequence; the non-pseudo-target subsequence and the target subsequence are deleted from the current event sequence, and the current monitoring data portion corresponding to the deleted portion in the current event sequence is deleted;

[0070] Preferably, the count value threshold is a preset value, for example, the count value threshold is set to 1; this is a setting method with a small tolerance, and the count value threshold can be comprehensively set according to the tolerance requirements and the length of the current event sequence;

[0071] Alternatively: the step S3A6 specifically includes the following steps:

[0072] Step S3A61: If the target subsequence does not appear in the event sequence corresponding to any environment type, the target subsequence is deleted from the current event sequence; the event sequence here is the event sequence corresponding to the historical monitoring data;

[0073] Step S3A62: intercepting a target subsequence whose weighted count value ctt is greater than the count value threshold; if there is a subsequence at the head of the target subsequence that constitutes the post-order part of the first characteristic event subsequence of any environment type k in the current event sequence (and the proportion of the post-order part in the first characteristic event subsequence exceeds a preset value, for example: 50%), or there is a subsequence at the tail of the target subsequence that constitutes the pre-order part of the first characteristic event subsequence of any environment type k in the current event sequence (and the proportion of the pre-order part in the first characteristic event subsequence exceeds a preset value, for example: 50%), then the target subsequence is regarded as a pseudo target subsequence; further, if there is a true subsequence in the target subsequence that satisfies the following conditions 1 and 2, then the target subsequence is regarded as a pseudo target subsequence;

[0074] Condition 1: The difference between the length of the true subsequence and the length of the target subsequence is equal to 1;

[0075] At this time, the target subsequence is corrected with a tolerance of 1 for complex environments; it can be set to other values ​​to increase the tolerance;

[0076] Condition 2: the true subsequence matches any first characteristic event subsequence;

[0077] Alternatively: Condition 2, specifically: the length difference between the true subsequence and any first characteristic event subsequence is equal to 1;

[0078] Alternative: Condition 2, specifically: the first characteristic event subsequence of the environment type k corresponding to the true subsequence and the target subsequence matches; Condition 2 is used in the scenario where the events of the environment type k intersect; the alternative form of Condition 2 is used in the scenario where the events of the environment type k do not intersect;

[0079] Step S3A63: taking the part of the pseudo target subsequence that is not the true subsequence as a residual subsequence, deleting the residual subsequence from the current event sequence, and deleting the current monitoring data part corresponding to the deleted part in the current event sequence; deleting the target subsequence that is not the pseudo target subsequence from the current event sequence, and deleting the current monitoring data part corresponding to the deleted part in the current event sequence;

[0080] Of course, the order of the two deletion steps can be adjusted according to the length of the current event sequence; or they can be performed alternately;

[0081] Step S4: performing prediction based on current monitoring data; specifically: inputting the current monitoring data into the prediction model to obtain a predicted value of the monitoring data;

[0082] Preferably: the prediction model is an artificial intelligence model;

[0083] Preferably: the prediction model is an LSTM model; the LSTM model is trained in advance, a data set is constructed based on historical monitoring data, and the monitoring data in the data set is preprocessed such as normalization; an input vector is constructed based on current monitoring data;

[0084] Preferably: the LSTM model comprises an input layer, the input layer is used to receive the input vector; the number of neurons in the input layer is equal to the number of elements in the input vector, that is, the number of types of monitoring data; the number of neurons in the output layer is equal to the number of types of monitoring data; the time for defining the input vector is not long, and the input of the current monitoring data is completed through repeated iterations of multiple steps;

[0085] Preferably: the time step is 8;

[0086] Preferred: Use mean absolute error as the loss function;

[0087] Based on the same inventive concept, the present invention also provides a target intelligent monitoring system based on a complex environment, and the system is used to implement the above-mentioned target intelligent monitoring method based on a complex environment;

[0088] The monitoring system includes a sensor device disposed at a target location, and an analysis server; the sensor device sends real-time monitoring data to the analysis server; the analysis server is provided with a prediction model;

[0089] Preferably: the server is a cloud server;

[0090] Preferably: the server is an artificial intelligence server;

[0091] Based on the same inventive concept, the present invention also provides a target intelligent monitoring server based on a complex environment, and the system is used to implement the above-mentioned target intelligent monitoring method based on a complex environment;

[0092] Based on the same inventive concept, the present invention also provides a target intelligent monitoring client based on a complex environment, and the system is used to implement the above-mentioned target intelligent monitoring method based on a complex environment;

[0093] A computer program (also referred to as a program, software, software application, script, or code) can be written in any form of a programming language, including an assembled or interpreted language, a declarative or procedural language, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program, or in multiple collaborative files (e.g., files storing one or more modules, subroutines, or code portions). A computer program can be deployed to execute on one computer or on multiple computers located at one site or distributed across multiple sites and interconnected by a communication network.

