An alarm system and method based on intelligent monitoring of a chicken farm

Abstract

The application discloses an alarm system and method based on intelligent monitoring of a chicken farm, and belongs to the technical field of burglar alarms. The alarm system specifically comprises the following steps: the anti-theft early warning scheme first generates corresponding baseline and enters the anti-theft baseline library based on the daily data of the perimeter conductive ring, the access lock spring sensing unit, the footstep pressure array, the chicken group voiceprint array and the feeding trough impedance decoy, then performs differential operation, extracts the circuit breaking event, the access lock spring unlocking sequence anomaly, the footstep pressure array track, the voiceprint disturbance and the impedance touch trace, and aggregates them into an evidence set; when the primary trigger condition is established, the feeding order is broadcasted and the chicken group voiceprint response time sequence is recorded to obtain the feeding order inquiry evidence result; the evidence set and the inquiry evidence result are combined to match the intrusion criterion rule to generate alarm information containing a trigger type identifier, a trigger position, a monitoring section index and an evidence sequence; and finally, the alarm information is used to determine the high, medium and low early warning levels according to the hierarchical rule table, and the corresponding monitoring section index is bound.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of anti-theft alarm technology, in particular to an alarm system and method based on intelligent monitoring of chicken farms. BACKGROUND

[0002] The existing farm safety protection technology mainly relies on the combination of perimeter protection facilities and traditional monitoring systems. For example, some farms install infrared beam detectors, video monitoring cameras, electronic fences and access control sensors on the boundaries of the farm area to realize real-time monitoring and recording of personnel and environment in the farm area. At the same time, the monitoring center usually receives data transmitted by sensors and cameras through wired or wireless communication networks, and sends an alarm prompt when an intrusion signal is detected. This type of alarm can improve the protection capability of the farm to some extent, but its technical characteristics are relatively single, and most of them rely only on the triggering signal of a single sensor or image change to determine the intrusion event, lacking multi-dimensional intelligent analysis and dynamic adaptation capability. Therefore, in the complex farm environment, the existing technology has the problems of insufficient monitoring accuracy and rigid alarm strategy.

[0003] In the actual operation scene of the chicken farm, in addition to natural factors, theft and human damage behavior become the main hidden danger threatening the safety of the farm. However, the existing alarm method often has a high false alarm rate in such a scene. These false alarms not only reduce the trust of the breeding personnel in the alarm system, but also increase the human cost of security disposal. As can be seen, the existing alarm method based on single signal triggering is difficult to realize effective theft and damage behavior detection in the dynamic and complex environment of the chicken farm, and it is urgent to introduce a more intelligent, multi-dimensional perception and analysis alarm method to solve the shortcomings of the existing technology. SUMMARY

[0004] The purpose of the present application is to provide an alarm system and method based on intelligent monitoring of chicken farms, which solves the problems in the background art:

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] An alarm method based on intelligent monitoring of chicken farms, comprising:

[0007] S1: generating corresponding baseline from daily data of perimeter conductive ring, access control lock spring sensing unit, footstep pressure array, chicken voiceprint array and feeding trough impedance decoy, and writing into the anti-theft baseline library;

[0008] S2: performing difference operation on real-time data and anti-theft baseline library to obtain circuit breaking event, access control lock spring unlocking sequence anomaly, footstep pressure array trajectory, voiceprint disturbance and impedance touch trace, and aggregate as evidence set;

[0009] S3: When the primary trigger condition is established, broadcast the feeding password, record the response timing of the chicken voiceprint array, and perform consistency verification on the voiceprint disturbance and footstep pressure array track in the evidence set to obtain the feeding password inquiry result;

[0010] S4: According to the evidence set and the feeding password inquiry result, match the intrusion criterion rule to generate alarm information, and the alarm information includes trigger type identification, trigger position, monitoring section index and evidence sequence;

[0011] S5: According to the hierarchical rule table and combined with the trigger type identification and evidence sequence in the alarm information, determine the high, medium or low warning level, and bind the monitoring section index;

[0012] S6: According to the warning level and the monitoring section index, drive the fence electric lock, corridor lighting, buzzer and communication dialer to execute access control locking, lighting guidance, sound and light prompt and communication reporting.

[0013] As a further scheme of the present application: in step S1, the process of generating corresponding baseline from the daily data of the perimeter conductive ring, the access lock spring sensing unit, the footstep pressure array, the chicken voiceprint array and the feeding trough impedance decoy, and writing into the anti-theft baseline library is:

[0014] During the daily feeding and night resting period when no one intrudes, the perimeter conductive ring data, the access lock spring sensing unit data, the footstep pressure array data, the chicken voiceprint array data and the feeding trough impedance decoy data are continuously collected and archived to the original data storage area;

[0015] The data is extracted from the original data storage area to generate on-off state sequence, access lock spring unlocking timing, footstep pressure array track segment, chicken voiceprint response segment and feeding trough impedance touch curve, and a daily feature table is generated in a unified field format;

[0016] Taking the night resting period as a reference, short-time noise segments are removed, and daily features are aggregated by period to obtain on-off templates, unlocking sequence templates, gait templates, voiceprint templates and impedance templates;

[0017] According to the monitoring section index, the channel number and the period label, each template is solidified as a corresponding baseline and written into the anti-theft baseline library, and the baseline version label and the effective identification are recorded.

[0018] As a further scheme of the present application: in step S2, the process of performing difference operation on the real-time data and the anti-theft baseline library to obtain the open circuit event, the access lock spring unlocking sequence anomaly, the footstep pressure array track, the voiceprint disturbance and the impedance touch trace, and aggregating them into the evidence set is:

[0019] Index the real-time data by monitoring section and segment it by uniform time window, and retrieve the same section and same time window template from the anti-theft baseline library to form a one-to-one corresponding aligned data pair;

[0020] Differential operation is performed on the aligned data pair to output the on-off deviation marker, the unlocking timing deviation marker, the pressure array position deviation marker, the voiceprint frequency band deviation marker, and the impedance touch deviation marker;

[0021] A circuit breaking event is generated according to the on-off deviation marker, an access control lock spring unlocking sequence anomaly is generated according to the unlocking timing deviation marker, and a footstep pressure array trajectory is generated according to the pressure array position deviation marker;

[0022] A voiceprint disturbance is generated according to the voiceprint frequency band deviation marker, an impedance touch trace is generated according to the impedance touch deviation marker, and the monitoring section index and the time window number are bound;

[0023] The circuit breaking event, the access control lock spring unlocking sequence anomaly, the footstep pressure array trajectory, the voiceprint disturbance, and the impedance touch trace are aggregated in time window order to generate an evidence set, taking the monitoring section index as the primary key.

