Fire fighting system-oriented event rule mining method and device

Abstract

The invention relates to an event rule mining method and device for a fire fighting system. The method comprises the following steps: acquiring equipment data of N fire fighting equipment detected in real time in the fire fighting system; n is a positive integer; extracting a target event in the fire fighting system based on the equipment data; determining M first prediction events associated with the target event according to the fire fighting system topology of the fire fighting system and the event similarity; m is a positive integer; and monitoring the target event, and obtaining an event rule mining result of the fire fighting system based on the relationship between the next occurrence event of the target event and the M first prediction events. Through the mode, whether unknown characterization of a known event or a completely unknown event exists or not can be effectively judged, so that the comprehensiveness of state analysis of the fire fighting system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an event rule mining method and device for a fire protection system. Background Art

[0002] With the continuous progress of sensor technology, fire protection means have gradually changed from past manually operated devices such as handheld fire extinguishers and fire hydrants to intelligent fire protection systems with autonomous perception and autonomous fire extinguishing. These intelligent fire protection systems include the coordinated use of various devices such as sensors, monitoring equipment, and alarm systems. These devices generate a large amount of data, including various information about the occurrence of fires. By mining and analyzing the events occurring in the fire protection system, we can better understand the causes, spread, and extinguishing process of fires, predict the development trend of fires, take preventive measures in advance to prevent the occurrence of fire accidents, and improve the response speed and processing effect of the fire protection system. In addition, the large amount of data accumulated in the fire protection system can also provide important evidence for the investigation and analysis of fire accidents, help improve fire protection facilities and preventive measures, and enhance the level of fire safety management. Therefore, event mining for the fire protection system has important practical significance and application value.

[0003] For event mining in the fire protection system, the classic method is mainly the template matching method, that is, comparing data features with the features of existing event data. With the continuous development of deep learning and machine learning, some novel methods have emerged, such as sequential pattern mining, time series analysis, etc. Through these technologies, hidden patterns and rules in event data can be discovered, which helps users better understand the reasons and mechanisms behind events, and can also provide prediction and decision-making support for the occurrence of events to help users make more accurate decisions. However, these technologies all have the same defect, that is, they can only judge data under the same task rule. For example, in the fire warning task, it is judged whether the data collected by the sensors of the intelligent fire protection system represents a fire according to its relevant rules (indicating the presence or absence of a fire). Although the existing technologies already have good performance, due to the existence of the above defects, they show two weaknesses. One is that they cannot comprehensively explore the data under the same task (the rules are determined, and there are some rules that can represent other hidden features that have not been discovered), and the other is that they do not have the multi-task ability (one model or one technology can only solve one task). Summary of the Invention

[0004] To solve the above problems of the existing technology, the present invention provides an event rule mining method and device for a fire protection system.

[0005] In a first aspect, an embodiment of the present application provides an event rule mining method for a fire protection system, including: obtaining device data of N fire protection devices detected in real time in the fire protection system; N is a positive integer; extracting target events in the fire protection system based on the device data; determining M first predicted events associated with the target events according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer; monitoring the target events, and obtaining an event rule mining result of the fire protection system based on the relationship between the next occurrence event of the target event and the M first predicted events.

[0006] In an optional implementation manner of the first aspect, the device data detected by each fire protection device includes at least one of operation data and perception data; the operation data is data detected for the operation of the fire protection device itself; the perception data is data obtained by the fire protection device for perceiving the environment.

[0007] In an optional implementation manner of the first aspect, the extracting target events in the fire protection system based on the device data includes: inputting the operation data and the perception data into an autoencoder, and after projecting the perception data to the same dimension as the operation data through the autoencoder, performing event recognition to extract the target events.

[0008] In an optional implementation manner of the first aspect, the autoencoder is a two-stage network, and the two-stage network includes a projection network and a visual self-attention network; the projection network is used to project the perception data to the same dimension as the operation data; the visual self-attention network is used to perform event recognition based on the perception data and operation data in the same dimension to extract the target events; the loss function of the projection network is based on KL divergence; the expression of the loss function of the projection network includes: ; where represents the loss function of the projection network; represents the KL divergence; represents the perception data; represents the operation data; represents the data with dimension alignment in the perception data; represents the data with dimension alignment in the operation data; represents the projection network; the expression of the loss function of the visual self-attention network includes: ; where represents the loss function of the visual self-attention network; represents the number of fire protection devices; represents the event metric regarding the operation data; Represents an event metric regarding the perceived data; Represents the visual self-attention network.

