An intelligent building security control system and control method based on big data analysis

Abstract

The present invention discloses an intelligent building security control system and control method based on big data analysis, relating to the field of big data analysis technology. During operation, the system enables real-time monitoring and analysis of key data such as the flow, density, location, and dwell time of people within a building. Through comprehensive data acquisition and sophisticated data preprocessing by a sensor group, it can extract detailed personnel flow characteristics, establish a Kalman filter model for trajectory prediction, and obtain an anomaly assessment index Dyc for personnel intersections and dwell times. The decision-making module further generates a matching state response strategy based on the anomaly assessment index, ensuring timely and effective personnel, broadcast, and equipment response measures in abnormal situations. This significantly enhances the intelligent level of building security and effectively addresses issues such as data lag, untimely response, and monitoring blind spots in traditional security systems, providing a more efficient and secure building environment management solution.

Description

Technical Field

[0001] The present invention relates to the field of building security technology, and in particular to an intelligent building security control system and control method based on big data analysis. Background Art

[0002] Existing building security systems typically rely on preset rules and simple sensor data feedback. These systems often exhibit slow response and lack of accuracy when faced with complex scenarios such as rapid movement, gatherings, and noise. For example, when the flow of people suddenly increases or the ambient noise rises sharply, traditional security systems may not be able to detect and respond accordingly in time, resulting in potential safety hazards. In addition, when processing multi-sensor data, existing systems lack effective data fusion and in-depth analysis capabilities, and are unable to comprehensively and accurately assess and predict security risks. These shortcomings make existing security systems limited in practical applications and difficult to cope with complex and changing security challenges.

[0003] The shortcomings of existing security systems mainly stem from their reliance on simple rules and the limitations of limited data analysis. In scenarios where people move and gather quickly, traditional systems find it difficult to respond and adjust quickly, resulting in delayed security measures. For example, when a large number of people gather in a building, the system cannot identify and evacuate the crowd in a timely manner, which may cause congestion and stampedes. In the case of a sharp increase in environmental noise, it is difficult to accurately identify the source of the noise and take appropriate measures. For example, adding security personnel or issuing evacuation alarms not only increases safety risks, but may also lead to property damage and casualties.

[0004] Therefore, how to carry out intelligent control of building security is a technical problem that technical personnel need to solve at present. Summary of the Invention

[0005] In response to the deficiencies of the prior art, the present invention provides an intelligent building security control system and control method based on big data analysis, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent building security control system and control method based on big data analysis, including a data acquisition module, a data preprocessing module, a trajectory prediction module and a decision module;

[0007] The data acquisition module collects data on personnel flow, density, location and dwell time through a sensor group placed in the building to form a collection data set;

[0008] The data preprocessing module preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stay time features to form a feature vector X;

[0009] The trajectory prediction module uses the feature vector X to establish a Kalman filter model, obtains the number of personnel intersection points and the number of personnel intersection stay time points by training the Kalman filter model, and obtains the position point abnormality evaluation index Dyc after synchronous fitting;

[0010] The decision module evaluates the personnel flow and stay matching status results using the location point anomaly assessment index Dyc, and generates a matching status response strategy plan based on the personnel flow and stay matching status results, and simultaneously performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response, and device response.

[0011] Preferably, the data acquisition module includes a data collection unit and a data transmission unit;

[0012] The data collection unit collects data through a sensor group placed in the building. The sensor group includes a camera sensor, an infrared sensor, a motion sensor, and a positioning sensor. The sensor group collects data on the flow of people, the density of people in a fixed area, the location of people, and the time people stay in the building in real time to form a collection data set.

[0013] The data transmission unit compresses and packages the collected data set by using the Internet of Things, wireless network and Ethernet, and simultaneously transmits the packaged and compressed collected data set to the data preprocessing module;

[0014] Compression includes compressing the collected data set using the data compression algorithm Zlib and the data compression algorithm Gzip, marking the compressed collected data set with a timestamp and a data source identifier, and then packaging and sending it.