[0024] As a further scheme of the present application, in the step S3, the process of obtaining the feeding password verification result is:

[0025] After determining that the primary trigger condition is met, the feeding password is broadcast through the loudspeaker according to the monitoring section index, and the password start time is recorded;

[0026] The chicken voiceprint array records the response start time and the response timing, and the footstep pressure array synchronously intercepts the footstep pressure array trajectory in the same time window;

[0027] Taking the password start time as the reference, the time windows of the voiceprint disturbance and the footstep pressure array trajectory are aligned, the consistency of the monitoring section is verified, and a consistency determination result is generated;

[0028] According to the consistency determination result and the monitoring section index, a feeding password verification result is formed, and the related evidence sequence number is bound.

[0029] As a further scheme of the present application, the specific content of the step of determining that the primary trigger condition is met is:

[0030] When a circuit breaking event or an access control lock spring unlocking sequence anomaly occurs, or a footstep pressure array trajectory and a voiceprint disturbance, a footstep pressure array trajectory and an impedance touch trace, or a voiceprint disturbance and an impedance touch trace exist simultaneously in the same monitoring section and time window, it is determined that the primary trigger condition is met.

[0031] As a further scheme of the present application, in the step S4, the process of matching the intrusion criterion rule according to the evidence set and the feeding password verification result to generate the alarm information is:

[0032] Aligning and merging the evidence set and the feeding order query result into an evidence mapping table according to the monitoring section index and the time window;

[0033] According to the feeding order query result, locating the corresponding rule branch and its matching field in the intrusion criterion rule;

[0034] According to the evidence mapping table, comparing the rule field item by item, generating a trigger type identifier, and determining the trigger position according to the monitoring section index;

[0035] Arranging the evidence entries of the monitoring section in the time window order, generating an evidence sequence and associating the trigger type identifier;

[0036] Integrating the trigger type identifier, the trigger position, the monitoring section index and the evidence sequence to generate a structured alarm information.

[0037] As a further scheme of the present application: in step S5, the process of determining the high, medium or low warning level according to the hierarchical rule table and combining the trigger type identifier and the evidence sequence in the alarm information, and binding the monitoring section index is:

[0038] Parsing the alarm information to extract the trigger type identifier, the evidence sequence and the monitoring section index as input items for hierarchical determination;

[0039] According to the trigger type identifier, locating the corresponding rule item in the hierarchical rule table, obtaining the evidence verification requirement and the warning level mapping of the item;

[0040] According to the rule item, verifying the evidence sequence item by item to confirm the consistency of the evidence and the time window, and determining the high, medium or low warning level;

[0041] Binding the warning level with the monitoring section index, and generating a hierarchical record together with the trigger type identifier and the evidence sequence.

[0042] An alarm system based on intelligent monitoring of chicken farms, comprising:

[0043] A baseline modeling module for generating corresponding baselines from the daily data of the perimeter conductive ring, the access lock spring sensing unit, the footstep pressure array, the chicken soundprint array and the feeding trough impedance decoy, and writing into the anti-theft baseline library;

[0044] An anomaly extraction module for performing difference operation on the real-time data and the anti-theft baseline library to obtain the open circuit event, the access lock spring unlocking sequence anomaly, the footstep pressure array trajectory, the soundprint disturbance and the impedance touch trace, and aggregate them into an evidence set;

[0045] The password inquiry module is configured to broadcast a feeding password when a primary trigger condition is met, record a response time sequence of a chicken soundprint array, and perform consistency verification on soundprint disturbance and footstep pressure array tracks in an evidence set according to the response time sequence, to obtain a feeding password inquiry result.

[0046] The rule matching module is configured to match an intrusion criterion rule according to the evidence set and the feeding password inquiry result, and generate an alarm information, wherein the alarm information includes a trigger type identifier, a trigger location, a monitoring section index, and an evidence sequence.

[0047] The early warning grading module is configured to determine a senior, intermediate or junior early warning level according to a grading rule table and in combination with the trigger type identifier and the evidence sequence in the alarm information, and bind the monitoring section index.

[0048] The linkage execution module is configured to drive a fence electric lock, a corridor lighting, a buzzer and a communication dialer according to the early warning level and the monitoring section index, to execute access control locking, lighting guidance, sound and light prompting and communication reporting.

[0049] The present application has the following advantages:

[0050] The present application can significantly reduce the false alarm rate of the chicken farm alarm system by introducing a multi-dimensional sensing unit and a baseline library comparison mechanism. In the technical implementation process, first, the baseline data under the daily operation state is collected by using a perimeter conductive ring, an access lock spring sensing unit, a footstep pressure array, a chicken soundprint array and a feeding trough impedance decoy, and is written into the anti-theft baseline library, thereby providing a standard reference for subsequent data determination. The circuit breaking event, the access lock spring sequence anomaly, the footstep pressure array track, the soundprint disturbance and the impedance touch mark obtained by difference operation can be compared with the anti-theft baseline in time sequence, so that the alarm system no longer relies on the trigger result of a single sensor, but generates an evidence set through the aggregation of multi-source data. Compared with the prior art, the scheme of the present application effectively overcomes the false alarm problem caused by wind blowing, grass movement or chicken activity in the traditional alarm mode, and enhances the accuracy of intrusion detection. Especially based on the response characteristics of the chicken soundprint array, the consistency of the abnormal soundprint disturbance can be verified, and whether there is human interference can be identified, which improves the judgment accuracy of theft and damage behavior, and ensures that the alarm information has sufficient credibility.

[0051] Further, the present application realizes hierarchical early warning and intelligent response control through the joint matching of the evidence set and the feeding order inquiry evidence result. The alarm system can not only generate structured alarm information according to the trigger type identifier, trigger position, monitoring section index and evidence sequence, but also can classify the alarm event as high, medium or low level according to the hierarchical rule table and bind it to a specific monitoring section. In this way, the farm manager can quickly understand the severity and location of the alarm when receiving the alarm information, facilitating targeted disposal. BRIEF DESCRIPTION OF DRAWINGS

[0052] The present application will be further described below in conjunction with the accompanying drawings.

[0053] Figure 1 is a flowchart of an alarm method for a chicken farm based on intelligent monitoring according to the present application.