[0009] In an alternative implementation of the first aspect, the determining the M first predicted events associated with the target event according to the fire protection system topology and event similarity of the fire protection system includes: obtaining G known events in the fire protection system; calculating the association degree between each known event and the target event based on the fire protection system topology of the fire protection system and the event similarity between each known event and the target event; and extracting the M first predicted events from the G known events according to the association degree between each known event and the target event.

[0010] In an alternative implementation of the first aspect, the calculation formula for the association degree between each known event and the target event includes: ; where Represents the target event; Represents any one of the known events; Represents the association degree between the target event and the known event; Represents the path length between the target event and the known event in the fire protection system topology diagram; Represents the similarity between the target event and the known event at the data level.

[0011] In an alternative implementation of the first aspect, the obtaining the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the target event and the M first predicted events includes: if the first event is not one of the M first predicted events, then determining M second predicted events associated with each first predicted event according to the fire protection system topology and event similarity of the fire protection system; where the first event is the next occurrence event of the target event; the total number of the second predicted events is ; if the second event is one of the second predicted events, then determining the second event as the hidden representation of the first event; where the second event is the next occurrence event of the first event; the event rule mining result of the fire protection system includes that the second event is the hidden representation of the first event.

[0012] In an alternative implementation of the first aspect, the method further includes: if the second event is not the If one of the second predicted events occurs, a new event rule is determined based on the target event, the first event, and the second event; wherein, the event rule mining result of the fire protection system includes the new event rule.

[0013] In an optional implementation manner of the first aspect, the method further includes: if the first event is one of the M first predicted events, determining M third predicted events associated with the first event according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer; monitoring the first event, and obtaining the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the first event and the M third predicted events.

[0014] In a second aspect, an event rule mining device for a fire protection system provided by an embodiment of the present application includes: a data acquisition module, configured to acquire device data of N fire protection devices detected in real time in the fire protection system; N is a positive integer; an event extraction module, configured to extract a target event in the fire protection system based on the device data; an activity model correlation module, configured to determine M first predicted events associated with the target event according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer; a rule mining module, configured to monitor the target event, and obtain the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the target event and the M first predicted events.

[0015] The beneficial effects of the present invention include: The embodiment of the present application provides an event rule mining method for a fire protection system. First, device data of N fire protection devices detected in real time in the fire protection system is acquired; then, a target event in the fire protection system is extracted based on the device data; then, M first predicted events associated with the target event are determined according to the fire protection system topology and event similarity of the fire protection system; finally, the target event is monitored, and the event rule mining result of the fire protection system can be obtained based on the relationship between the next occurrence event of the target event and the M first predicted events. That is, the embodiment of the present application proposes a new rule mining method, which extracts target events by using the device data of N fire protection devices in the fire protection system, combines the relevance of events for prediction, and uses the prediction results and actual monitoring situations for rule mining. Through this method, it can be determined whether there are unknown representations of known events or completely unknown events, thereby improving the comprehensiveness of the analysis of the state of the fire protection system. Description of the Drawings

[0016] Figure 1 It is a step flow chart of an event rule mining method for a fire protection system provided by an embodiment of the present invention; Figure 2A schematic diagram of a module of a fire protection system provided by an embodiment of the present invention; Figure 3 A block diagram of a module of an event rule mining device for a fire protection system provided by an embodiment of the present invention; Figure 4 A block diagram of a module of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0017] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0018] It should be understood that when used in the specification and the appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0019] It should also be understood that the term "and / or" used in the specification and the appended claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0020] As used in the specification and the appended claims of the present application, the term "if" can be interpreted as "when", "once", "in response to determining", or "in response to detecting" depending on the context.