[0015] Preferably, the data preprocessing module includes a data filtering unit and an extraction unit;

[0016] The data filtering unit preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, wherein the data cleaning preprocessing includes noise detection, outlier detection and filling missing values, specifically including using standard deviation method preprocessing, box plot method and isolation forest method, mean filling method, median filling method and nearest neighbor interpolation method for specific processing; data normalization processing includes using maximum and minimum normalization processing and Z-Score normalization method to convert data of different dimensions into the same range;

[0017] The extraction unit then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stop time features to form a feature vector X, specifically X={v, , P(x, y), Ts};

[0018] Among them, the personnel flow characteristics include calculating the personnel's moving speed v, through The formula is calculated and obtained, where v represents the moving speed of the personnel, Indicates the time interval The distance moved within Indicates a time interval;

[0019] The population density feature includes calculating the population density ,pass The formula is calculated and obtained, where represents the density of people, N represents the number of people in the detection area, and a represents the area of ​​the detection area;

[0020] Personnel position features include extracting the coordinate position P(x, y) of the person, where x and y represent the horizontal and vertical positions of the person in the detection area, respectively.

[0021] The personnel dwell time feature includes calculating the personnel dwell time Ts, through The formula is calculated and obtained, where Indicates the time when the personnel moves to the stopping position, Indicates the time when the person leaves the stopping position.

[0022] Preferably, the data preprocessing module includes a data filtering unit and an extraction unit;

[0023] The data filtering unit preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, wherein the data cleaning preprocessing includes noise detection, outlier detection and filling missing values, specifically including using standard deviation method preprocessing, box plot method and isolation forest method, mean filling method, median filling method and nearest neighbor interpolation method for specific processing; data normalization processing includes using maximum and minimum normalization processing and Z-Score normalization method to convert data of different dimensions into the same range;

[0024] The extraction unit then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stop time features to form a feature vector X, specifically X={v, , P(x, y), Ts};

[0025] Among them, the personnel flow characteristics include calculating the personnel's moving speed v, through The formula is calculated and obtained, where v represents the moving speed of the personnel, Indicates the time interval The distance moved within Indicates a time interval;

[0026] The population density feature includes calculating the population density ,pass The formula is calculated and obtained, where represents the density of people, N represents the number of people in the detection area, and a represents the area of ​​the detection area;

[0027] Personnel position features include extracting the coordinate position P(x, y) of the person, where x and y represent the horizontal and vertical positions of the person in the detection area, respectively.

[0028] The personnel dwell time feature includes calculating the personnel dwell time Ts, through The formula is calculated and obtained, where Indicates the time when the personnel moves to the stopping position, Indicates the time when the person leaves the stopping position.

[0029] Preferably, the location point abnormality assessment index Dyc is obtained by the following calculation formula:

[0030] ;

[0031] Where, represents the location of the i-th trajectory intersection point, Represents the fitting value of the intersection position, specifically by fitting all the trajectory intersection positions Perform fitting to obtain the center point trend and center representative position, Represents the standard deviation of the intersection position, specifically by counting the intersection positions of all trajectories The distribution of , calculate the standardization to obtain, Indicates the length of time people stay at the i-th intersection, It represents the fitting value of the length of time people stay at the intersection, by calculating the length of time people stay at all intersections Perform fitting to obtain the center point trend and representative value, It represents the standard deviation of the length of time people stay at the intersection, specifically by counting the length of time people stay at all intersections The distribution of is obtained by calculating the standard deviation, where n represents the total number of intersection locations.

[0032] Preferably, the modified state vector The predicted trajectory is iteratively corrected through the following training steps: Step 1: Define the state vector of the Kalman filter model :

[0033] ;

[0034] Where, and They represent the position coordinates of the personnel at time t, and They represent the speed of the person at time t;

[0035] Step 2: Create a state vector Transfer equation:

[0036] ;

[0037] Where A represents the state vector Transfer matrix, B represents the control matrix, represents the control input, represents process noise, and the state vector The transfer matrix A is as follows, which is used to describe the state vector The state change relationship from time t to time t+1 is:

[0038] ;

[0039] Where, Indicates a time interval;

[0040] Step 3: Establish the observation equation:

[0041] ;

[0042] Where H represents the observation matrix, Represents measurement noise, specifically the random error in the observation process. The observation matrix H is as follows: the observation matrix H transforms the state vector Mapped to the observation space,