[0054] Figure 2 is a structural diagram of an alarm system for a chicken farm based on intelligent monitoring according to the present application. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0056] Embodiment one

[0057] Please refer to Figure 1 The present application is an alarm method for a chicken farm based on intelligent monitoring, which comprises:

[0058] S1: generate corresponding baseline from daily data of the perimeter conductive ring, access control lock spring sensing unit, footstep pressure array, chicken voiceprint array and feeding trough impedance decoy, and write into the anti-theft baseline library;

[0059] S2: perform difference operation on real-time data and the anti-theft baseline library to obtain open circuit events, access control lock spring unlocking sequence anomalies, footstep pressure array trajectories, voiceprint disturbances and impedance touch traces, and aggregate them into an evidence set;

[0060] S3: when the primary trigger condition is established, broadcast a feeding order, record the response timing of the chicken voiceprint array, and perform consistency check on the voiceprint disturbances and footstep pressure array trajectories in the evidence set according to the feeding order inquiry evidence result;

[0061] S4: According to the evidence set and the feeding order inquiry result, the intrusion criterion rule is matched to generate an alarm information, and the alarm information includes trigger type identification, trigger position, monitoring section index and evidence sequence;

[0062] S5: According to the hierarchical rule table and combined with the trigger type identification and evidence sequence in the alarm information, the senior, intermediate or low level early warning level is determined, and the monitoring section index is bound;

[0063] S6: According to the early warning level and the monitoring section index, the fence electric lock, the corridor lighting, the buzzer and the communication dialer are driven to execute the access control lock, the lighting guidance, the sound and light prompt and the communication reporting.

[0064] In the step S1, the process of generating corresponding baseline for the daily data of the perimeter conductive ring, the access lock spring sensing unit, the footstep pressure array, the chicken voiceprint array and the feeding trough impedance decoy, and writing into the anti-theft baseline library is as follows:

[0065] Data collection is carried out under the premise of no intrusion. Two relatively stable time periods are selected in this embodiment: one is the daily feeding period (for example, 6:30-7:30 and 16:30-17:30 every day), and the other is the night resting period (for example, 20:00 every day-05:00 the next day). In order to ensure the condition of no intrusion, first, cross verification is carried out through nearly seven days of video playback and historical alarm records to confirm that no one enters the chicken house or the fence in these time periods. Then, according to the unified sampling frequency (for example, 10 times per second), the on-off current value of the perimeter conductive ring, the stress and release displacement of the access lock spring sensing unit, the pressure distribution matrix of the footstep pressure array, the audio signal of the chicken voiceprint array and the resistance-capacitance comprehensive change value of the feeding trough impedance decoy are continuously collected. The collected data is in the format of "time stamp-monitoring section index-channel number-original reading", and is written into the "original data storage area" in real time. If short-time abnormality occurs during the collection process (such as sudden strong wind causing slight swing of the net, mice passing through causing instantaneous jump of a certain pressure array point), only abnormality marking is made without triggering alarm, and original record is preserved for subsequent cleaning.

[0066] After extracting data from the raw data storage, feature processing is performed to form a unified daily feature table. Specifically, for the perimeter conductive ring, the continuous electrical signal is discretized into a "on-off" state sequence according to the current threshold, similar to writing the entire day's door status as a series of "1 / 0" time trajectory; for the access control lock spring sensing unit, the starting time, peak displacement and reset time of the unlocking action are extracted to generate two key fields of "unlocking time sequence" and "strength" to distinguish between normal feeders opening the door and abnormal prying; for the footstep pressure array, the two-dimensional pressure matrix is spliced on the time axis to identify the continuous landing points of the gait and the moving path of the force center, thereby obtaining the "footstep pressure array trajectory segment", which can be understood as a "light-heavy-light" force trace curve left by a person walking on the carpet; for the chicken flock voiceprint array, according to the response law of the chicken flock to the feeding order or fixed environmental sound (such as daily scheduled feeding sound), the "voiceprint response segment" is cut out, and parameters such as main frequency, resonance peak spacing and energy envelope are extracted; for the feeding trough impedance bait, the impedance touch curve caused by hand or beak contact is recorded, including the rising slope, peak impedance and recovery time. Finally, with time label-section index-channel number-on-off state-unlocking time sequence-gait segment-voiceprint segment-impedance curve-abnormality label as unified fields, a structured daily feature table is generated. For example, if there is a normal feeding door opening at 7:05 am in A1 section, No. 03 channel, the record will show: the on-off sequence is continuous "on", the unlocking time sequence has a single peak and rapid reset, the gait segment shows "short step fast frequency" characteristics, the voiceprint segment has obvious group call, and the impedance curve shows short touch followed by rapid recovery.

[0067] The template is aggregated based on the night resting period to obtain a typical and stable resting baseline. First, denoising is performed: short noise segments with a duration less than a set threshold (for example, 1-2 seconds) are removed, and each channel is smoothed with a sliding window to avoid the influence of occasional burrs on the overall judgment. Second, the data is aggregated by time period: the period from 20:00 to 05:00 is divided into several time blocks (for example, each 30 minutes is a block), and the stable mode of each feature is counted. For the perimeter conductive ring, the on-off template during the resting period should maintain a flat shape of "continuous on"; for the access control lock spring, the release sequence template during the resting period should be empty or only have weak fluctuations generated by device self-checking; for the foot pressure array, the gait template is a low-frequency scattered distribution, reflecting the occasional small pressure changes of the chicken flock moving the roost at night, and will not appear "continuous, directional, and heavy" human gait trajectories; for the voiceprint template, the main frequency energy should be significantly lower than that during the day, and there should be no group response peak to the password; for the impedance template, the curve is mainly stable at the baseline, and occasional short touches are occasionally seen. Through this "denoising, blocking, and statistical" method, complex and messy data is summarized as "on-off template, release sequence template, gait template, voiceprint template, and impedance template". For example, the voiceprint template during the period from 02:00 to 02:30 often shows slight fluctuations in background noise without group call peaks, which becomes the standard for judging abnormal voiceprint disturbances at night.