[0021] In addition, in the description of the specification and the appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0022] The reference to "one embodiment" or "some embodiments" etc. described in the specification of the present application means that a specific feature, structure, or characteristic described in conjunction with the embodiment is included in one or more embodiments of the present application. Thus, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0023] For event mining in a fire protection system, the classic method is mainly the template matching method, which compares data features with the features of existing event data. With the continuous development of deep learning and machine learning, some novel methods have emerged, such as sequential pattern mining, time series analysis, etc. Through these technologies, hidden patterns and regularities in event data can be discovered, helping users better understand the reasons and mechanisms behind events, and can also provide predictions and decision-making support regarding the occurrence of events, helping users make more accurate decisions. However, all these technologies have the same defect, that is, they can only judge data under the same task rules. For example, in the fire warning task, according to its relevant rules, it is judged whether the data collected by the sensors of the intelligent fire protection system represents a fire (indicating a fire or indicating no fire). Although the existing technologies already have good performance, due to the existence of the above defects, they show two weaknesses. One is that they cannot comprehensively explore the data under the same task (the rules are fixed, and there are some rules that can represent other hidden features that have not been discovered), and the other is that they do not have multi-task capabilities (one model or one technology can only solve one task).

[0024] In view of the above problems, the present application proposes the following embodiments to solve the above technical problems.

[0025] Please refer to Figure 1 , in a first aspect, an event rule mining method for a fire protection system provided by an embodiment of the present application includes: steps 101 to 104.

[0026] Step 101: Obtain the device data of N fire protection devices detected in real time in the fire protection system; N is a positive integer.

[0027] First, all fire protection devices in the fire protection system will be detected in real time to obtain their device data. Here, it is assumed that there are N fire protection devices in the fire protection system. The specific number of N can be determined according to the actual configuration of the fire protection system.

[0028] Please refer to Figure 2 , Figure 2 shows some fire protection devices in the fire protection system. The fire protection devices can include various sensors, monitoring devices, and alarm devices, etc. For example, the fire protection devices can specifically be temperature sensors, smoke sensors, humidity sensors, cameras, gas detectors, fire pumps, sirens, etc. It should be noted that Figure 2 the fire protection devices shown are only examples. In specific applications, the fire protection system can include more fire protection devices than Figure 2 shown, or different fire protection devices from Figure 2 shown. The present application does not make any limitations.

[0029] Step 102: Extract target events in the fire protection system based on device data.

[0030] Here, the target events in the fire protection system are extracted based on device data, that is, the target events are obtained through the analysis and processing of device data.

[0031] It should be noted that an event can be understood as a definable thing / situation that occurs in the fire protection system. For example, the camera is overheated, the current of the smoke alarm is 0 (open circuit fault), the water pressure at the water spray nozzle is lower than the threshold (insufficient water pressure), a fire with smoke but no open flame, a fire with smoke and open flame, no fire, etc.

[0032] The operation process of the fire protection system can be understood as a series of events connected in sequence over time. For example, as the fire spreads, event one: the carbon monoxide concentration reaches 1.1% transitions to event two: the carbon monoxide concentration reaches 1.7%.

[0033] Step 103: Determine M first predicted events associated with the target event according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer.

[0034] Then, the fire protection system topology of the fire protection system and the known events in the fire protection system can be obtained, and M first predicted events associated with the target event are determined based on the fire protection system topology of the fire protection system and the event similarity with the known events.

[0035] In the embodiment of the present application, M is 5. That is, 5 first predicted events associated with the target event are determined according to the fire protection system topology and event similarity of the fire protection system.

[0036] Of course, in other embodiments, M can also be 3, 7, 9, etc.

[0037] The above determination process can extract M first predicted events in order from the largest to the smallest degree of association.

[0038] Step 104: Monitor the target event, and obtain the event rule mining result of the fire protection system based on the relationship between the next occurring event of the target event and the M first predicted events.

[0039] Finally, monitor the target event that occurs at the current moment, and when the target event transitions to the next occurring event, judge the relationship between the next occurring event and the M first predicted events, and then obtain the event rule mining result of the fire protection system.

[0040] In summary, the embodiments of the present application provide an event rule mining method for a fire protection system. First, device data of N fire protection devices detected in real time in the fire protection system is obtained; then, target events in the fire protection system are extracted based on the device data; then, M first predicted events associated with the target events are determined according to the fire protection system topology and event similarity of the fire protection system; finally, the target events are monitored, and based on the relationship between the next occurrence event of the target event and the M first predicted events, the event rule mining result of the fire protection system can be obtained. That is, the embodiments of the present application propose a new rule mining method, which uses the device data of N fire protection devices in the fire protection system to extract target events, combines the relevance of events for prediction, and uses the prediction results and actual monitoring situations for rule mining. In this way, it can be determined whether there are unknown representations of known events or completely unknown events, thereby improving the comprehensiveness of the analysis of the state of the fire protection system.