[0094] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0097] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent monitoring of targets in complex environments, characterized in that: The method comprises: Step S1: Obtain historical monitoring data for the target; filter the historical monitoring data based on the environmental monitoring tag to obtain event sequences sq corresponding to different environmental types k under the same historical monitoring data k , the event sequence is the events corresponding to the historical monitoring data that meet the environmental monitoring tags, which are sorted in chronological order; Step S2: For each environment type, based on the event sequence sq k =(ac k，m )=(ac k，1 , ac k，2 , ..., ac k,N.k ) determining a characteristic event subsequence, wherein the characteristic event subsequence includes a first characteristic event subsequence and a second characteristic event subsequence; wherein: the first characteristic event subsequence is a typical occurrence event subsequence; the second characteristic event subsequence is a typical non-occurrence event subsequence; ac k，n is the nth event under environment type k, n = 1 ~ Nk; Nk is the number of event elements corresponding to the environmental monitoring tag of environment type k; Step S3: Perform real-time monitoring to obtain current monitoring data, obtain current event sequences corresponding to different environment types based on the current monitoring data, and process the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence; The processing of the current event sequence based on the first characteristic event subsequence and the second characteristic event subsequence specifically includes the following steps: Step S3A1: Determine the interference weights win of the first characteristic event subsequence and the second characteristic event subsequence of each environment type k ; Step S3A2: Obtain an unprocessed environment type k; Step S3A3: Determine whether there is a second characteristic event subsequence in the current event sequence. If a subsequence matching the second characteristic event subsequence of environment type k appears in the current event sequence, it is called the target subsequence. Then, the count value ct of the target subsequence is set to j Increase win k ; Repeat this step until there is no subsequence matching its second characteristic event subsequence; Step S3A4: if there is still unprocessed environment type k, return to step S3A2; otherwise, proceed to the next step; Step S3A5: Calculate the weighted count value ctt=∑ k ct j ; Step S3A6: intercepting a target subsequence whose weighted count value ctt is greater than the count value threshold; if there is a subsequence at the head of the target subsequence that constitutes the post-order part of the first characteristic event subsequence of any environment type k in the current event sequence, or there is a subsequence at the tail of the target subsequence that constitutes the pre-order part of the first characteristic event subsequence of any environment type k in the current event sequence, then the target subsequence is regarded as a pseudo target subsequence; deleting the non-pseudo target subsequence from the current event sequence, and deleting the current monitoring data portion corresponding to the deleted portion in the current event sequence; Step S4: Make predictions based on current monitoring data.

2. The method for intelligent target monitoring in a complex environment according to claim 1 is characterized in that: The prediction model is an artificial intelligence model.

3. The method for intelligent target monitoring in a complex environment according to claim 2 is characterized in that: The prediction model is an LSTM model.

4. The method for intelligent target monitoring in a complex environment according to claim 3 is characterized in that: Pre-train the LSTM model.

5. The method for intelligent target monitoring in a complex environment according to claim 4 is characterized in that: Monitoring data contains one or more types.

6. The method for intelligent target monitoring in a complex environment according to claim 5 is characterized in that: The count value threshold is a preset value.

7. The method for intelligent target monitoring in a complex environment according to claim 6 is characterized in that: Set the count value threshold to 1.

8. The method for intelligent target monitoring in a complex environment according to claim 7 is characterized in that: The count value ct of the target subsequence j The initial value of is 0.

9. A target intelligent monitoring system based on a complex environment, characterized in that: The target intelligent monitoring system based on a complex environment is used to implement the target intelligent monitoring method based on a complex environment described in any one of claims 1 to 6.

10. A target intelligent monitoring server based on a complex environment, characterized in that: The target intelligent monitoring server based on a complex environment is used to implement the target intelligent monitoring method based on a complex environment described in any one of claims 1 to 6.

Images