[0068] Finally, the above templates are solidified according to the monitoring section index, channel number, and time period label, and written into the "anti-theft baseline library", while recording the "baseline version label" and "effective identification". In implementation, each baseline is assigned a unique version label (for example, "V2025-09-A1-03-night resting 1"), and the "effective identification" is set to "effective" only after the new template passes quality check. Quality check includes: whether the data coverage meets the standard (such as complete sampling for more than three consecutive days), whether the template stability meets the threshold (such as the variance of each statistical indicator being lower than a set value), and consistency comparison with historical effective baseline (to prevent false updates due to occasional environmental changes). If the check fails, the previous version remains in effect to avoid affecting subsequent real-time comparison and alarm. When writing the library, the "monitoring section index" (such as A1, A2, etc.), "channel number" (such as 03 corresponding to a certain conductive ring or a certain pressure array), and "time period label" (such as "daytime feeding one" and "night resting two") are also associated to ensure that the corresponding reference baseline can be quickly found at any time and location. In this way, when performing real-time difference operation in step S2, the corresponding "on-off, release, gait, voiceprint, and impedance" baseline can be accurately called according to the current time period and specific channel, improving the reliability and explainability of abnormal identification; once a deviation occurs, it can be traced back to the version and parameters used.

[0069] The process of aggregating the circuit breaking events, the door lock spring unlocking sequence abnormalities, the footstep pressure array trajectory, the voiceprint disturbance, and the impedance touch traces into an evidence set in the step S2 is as follows:

[0070] In this embodiment, the real-time data stream from each sensing channel is first classified according to the "monitoring section index" (such as A1, A2, etc.), and then sliced with a unified time window, for example, each window is set to 60 seconds and updated in a rolling manner. In order to reduce clock errors, the same time source is used to correct the timestamps within each section, and linear interpolation is performed on the sampling points as necessary to make the sampling points of different sensors fall within the same window boundary. Subsequently, the templates with the same section and time window label are retrieved from the anti-theft baseline library, including the on-off template, the unlocking sequence template, the gait template, the voiceprint template, and the impedance template, and the real-time data of the window form a one-to-one corresponding data pair. If a window appears for the first time and there is no template, it is returned to the "adjacent window template" or "same period historical template" of the same section, and marked as "low confidence alignment"; when the next round of offline training generates a new template, it is automatically replaced. For example, the real-time slice of A1 section at 20:00-20:01 will be aligned with the baseline template of A1 section at 20:00-20:01, thereby providing a stable reference for subsequent differential comparison.

[0071] After alignment, each type of sensing feature is compared item by item. For the perimeter conductive ring, the real-time "on-off" sequence is compared with the continuous section of the template. If there is a discontinuity that is inconsistent with the template and lasts more than a set threshold (such as 2 seconds), a "on-off deviation marker" is generated. For the unlocking timing of the access control lock spring, the key parameters such as the action start time, peak amplitude, duration, and reset gradient are compared. If any of the indicators deviates from the template tolerance band (such as ±10%) and the combination logic is abnormal, an "unlocking timing deviation marker" is output. For the footstep pressure array, the spatial difference between the force center trajectory and the template trajectory is calculated, and the changes in step length, rhythm, and force peak value are integrated. If the trajectory is continuously offset, strongly directional, and inconsistent with the "nighttime sporadic movement" template, a "pressure array position deviation marker" is output. For the chicken voiceprint array, the main frequency band energy distribution is compared with the template spectrum envelope. If there is a sudden increase in frequency band that is not in the template, a sustained human voice feature, or a misaligned group response timing, it is recorded as a "voiceprint frequency band deviation marker". For the feeding trough impedance bait, the rising slope of the touch curve, the peak value, and the recovery time are compared. If a "human hand long press" type wide peak or repeated touch is formed while the template is "silent", it is marked as an "impedance touch deviation marker". For example, if the pressure array in a window shows "heavy-heavy-heavy" continuous landing points with a step length close to adult gait, while the template is "light scattered points", it is judged as a significant deviation.

[0072] After obtaining various types of deviation markers, they are converted into event or trajectory objects with explicit semantics. For on-off deviation, if the conductive loop appears continuous disconnection and no self-check record, a "circuit breaking event" is generated, and the start and end time and estimated length are given. For access lock spring, if the combination of unlocking parameters is obviously different from the "normal opening door" template (for example, the duration is too long, the reset is delayed or the number of unlocking is abnormal), an "access lock spring unlocking sequence anomaly" is generated, and the suspected means characteristics (such as multiple light attempts) are recorded. For foot pressure array, discrete "pressure array position deviation markers" are connected into continuous "foot pressure array trajectory", and trajectory segment, estimated step, rhythm and direction are given. For example, A2 segment appears continuous eastward migration of force center in 23:15-23:16 window, step is about 0.7 meters, rhythm is stable, and a "suspected walking trajectory" is formed. Normal chicken flock activity will not form such regular trajectory line.

[0073] For the two types of deviation of voiceprint and impedance, "voiceprint disturbance" and "impedance touch trace" objects are generated respectively. Voiceprint disturbance will contain the frequency band range, energy peak, duration, and whether it is suspected human voice or related to feeding order, etc. labels; impedance touch trace contains the number of touches, peak and duration of each touch, which is used to distinguish "short pecking" and "human touch". All objects are bound to "monitor segment index" and "time window number" when created, which is convenient for subsequent cross-source association and tracing. For example: if a continuous significant peak appears near 1.8 kHz in the 22:30-22:31 window of A1 segment, accompanied by harmonic structure unique to chicken flock, a "voiceprint disturbance" is generated, with the judgment of "suspected whistle or human voice"; if two times of high resistance contact of more than 3 seconds are recorded in the trough impedance channel, two "impedance touch traces" are generated. This kind of "sound-touch" concurrency is often seen in the scene of people approaching the trough to explore.

[0074] Finally, with the "monitoring section index" as the primary key, the various types of objects generated in the same section and the same time period in the continuous window are aggregated in chronological order into an "evidence set". The aggregation process includes: removing duplicates (for example, overlapping segments of the same trajectory across two windows are only kept once), association integration (for example, "sound disturbance" and "impedance touch trace" in the same time period are associated as the same chain), and weighted sorting (for example, the weight of the circuit breaking event is higher than that of a single touch trace). The generated evidence set is a sequence unit with a window, clearly showing the evolution relationship from "premonition-trigger-continuation-end", facilitating subsequent steps for verification and early warning. For example, "sound disturbance", "impedance touch trace" and "footstep pressure array trajectory" appear in A1 section from 22:30 to 22:33, and there is no normal feeding order, which should be considered as a continuous suspicious link; the evidence set records this link as a time sequence of "sound, touch, and movement", and presents it together with any possible "circuit breaking event" or "access control anomaly".

[0075] In the step S3, when the primary trigger condition is met, the feeding order is broadcast, the response time sequence of the chicken sound array is recorded, and the consistency check of the sound disturbance and the footstep pressure array trajectory in the evidence set is performed according to the response time sequence, to obtain the feeding order verification result.