[0041] Optionally, the device data detected by each fire protection device includes at least one of operation data and perception data; the operation data is the data detected for the operation of the fire protection device itself; the perception data is the data obtained by the fire protection device for perceiving the environment.

[0042] Exemplarily, the fire protection system is composed of N fire protection devices, and the operation data of the fire protection device can be expressed as ; among them, the operation data may specifically include: data of the fire protection device itself such as voltage, current, and water pressure.

[0043] The perception data of the fire protection device can be expressed as ; among them, can be used to represent the type of perception data collected by the device. The type of perception data collected by a single fire protection device consists of a single type or multiple types. The perception data may specifically include: data such as smoke concentration, temperature, and humidity obtained by the fire protection device for perception.

[0044] Correspondingly, step 102 of extracting target events in the fire protection system based on device data may specifically include: inputting the operation data and the perception data into an autoencoder, and after projecting the perception data to the same dimension as the operation data through the autoencoder, performing event recognition to extract the target events.

[0045] It should be noted that before inputting the operation data and the perception data into the autoencoder, it is necessary to align the operation data and the perception data according to the time stamp, which can effectively improve the accuracy of subsequent event extraction and the accuracy of event rule mining.

[0046] Specifically, align the operation data and the perception data according to the time stamp to obtain: , and so on.

[0047] The network structure of the autoencoder will be described below. In the embodiments of the present application, the autoencoder is a two-stage network, and the two-stage network includes a projection network and a visual self-attention network.

[0048] The autoencoder can be represented by the following formula: ; where the above formula represents the working principle of the autoencoder, that is, first project the perceptual data onto the same dimension as the running data , and then perform joint feature extraction. In this formula, represents the data with dimension alignment in the perceptual data; represents the data with dimension alignment in the running data; represents the projection network; represents the visual self-attention network (i.e., the ViT network).

[0049] Specifically, the projection network is used to project the perceptual data onto the same dimension as the running data; the visual self-attention network is used to perform event recognition based on the perceptual data and the running data of the same dimension to extract the target event.

[0050] The loss function of the projection network is based on the KL divergence; the expression of the loss function of the projection network includes: ; where represents the loss function of the projection network; represents the KL divergence; represents the said perceptual data; represents the running data; represents the data with dimension alignment in the perceptual data; represents the data with dimension alignment in the running data; The expression of the loss function of the visual self-attention network includes: ; where represents the loss function of the visual self-attention network; represents the number of fire-fighting devices; represents the event index regarding the running data; represents the event index regarding the perceptual data. When training the above autoencoder, make + that's it.

[0051] In summary, in the embodiments of the present application, the perceptual data is first passed through the autoencoder Projected onto the same dimension as the operation data and then joint feature extraction is performed, and finally the target event is output. That is, the present invention uses the perception data and the operation data to complete the joint event extraction through projection. The extraction result can not only be used for rule mining, but also for all downstream tasks involving perception data and operation data, making the analysis of the fire protection system more diverse, and there is no need to build a specific model for a specific task as in the prior art.

[0052] Optionally, M first predicted events associated with the target event are determined according to the fire protection system topology and event similarity of the fire protection system, including: obtaining G known events in the fire protection system; calculating the association degree between each known event and the target event based on the fire protection system topology of the fire protection system and the event similarity between each known event and the target event; and extracting M first predicted events from the G known events according to the association degree between each known event and the target event.

[0053] It should be noted that the operation process of the fire protection system is obtained by connecting events in series in time. Assuming that these events are known in advance, the operation process of the fire protection system is transformed into an event assignment task, and a specific event is assigned to the system at each specific moment. A series of assigned behaviors are defined as activities. Assuming that the specific time period is ~ , at the moment (start), the assigned event is , the duration of is , then the next event that occurs should have the highest association degree with ~ The sum of the durations of the events started within the time period should be greater than - .

[0054] Optionally, in the embodiment of the present application, the calculation formula for the association degree between the above-mentioned known event and the target event includes: ; In this formula, represents the target event; represents any known event; represents the association degree between the target event and the known event; represents the path length between the target event and the known event in the fire protection system topology diagram; it should be noted that here, the target event and the known event The path length between the target events Source device to known events The access step length of the source device is determined by the traversal method, which is depth-optimal traversal. Indicates the similarity between the target event and the known event at the data level. The similarity at the data level can be understood as the similarity in data changes, such as the spread of fire, the similarity of carbon monoxide concentration and indoor temperature changes. The above similarity can adopt Jcard similarity, which is not limited in this application.