[0076] After meeting the primary trigger condition, the loudspeaker in the section is selected according to the "monitoring section index" to perform directional broadcast, avoiding interference with other sections. The feeding order uses fixed and clear Chinese phrases, such as "gather, eat", to reduce ambiguity. A unified timestamp is generated at the same time as the order control instruction is issued, and is recorded as the "order start time". To ensure time consistency between different devices, time calibration is completed before broadcasting, and the "order end time" is recorded when the order broadcast ends. If there are multiple loudspeakers in the section, the one closest to the feeding trough is selected according to the priority table, and the loudspeaker number and volume level used are written in the log. For example, after the "access control lock spring sequence anomaly" occurs in A1 section, the order is broadcast on A1-loudspeaker 01, and the order start time is recorded as 22:31:05, which facilitates subsequent alignment and verification of the chicken's response and footstep pressure array data.

[0077] Specifically, the primary trigger condition is used to determine whether to start the "password challenge". The specific judgment is as follows: first, if a "circuit breaking event" (such as a persistent disconnection of the perimeter conductive ring) is detected, it is established; second, if an "access lock spring unlocking sequence anomaly" (such as a long unlocking time, repeated unlocking and no reset) occurs, it is also established; third, if "footstep pressure array trajectory" and "voiceprint disturbance" are observed at the same time in the same monitoring section and the same time window (which may indicate that someone is approaching while making a sound), it is also considered to be established; fourth, if "footstep pressure array trajectory" and "impedance touch trace" appear at the same time (which may indicate that someone is approaching and touching the feeding trough or fence); fifth, if "voiceprint disturbance" and "impedance touch trace" occur at the same time (which may indicate that someone is making a sound while touching the device). For example, if a regular human gait trajectory appears in the A2 section in the 22:48-22:49 window, and the feeding trough impedance records a continuous 3-second or more high-resistance touch, the primary condition is triggered; then the password is broadcast for challenge. If the chicken group does not respond normally, and the pressure array trajectory continues to move towards the exit, it will be most likely determined to be "inconsistent" in the subsequent steps, providing strong evidence for classification warning and subsequent disposal. Through the above mechanism, the chain of "primary trigger-password challenge-consistency determination-result binding" is closed, which can quickly rule out disturbances caused by normal feeding, and can also identify suspicious intrusion behavior in time.

[0078] After the password is broadcast, the chicken voiceprint array enters the "response listening" state, automatically labels the time when the first obvious response peak appears as the "response start time", and continuously records the response time sequence and intensity change in the next few seconds (such as 10-20 seconds) to form a "response time sequence curve". In order to rule out environmental noise interference, the stability of the main frequency, harmonic structure and energy envelope is also considered in the acoustic characteristics, and only when these indicators meet the threshold value together can the response be confirmed. At the same time, the footstep pressure array module intercepts the same time window data overlapping with the password broadcast, and outputs the footstep pressure array trajectory, including the force center moving path, step length and rhythm. Intuitively, this step is like "listening to the sound and looking at the reaction" at the same time "looking at the ground and looking at the footstep": if the chicken group gathers due to the password, the pressure array graph will appear multiple light convergence points near the feeding trough; if someone approaches, the trajectory will show a regular, directional and heavy path. For example, at 22:31:07, the voiceprint array detects a group of short response peaks, and the pressure array appears scattered points in front of the feeding trough, which is usually considered as a normal feeding response.

[0079] The voiceprint disturbance is aligned with the time window of the footstep pressure array track based on the password starting time. If there is a slight drift in the timestamps of the two data sources, the drift is corrected within the allowed tolerance range (e.g. ±2 seconds). Then, it is verified whether the two types of data come from the same "monitoring section index" to avoid the cross situation that the password is announced in A1 section and the response comes from A2 section. After alignment, the "consistency determination result" is given based on the timing matching degree, spatial aggregation degree and behavior pattern matching degree, which can be divided into "consistent", "inconsistent" and "insufficient evidence". For example, if the chicken-specific response peak appears within 2-5 seconds after the password, and the pressure array forms a multi-point slight aggregation around the feeding trough, and both come from the same section where the password is announced, it is determined as "consistent"; otherwise, if the voiceprint shows long-term human voice characteristics and the pressure array shows a single path quickly passing through the feeding trough area, it is determined as "inconsistent".

[0080] After obtaining the consistency determination, the "feeding password inquiry evidence result" is generated, which includes at least: monitoring section index, password starting and ending time, response starting time and response timing summary, footstep pressure array track summary (such as step length, rhythm, aggregation position), consistency level, and associated evidence sequence number. The evidence sequence number is used to point to the "break event, access lock spring unlocking sequence anomaly, footstep pressure array track, voiceprint disturbance and resistance touch trace" formed in step S2, realizing the two-way tracing of "password inquiry" and "original evidence". A brief explanation is also attached: if it is "consistent", it means that the current behavior is closer to the normal feeding mode; if it is "inconsistent", it indicates that there is abnormal intervention; if it is "insufficient evidence", it is recommended to extend the listening time or rebroadcast the password for secondary verification. For example, the inquiry evidence result of A1 section shows "consistent" and binds evidence numbers A1-20250915-223105-01 (voiceprint) and A1-20250915-223105-02 (pressure array), which can be directly referenced in the subsequent rule matching stage.

[0081] In step S4, according to the evidence set and the feeding password inquiry evidence result, the intrusion criterion rule is matched to generate an alarm information, and the alarm information includes trigger type identification, trigger position, monitoring section index and evidence sequence. The process is as follows:

[0082] In this embodiment, the evidence set formed in step S2 (including circuit breaking events, access lock spring unlocking sequence abnormalities, footstep pressure array trajectories, voiceprint disturbances, and impedance touch traces) and the feeding password query evidence results output in step S3 are strictly aligned with the unified primary key of the monitoring section index + the time window number. The alignment strategy is: first match the section index, and then match the time window; if there is a slight deviation in the window boundary, merge within the allowed tolerance (such as ±2 seconds). After alignment, the two types of information are combined into an evidence mapping table, where each row corresponds to a specific section and window, and the fields include: window timestamp, evidence item list, password consistency level (such as consistent / inconsistent / insufficient evidence), and the numbers pointing to each other. For example, the A1 section in the 22:31:00-22:31:30 window mapping row contains the summary of “footstep pressure array trajectory, voiceprint disturbance, and impedance touch trace”, and also contains the conclusion of “password consistency = inconsistent”, facilitating subsequent one-stop comparison and determination.

[0083] The feeding password query evidence results of the window are read first to select the corresponding branch in the intrusion criterion rule library. The rule library at least includes three types: “password consistent branch”, “password inconsistent branch”, and “insufficient evidence branch”: when consistent, it is biased towards normal feeding or chicken gathering scenarios; when inconsistent, it enters strict criteria, focusing on the combination of fields such as “human gait”, “abnormal touch”, “access anomaly”, and “perimeter circuit breaking”; when the evidence is insufficient, a conservative strategy is adopted and further observation is prompted. After positioning the branch, the matching field set required by the branch is automatically loaded, such as “whether there is a continuous human gait trajectory”, “whether there is a long period of high resistance touch”, “whether the access unlocking exceeds the threshold”, “whether the perimeter is continuously disconnected”, “whether the voiceprint contains human voice feature peaks”, etc., and each field is accompanied by a threshold and a weight. In this way, the positioning of the rule branch is like choosing a “special ruler”, and this ruler will be used to measure the evidence in the mapping table item by item.

[0084] After determining the rule branch, the fields of the “evidence mapping table” are compared item by item. If the on-off deviation of the perimeter conductive ring reaches the continuous threshold, it generates “trigger type identification = circuit breaking”; if the access unlocking duration exceeds the limit or there are multiple abnormal attempts, it is identified as “access anomaly”; if the pressure array trajectory presents regularity, obvious directionality, and step length close to the adult value range, accompanied by the appearance of human voice feature peaks in the voiceprint or long press impedance curve, it is classified as “suspected intrusion”. When multiple conditions occur simultaneously, the main trigger type is output according to the priority strategy and the secondary label is retained to ensure clear meaning without conflict. The trigger position is determined according to the “monitoring section index” combined with the geographical mapping relationship of the section, such as the A1 section corresponding to the “eastern segment of the north fence”. For example: in the A1 section 22:31 window, “human gait + long press impedance + password inconsistency” appears, the main trigger type is identified as “suspected intrusion”, and the trigger position is located at “A1 – eastern segment of the north fence”.

[0085] The evidence entries of the same section at continuous time are then time-sequenced to form an "evidence sequence". The arrangement rules are: sorting by window sequence; splicing the tracks that continue across windows; chain binding of "voiceprint-touch-gait" entries that point to each other; and deduplication and fusion of repeated or overlapping entries. The head of the sequence is usually a precursor (such as a weak voiceprint disturbance at a remote end), the middle section is the main behavior (such as a stable human gait approaching and a long touch), and the tail may be a result (such as an abnormal access control or perimeter break). The overall evidence sequence is bound to the "trigger type identification" to form a clear chain of "type-process-result". Taking A1 as an example, "voiceprint disturbance, footstep pressure array track, impedance long press, and access control anomaly" occur at 22:30-22:33, which are arranged into a complete sequence and saved together with the "suspicious intrusion" trigger type for subsequent review and review.

[0086] Finally, the key elements are integrated into structured alarm information. The information contains at least four items:

[0087] I. Trigger type identification (such as break, access control anomaly, and suspicious intrusion);

[0088] II. Trigger location (mapped from the section to a specific position, such as "A1 - north fence east section");

[0089] III. Monitoring section index (such as A1);

[0090] IV. Evidence sequence (evidence entries arranged by time window and their numbers).

[0091] For ease of understanding, the alarm information is output in a structured field, which can be automatically processed by the platform and quickly interpreted by humans. Example:

[0092] Trigger type identification: suspicious intrusion;

[0093] Trigger location: A1 - north fence east section;

[0094] Monitoring section index: A1;

[0095] Evidence sequence: [22:30 - voiceprint disturbance (number A1-223000-01); 22:31 - footstep pressure array track (number A1-223100-02); 22:31 - impedance touch trace (number A1-223100-03, long press 3 seconds); 22:32 - access control lock spring release sequence anomaly (number A1-223200-04)].

[0096] In the step S5, the process of determining the high, medium, or low warning level according to the hierarchical rule table and combining the trigger type identification and evidence sequence in the alarm information, and binding the monitoring section index is:

[0097] In this embodiment, the hierarchical module first receives the structured alarm information generated in step S4, and performs field parsing thereon. Three types of key input items are extracted in a predetermined order: first, a "trigger type identifier" for indicating the main cause of this alarm (such as a circuit break, an access control anomaly, a suspected intrusion, etc.); second, an "evidence sequence", which is a list of evidence items arranged in a time window (including evidence number, source channel, start and end time, intensity index, and confidence level); and third, a "monitoring section index" for locking the physical location where the alarm is located. To avoid misjudgment across sections, the parsing stage will check whether the section labels of all items in the "evidence sequence" are consistent with the section index in the alarm information; if there are cross-section items, they will be marked as "manual review required" or "evidence excluded" before entering the classification. For example, if the alarm information shows that the trigger type identifier is "suspected intrusion", the monitoring section index is A1, and the evidence sequence is "22:30 sound disturbance, 22:31 footstep pressure array track, 22:31 impedance long press, 22:32 access control anomaly" in turn, the hierarchical module will cache these four evidences in time sequence together with section A1 as input for subsequent rule comparison.

[0098] After parsing, the "trigger type identifier" is used as a search key to locate the corresponding rule item in the "hierarchical rule table". The rule table shows the required evidence verification requirements and warning level mapping relationship in rows and columns. Taking "suspected intrusion" as an example, its rule item usually requires: there is a continuous human gait track or equivalent strong evidence; at least one contact type evidence (such as impedance long press reaching threshold) is concurrent; within a limited time window, a causal chain is formed with soundprint anomaly or access control anomaly. This item also provides mapping from "satisfying all conditions, high-level warning" to "satisfying some key conditions, medium-level warning" to "only satisfying weak conditions, low-level warning". For "circuit break" triggers, the rule item will emphasize more on the duration and impact range; for "access control anomaly" triggers, it will pay more attention to the parameters such as undocking duration, repeated attempt times, and reset failure, etc. After locating the item, the corresponding numerical threshold (such as continuous step interval, touch duration lower limit, time window merging width) and weight coefficient will be loaded simultaneously, serving as a benchmark for subsequent item-by-item verification.