[0055] Optionally, in the above step 104, based on the relationship between the next occurrence event of the target event and the M first predicted events, the event rule mining result of the fire protection system is obtained, which may specifically include the following three situations: In the first case, if the first event is not one of the M first predicted events, then M second predicted events associated with each first predicted event are determined according to the fire system topology and event similarity of the fire system; wherein the first event is the next occurrence event of the target event; the total number of second predicted events is If the second event is If one of the second predicted events is selected, the second event is determined to be a hidden representation of the first event; wherein the second event is the next occurrence event of the first event; and the event rule mining result of the fire protection system includes that the second event is a hidden representation of the first event.

[0056] In other words, the first case can mine hidden representations between events, that is, mine unknown representations.

[0057] For example, when M=5, when acquiring the target event Then, five associated first predicted events are obtained through step 103, including ( , , , , ), then, for the target event Monitor and when an event changes, enter the next event (the first event ), judge the first event Does it belong to ( , , , , ), if not, then based on the above step 103, five second predicted events associated with each first event are determined, for example, The associated 5 second prediction events include ( , , , , ); The five second prediction events associated with the first prediction event include ( , , , , ); The five second prediction events associated with the first prediction event include ( , , , , ); The five second prediction events associated with the first prediction event include ( , , , , ); The five second prediction events associated with the first prediction event include ( , , , , ). The total number of second prediction events is 25. Then, continue to judge the next occurrence event of the first event (i.e., the second event ). If the second event belongs to one of the above 25 second prediction events, then it is determined that the second event is the hidden representation of the first event .

[0058] The second case: If the second event is not one of the second prediction events, then a new event rule is determined based on the target event, the first event, and the second event; among them, the event rule mining result of the fire protection system includes the new event rule.

[0059] That is, continue to illustrate with the previous example. If the second event does not belong to one of the above 25 second prediction events, then based on the target event , the first event , and the second event a new event rule is determined. At this time, the staff can be notified to extract the specific operating status and perception data of the equipment and analyze the specific representation of this rule.

[0060] In the third case, if the first event is one of the M first predicted events, determine M third predicted events associated with the first event according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer; monitor the first event, and obtain the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the first event and the M third predicted events.

[0061] In other words, when the first event is one of the M first predicted events, it represents the prediction of a known activity. Then, continue to perform the above method on the next event to continue monitoring and mining.

[0062] Exemplarily, M = 5 above. After obtaining the target event then obtain 5 associated first predicted events through step 103, including ( , , , , ). Then, monitor the target event . When the occurrence event changes, that is, when entering the next event (the first event ), determine whether the first event belongs to one of ( , , , , ). If so, based on the above step 103, determine M third predicted events associated with the first event, and continue to monitor the first event , and obtain the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the first event and the M third predicted events.

[0063] Please refer to Figure 3 . Based on the same inventive concept, an event rule mining device 300 for a fire protection system provided by an embodiment of the present application includes: A data acquisition module 301, configured to acquire device data of N fire protection devices detected in real time in the fire protection system; N is a positive integer.

[0064] An event extraction module 302, configured to extract a target event in the fire protection system based on the device data.

[0065] An activity model correlation module 303, configured to determine M first predicted events associated with the target event according to the fire protection system topology and event similarity of the fire protection system; M is a positive integer.

[0066] A rule mining module 304 is configured to monitor the target event and obtain an event rule mining result of the fire protection system based on the relationship between the next occurrence event of the target event and the M first prediction events.

[0067] Optionally, the device data detected by each fire protection device includes at least one of operation data and perception data; the operation data is data detected for the operation of the fire protection device itself; the perception data is data obtained by the fire protection device for perceiving the environment.

[0068] Optionally, the event extraction module 302 is further specifically configured to input the operation data and the perception data into an autoencoder, project the perception data to the same dimension as the operation data through the autoencoder, and then perform event recognition to extract the target event.