[0099] In this step, the evidence sequence is checked item by item for temporal consistency according to the positioned rule items. First, the "time consistency" check is performed: based on the start and end time of the evidence, it is judged whether the multi-source evidence appears continuously within the specified merging window (for example, the interval between any two key evidence start and end is not more than 90 seconds, and the overall link is not interrupted by a threshold of empty window period). Second, the spatial consistency check is performed: the monitoring segment index and channel number of the evidence should be consistent, and if the trajectory crosses to the adjacent segment, it needs to appear in the same direction and meet the "cross-segment intrusion" additional condition to be counted. Third, the "intensity consistency" test is performed: for example, the stride and rhythm fall into the adult gait interval, the impedance touch duration reaches the long press criterion, the human voice feature peak is significant and the difference with the chicken response is obvious, the access control unlocking duration exceeds the limit, etc. The above verification results are summarized as a comprehensive score according to the weight, and then the warning level is determined according to the mapping given by the rule item. For example: if the "human gait trajectory, impedance long press ≥ 3 seconds, access control anomaly" appears continuously in the A1 segment within three minutes, and the feeding order inquiry evidence result is "inconsistent", it meets the strong cause and effect chain and intensity criterion, and is directly judged as "high-level warning"; if only "human gait trajectory + voice disturbance" appears and the duration is short, it can be judged as "medium-level warning"; if only single voice disturbance occurs without other support, it is "low-level warning". In the case of insufficient evidence or time sequence fragmentation, the level will be lowered, and a "suggestion to continue observation" prompt will be attached.

[0100] After the level determination is completed, a grading record is generated, and the "warning level" and "monitoring segment index" are strongly bound to ensure that the subsequent disposal link can quickly locate the specific area and equipment. The grading record is output in a structured field, which at least includes: monitoring segment index, warning level (high / medium / low), trigger type identification, evidence sequence details (listed according to time window), verification point summary (for example, "time consistency: passed; spatial consistency: passed; intensity consistency: gait is significant, long press meets the standard, unlocking exceeds the limit"), determination basis version (corresponding to the rule table version number) and generation time. Still taking A1 as an example, the grading record can be presented as: the segment is A1; the warning level is high; the trigger type is suspicious intrusion; the evidence sequence is [22:30 voice disturbance number A1-01, 22:31 footstep pressure array trajectory number A1-02, 22:31 impedance long press number A1-03, 22:32 access control anomaly number A1-04]; the verification points are "form a complete link within 180 seconds, and the intensity and time sequence both meet the high-level threshold"; the rule table version is "V2025-09"; the generation time is "2025-09-15 22:33". The grading record will be directly input to the subsequent step S6, driving the corresponding access control locking, lighting guidance, sound and light prompt, and communication reporting actions, realizing smooth connection and closed-loop management from "alarm generation" to "disposal execution".

[0101] Example Two

[0102] Referring to Figure 2 As shown in the drawings, the application further provides an alarm system based on intelligent monitoring of a chicken farm, comprising:

[0103] A baseline modeling module is configured to generate corresponding baselines for daily data of the perimeter conductive ring, the access control lock spring sensing unit, the footstep pressure array, the chicken soundprint array, and the feed trough impedance decoy, and write the baselines into a theftproof baseline library.

[0104] An anomaly extraction module is configured to perform difference operation on real-time data and the theftproof baseline library to obtain a circuit breaking event, an access control lock spring unlocking sequence anomaly, a footstep pressure array trajectory, a soundprint disturbance, and an impedance touch trace, and aggregate the above into an evidence set.

[0105] A password inquiry module is configured to broadcast a feeding password when a primary trigger condition is met, record a response timing of the chicken soundprint array, and perform consistency check on the soundprint disturbance and the footstep pressure array trajectory in the evidence set according to the response timing to obtain a feeding password inquiry result.

[0106] A rule matching module is configured to match an intrusion criterion rule according to the evidence set and the feeding password inquiry result, generate an alarm information, and the alarm information contains a trigger type identifier, a trigger location, a monitoring section index, and an evidence sequence.

[0107] A pre-warning grading module is configured to determine a high, medium or low pre-warning level according to a grading rule table and in combination with the trigger type identifier and the evidence sequence in the alarm information, and bind the monitoring section index.

[0108] A linkage execution module is configured to drive a fence electric lock, a corridor lighting, a buzzer, and a communication dialer according to the pre-warning level and the monitoring section index to execute access control locking, lighting guidance, sound and light prompting, and communication reporting.

[0109] The above describes one embodiment of the application in detail, but the content described is only a preferred embodiment of the application and cannot be considered as limiting the scope of the implementation of the application. Any equivalent changes and improvements made according to the scope of the application should still belong to the patent coverage of the application.

Claims

1. An alarm method for monitoring a chicken farm based on intelligence, characterized in that, The method comprises the following steps: S1: generate baseline corresponding to daily data of the perimeter conductive ring, access control lock spring sensing unit, footstep pressure array, chicken voiceprint array and feeding trough impedance decoy, and write into the anti-theft baseline library; S2: perform difference operation on real-time data and the anti-theft baseline library to obtain a circuit breaking event, access control lock spring unlocking sequence anomaly, footstep pressure array trajectory, voiceprint disturbance and impedance touch trace, and aggregate them into an evidence set; S3: when the primary trigger condition is established, broadcast a feeding order, record the response timing of the chicken voiceprint array, and perform consistency check on the voiceprint disturbance and footstep pressure array trajectory in the evidence set according to the feeding order inquiry result; the specific process is as follows: After determining that the primary trigger condition is established, broadcast the feeding order through the loudspeaker according to the monitoring section index, and record the order starting time; The chicken voiceprint array records the response starting time and response timing, and the footstep pressure array synchronously intercepts the footstep pressure array trajectory in the same time window; Align the time windows of the voiceprint disturbance and footstep pressure array trajectory based on the order starting time, verify the consistency of the monitoring section, and generate a consistency determination result; According to the consistency determination result and the monitoring section index, the feeding order inquiry result is formed, and the related evidence sequence number is bound; S4: according to the evidence set and the feeding order inquiry result, match the intrusion criterion rule to generate an alarm information, which contains trigger type identification, trigger position, monitoring section index and evidence sequence; the specific process of generating the alarm information is as follows: Align and merge the evidence set and the feeding order inquiry result into an evidence mapping table according to the monitoring section index and the time window; According to the feeding order inquiry result, locate the corresponding rule branch and its matching field in the intrusion criterion rule; According to the evidence mapping table, compare the rule fields one by one to generate the trigger type identification, and determine the trigger position according to the monitoring section index; Arrange the evidence items of the monitoring section in the time window order to generate the evidence sequence and associate the trigger type identification; Integrate the trigger type identification, trigger position, monitoring section index and evidence sequence to generate a structured alarm information; S5: according to the hierarchical rule table and combined with the trigger type identification and evidence sequence in the alarm information, determine the high, medium or low warning level, and bind the monitoring section index; S6: according to the warning level and the monitoring section index, drive the fence electric lock, corridor lighting, buzzer and communication dialer to perform access control locking, lighting guidance, sound and light prompt and communication reporting.

2. The alarm method based on intelligent monitoring of a chicken farm according to claim 1, characterized in that, In step S1, the process of generating baseline corresponding to daily data of the perimeter conductive ring, access control lock spring sensing unit, footstep pressure array, chicken voiceprint array and feeding trough impedance decoy, and writing into the anti-theft baseline library is as follows: During the daily feeding and night resting period when there is no intrusion, continuously collect perimeter conductive ring data, access control lock spring sensing unit data, footstep pressure array data, chicken voiceprint array data and feeding trough impedance decoy data, and archive them to the original data storage area; Extract the data from the original data storage area to generate on-off state sequence, access control lock spring unlocking timing, footstep pressure array trajectory segment, chicken voiceprint response segment and feeding trough impedance touch curve, and generate a daily feature table in a unified field format; Based on the nocturnal quiet period, short noise segments are removed, and daily features are aggregated by period to obtain on-off templates, unlatching sequence templates, gait templates, voiceprint templates, and impedance templates; According to the monitoring section index, channel number, and period label, each template is solidified into the corresponding baseline and written into the anti-theft baseline library. The baseline version label and effective identification are recorded.

3. The alarm method based on intelligent monitoring of a chicken farm according to claim 1, characterized in that, In step S2, the process of performing difference operation on real-time data and anti-theft baseline library to obtain circuit breaking events, access lock spring unlatching sequence abnormalities, footstep pressure array trajectories, voiceprint disturbances, and impedance touch traces, and aggregating them into evidence sets is as follows: Segment the real-time data according to the monitoring section index and the unified time window, and retrieve the same section and time window templates in the anti-theft baseline library to form a one-to-one aligned data pair; Perform difference operation on the aligned data pair to output on-off deviation markers, unlatching time sequence deviation markers, pressure array position deviation markers, voiceprint frequency band deviation markers, and impedance touch deviation markers; Generate circuit breaking events based on on-off deviation markers, generate access lock spring unlatching sequence abnormalities based on unlatching time sequence deviation markers, and generate footstep pressure array trajectories based on pressure array position deviation markers; Generate voiceprint disturbances based on voiceprint frequency band deviation markers, generate impedance touch traces based on impedance touch deviation markers, and bind the monitoring section index and the time window number; Aggregate circuit breaking events, access lock spring unlatching sequence abnormalities, footstep pressure array trajectories, voiceprint disturbances, and impedance touch traces in time window order to generate evidence sets using the monitoring section index as the primary key.

4. The alarm method based on intelligent monitoring of a chicken farm according to claim 1, characterized in that, The specific content of determining that the primary trigger condition is met is as follows: When a circuit breaking event or an access lock spring unlatching sequence abnormality occurs, it is determined that the primary trigger condition is met.

5. The alarm method based on intelligent monitoring of a chicken farm according to claim 1, characterized in that, In step S5, the process of determining the senior, intermediate, or junior warning level according to the hierarchical rule table and combining the trigger type identifier and evidence sequence in the alarm information, and binding the monitoring section index is as follows: Parse the alarm information to extract the trigger type identifier, evidence sequence, and monitoring section index as input items for hierarchical determination; Locate the corresponding rule item in the hierarchical rule table according to the trigger type identifier to obtain the evidence verification requirements and warning level mapping of the item; According to the rule item, verify each item in the evidence sequence to confirm the consistency of the evidence and the time window, and determine the senior, intermediate, or junior warning level; Bind the warning level and the monitoring section index, and generate a hierarchical record together with the trigger type identifier and the evidence sequence.

6. An alarm system based on intelligent monitoring of a chicken farm, applied to the alarm method based on intelligent monitoring of a chicken farm in any one of claims 1-5, characterized in that, It includes: Baseline modeling module for generating corresponding baselines from the daily data of the perimeter conductive ring, access lock spring sensing unit, footstep pressure array, chicken voiceprint array, and feeding trough impedance decoy, and writing them into the anti-theft baseline library; Abnormality extraction module for performing difference operation on real-time data and anti-theft baseline library to obtain circuit breaking events, access lock spring unlatching sequence abnormalities, footstep pressure array trajectories, voiceprint disturbances, and impedance touch traces, and aggregating them into evidence sets; Password inquiry module for broadcasting a feeding password when the primary trigger condition is met, recording the response time sequence of the chicken voiceprint array, and performing consistency verification on the voiceprint disturbances and footstep pressure array trajectories in the evidence set according to the response time sequence to obtain a feeding password inquiry result; The specific process is as follows: After determining that the primary trigger condition is met, the feeding password is broadcasted through the speaker according to the monitoring section index, and the password start time is recorded; The response start time and response timing are recorded by the chicken flocks' voiceprint array, and the foot pressure array trajectory in the same time window is synchronously intercepted; The time window of the voiceprint disturbance and the foot pressure array trajectory is aligned based on the password start time, the consistency of the monitoring section is verified, and a consistency determination result is generated; According to the consistency determination result and the monitoring section index, a feeding password inquiry result is formed, and the related evidence sequence number is bound; The rule matching module is used to match the intrusion criterion rule according to the evidence set and the feeding password inquiry result, generate an alarm information, and the alarm information includes trigger type identification, trigger position, monitoring section index and evidence sequence; The specific process of generating the alarm information is as follows: According to the monitoring section index and the time window, the evidence set and the feeding password inquiry result are aligned and merged into an evidence mapping table; According to the feeding password inquiry result, the corresponding rule branch and its matching field in the intrusion criterion rule are located; According to the evidence mapping table, the rule fields are compared item by item, the trigger type identification is generated, and the trigger position is determined according to the monitoring section index; According to the time window order, the evidence items of the monitoring section are arranged, the evidence sequence is generated and the trigger type identification is associated; The trigger type identification, trigger position, monitoring section index and evidence sequence are integrated to generate a structured alarm information; The early warning grading module is used to determine the high, medium or low early warning level according to the grading rule table and the trigger type identification and evidence sequence in the alarm information, and bind the monitoring section index; The linkage execution module is used to drive the fence electric lock, corridor lighting, buzzer and communication dialer according to the early warning level and monitoring section index, and execute access control locking, lighting guidance, sound and light prompt and communication reporting.

Images