[0069] Optionally, the autoencoder is a two-stage network, and the two-stage network includes a projection network and a visual self-attention network; the projection network is used to project the perception data to the same dimension as the operation data; the visual self-attention network is used to perform event recognition based on the perception data and operation data of the same dimension to extract the target event; The loss function of the projection network is based on KL divergence; the expression of the loss function of the projection network includes: ; where represents the loss function of the projection network; represents the KL divergence; represents the perception data; represents the operation data; represents the data with dimension alignment in the perception data; represents the data with dimension alignment in the operation data; represents the projection network; The expression of the loss function of the visual self-attention network includes: ; where represents the loss function of the visual self-attention network; represents the number of fire protection devices; represents the event metric regarding the operation data; represents the event metric regarding the perception data; represents the visual self-attention network.

[0070] Optionally, the active model correlation module 303 is further configured to obtain G known events in the fire protection system; calculate the correlation degree between each of the known events and the target event based on the fire protection system topology of the fire protection system and the event similarity between each of the known events and the target event; and extract the M first predicted events from the G known events according to the correlation degree between each of the known events and the target event.

[0071] Optionally, the calculation formula for the correlation degree between each of the known events and the target event includes: ; where represents the target event; represents any one of the known events; represents the correlation degree between the target event and the known event; represents the path length between the target event and the known event in the fire protection system topology diagram; represents the similarity between the target event and the known event at the data level.

[0072] Optionally, the rule mining module 304 is further specifically configured to, if the first event is not one of the M first predicted events, determine M second predicted events associated with each of the first predicted events according to the fire protection system topology of the fire protection system and the event similarity; where the first event is the next occurrence event of the target event; the total number of the second predicted events is ; if the second event is one of the second predicted events, determine the second event as the hidden representation of the first event; where the second event is the next occurrence event of the first event; the event rule mining result of the fire protection system includes that the second event is the hidden representation of the first event.

[0073] Optionally, the rule mining module 304 is further specifically configured to, if the second event is not one of the second predicted events, determine a new event rule based on the target event, the first event, and the second event; where the event rule mining result of the fire protection system includes the new event rule.

[0074] Optionally, the rule mining module 304 is further specifically configured to, if the first event is one of the M first predicted events, determine M third predicted events associated with the first event according to the fire protection system topology of the fire protection system and the event similarity; M is a positive integer; monitor the first event, and obtain the event rule mining result of the fire protection system based on the relationship between the next occurrence event of the first event and the M third predicted events.

[0075] Please refer to Figure 4 Figure 4 , based on the same inventive concept, an embodiment of the present application provides a module housing of an electronic device 400 that applies the above-mentioned event rule mining method for a fire protection system. The electronic device 400 includes: at least one processor 401 ( Figure 4 only one is shown in the figure), a memory 402, and a computer program 403 stored in the memory 402 and executable on at least one processor 401. When the processor 401 executes the computer program 403, the steps of the event rule mining method for a fire protection system in any of the foregoing embodiments are implemented.

[0076] The electronic device 400 may be a personal computer, a laptop computer, etc., and the electronic device 400 may also be the imaging device itself. When the electronic device 400 is a personal computer, the personal computer is electrically connected or communicatively connected to the imaging device.

[0077] The processor 401 and the memory 402. Those skilled in the art can understand that Figure 4 merely examples of the electronic device 400, which do not constitute a limitation on the electronic device 400, may include more or fewer components than shown in the figure, or combine certain components, or different components.

[0078] The so-called processor 401 may be a central processing unit (CPU), and the processor 401 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0079] In some embodiments, the memory 402 may be an internal storage unit of the electronic device 400, such as the hard disk or memory of the electronic device 400. In other embodiments, the memory 402 may also be an external storage device of the electronic device 400, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 400. Further, the memory 402 may also include both an internal storage unit and an external storage device of the electronic device 400.

[0080] It should be noted that for the information interaction, execution process, etc. between the above-mentioned devices / modules, since they are based on the same concept as the method embodiments of this application, for their specific functions and the technical effects brought about, reference can be specifically made to the method embodiment part, and details will not be elaborated here.

[0081] Those skilled in the art can clearly understand that for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In practical applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, and details will not be elaborated here.

[0082] The embodiments of this application also provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0083] The embodiments of this application provide a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can execute the steps in the above-mentioned method embodiments.

[0084] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-mentioned embodiment methods of this application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the photographing device / electronic device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0085] In the above embodiments, the descriptions of the various embodiments have their own focuses. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0086] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0087] In the embodiments provided in this application, it should be understood that the disclosed device / network device and method can be implemented in other ways. For example, the device / network device embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in an electrical, mechanical, or other form.

[0088] The unit described as a separation component may or may not be physically separated. The component displayed as a unit may or may not be a physical unit, that is, it may be located in one place or distributed across multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0089] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. An event rule mining method for a fire protection system, characterized in that, Including: Obtaining device data of N fire-fighting devices detected in real time in the fire-fighting system; N is a positive integer; Extracting a target event in the fire-fighting system based on the device data; Determining M first predicted events associated with the target event according to the fire-fighting system topology and event similarity of the fire-fighting system; M is a positive integer; Monitoring the target event and obtaining an event rule mining result of the fire-fighting system based on the relationship between the next occurrence event of the target event and the M first predicted events.

2. The method according to claim 1, characterized in that The device data detected by each fire-fighting device includes at least one of operation data and perception data; the operation data is data detected for the operation of the fire-fighting device itself; the perception data is data obtained by the fire-fighting device for perceiving the environment.

3. The method according to claim 2, wherein The extracting the target event in the fire-fighting system based on the device data includes: Inputting the operation data and the perception data into an autoencoder, and after projecting the perception data to the same dimension as the operation data through the autoencoder, performing event recognition to extract the target event.

4. The method according to claim 3, characterized in that, The autoencoder is a two-stage network, and the two-stage network includes a projection network and a visual self-attention network; the projection network is used to project the perception data to the same dimension as the operation data; the visual self-attention network is used to perform event recognition based on the perception data and operation data of the same dimension to extract the target event; The loss function of the projection network is based on KL divergence; The expression of the loss function of the projection network includes: ; wherein, represents the loss function of the projection network; represents the KL divergence; represents the sensed data; represents the running data; represents the data with dimension alignment in the sensed data; represents the data with dimension alignment in the running data; represents the projection network; The expression of the loss function of the visual self-attention network includes: ; wherein, represents the loss function of the visual self-attention network; represents the number of the fire-fighting devices; represents the event metric regarding the operation data; represents the event metric regarding the perception data; represents the visual self-attention network.

5. The method according to claim 1, characterized in that, The determining M first predicted events associated with the target event according to the fire-fighting system topology and event similarity of the fire-fighting system includes: Obtaining G known events in the fire-fighting system; Calculating the association degree between each known event and the target event based on the fire-fighting system topology of the fire-fighting system and the event similarity between each known event and the target event; Extracting the M first predicted events from the G known events according to the association degree between each known event and the target event.

6. The method according to claim 5, wherein The calculation formula of the association degree between each known event and the target event includes: ; wherein, represents the target event; represents any one of the known events; represents the degree of association between the target event and the known event; represents the path length between the target event and the known event in the topological diagram of the fire protection system; represents the similarity between the target event and the known event at the data level.

7. The method according to claim 1, characterized in that The obtaining the event rule mining result of the fire-fighting system based on the relationship between the next occurrence event of the target event and the M first predicted events includes: If the first event is not one of the M first predicted events, then M second predicted events associated with each of the first predicted events are determined according to the fire protection system topology of the fire protection system and the event similarity; wherein, the first event is the next occurring event of the target event; the total number of the second predicted events is pieces; If the second event is one of the second predicted events, determine that the second event is a hidden representation of the first event; wherein the second event is the next occurring event of the first event; the event rule mining result of the fire protection system includes that the second event is a hidden representation of the first event.

8. The method according to claim 7, wherein The method further includes: If the second event is not one of the second predicted events, a new event rule is determined based on the target event, the first event, and the second event; wherein, the event rule mining result of the fire protection system includes the new event rule.

9. The method according to claim 7, wherein The method further includes: If the first event is one of the M first predicted events, determining M third predicted events associated with the first event according to the fire-fighting system topology and event similarity of the fire-fighting system; M is a positive integer; Monitoring the first event and obtaining an event rule mining result of the fire-fighting system based on the relationship between the next occurrence event of the first event and the M third predicted events.

10. An event rule mining device for a fire protection system, characterized in that, Including: A data acquisition module for obtaining device data of N fire-fighting devices detected in real time in the fire-fighting system; N is a positive integer; An event extraction module, configured to extract a target event in the fire protection system based on the device data; An activity model correlation module, configured to determine M first predicted events associated with the target event according to the fire protection system topology of the fire protection system and event similarity; M is a positive integer; A rule mining module, configured to monitor the target event, and obtain an event rule mining result of the fire protection system based on the relationship between the next occurrence event of the target event and the M first predicted events.

Images