[0043] ;

[0044] Step 4: Establish the prediction step, including establishing the prediction prior state vector Estimate and predict the a priori error covariance:

[0045] Predict the prior state vector estimate:

[0046] ;

[0047] Where, represents the prior state vector, represents the posterior state vector estimate at time t-1;

[0048] Forecast a priori error covariance:

[0049] ;

[0050] Where, Represents the prior error covariance matrix, specifically the prediction of the error covariance at time t, reflecting the predicted state vector uncertainty, represents the posterior error covariance matrix at time t-1, specifically the error covariance corrected after combining the observed data at time t-1, Q represents the process noise covariance matrix, specifically the covariance of the process noise, reflecting the system in the state vector uncertainty in the transfer process;

[0051] Step 5: Establish the update step and calculate the Kalman gain:

[0052] ;

[0053] Where, represents the Kalman gain, specifically the weight value used to balance the predicted value and the observed value, and R represents the measurement noise covariance matrix, specifically the covariance of the observation noise, reflecting the uncertainty in the observation process.

[0054] Preferably, step five: establishing an update step further includes estimating the updated posterior state vector and updating the posterior error covariance:

[0055] Updated posterior state vector estimate:

[0056] ;

[0057] Where, Represents the updated posterior state vector estimate, specifically the state vector corrected after combining the observed data estimate, Represents the prior state vector, specifically the state vector at time t The predicted value of is not corrected with the observed data;

[0058] Update the posterior error covariance:

[0059] ;

[0060] Where, Represents the updated posterior error covariance matrix, specifically the error covariance corrected after combining the observed data, reflecting the uncertainty of the posterior state vector estimate.

[0061] Preferably, the decision module includes a matching unit and an execution unit;

[0062] The matching unit evaluates the personnel flow and stay matching status results by comparing the location point abnormality evaluation index Dyc with the preset abnormality evaluation matching threshold PG, and generates a matching status response strategy scheme for the personnel flow and stay matching status results according to the matching status results;

[0063] The execution unit performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response and device response.

[0064] Preferably, the matching status response strategy solution is obtained by the following matching method:

[0065] When the location point abnormality evaluation index Dyc is less than the abnormality evaluation matching threshold PG, the personnel flow and stay matching status is obtained without abnormal results, and the matching status is generated without executing the response strategy plan;

[0066] When the location point abnormality assessment index Dyc ≥ the abnormality assessment matching threshold PG, the abnormal results of the personnel flow and stay matching status are obtained, and a matching status execution response strategy plan is generated, including notifying relevant personnel to go for maintenance, adjusting the operating status of environmental equipment, and increasing the monitoring detection frequency of the abnormal location point to 150% of the original detection frequency;

[0067] When the location point anomaly assessment index Dyc ≥ twice the anomaly assessment matching threshold PG, obtain the personnel flow and stay matching status warning results, generate the matching status warning execution response strategy plan, including adjusting the access control status, turning on the broadcast notification, dispatching relevant personnel to deal with the situation, and adjusting the operating status of the environmental equipment.

[0068] An intelligent building security control system and control method based on big data analysis, comprising the following steps:

[0069] Step 1: The data collection module collects data on personnel flow, density, location, and dwell time through a sensor group placed in the building to form a collection data set;

[0070] Step 2: The data preprocessing module preprocesses the collected data set, including data cleaning and normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features, and personnel dwell time features to form the feature vector X;

[0071] Step 3: The trajectory prediction module uses the feature vector X to establish a Kalman filter model. By training the Kalman filter model, the number of intersection points and the number of intersection duration points are obtained. After simultaneous fitting, the location point anomaly assessment index Dyc is obtained.

[0072] Step 4: The decision module evaluates the personnel flow and stay matching status results using the location point anomaly assessment index Dyc, and generates a matching status response strategy plan based on the personnel flow and stay matching status results. It also performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response, and device response.

[0073] The present invention provides an intelligent building security control system and control method based on big data analysis, which has the following beneficial effects:

[0074] (1) When the system is running, it realizes real-time monitoring and analysis of key data such as personnel flow, density, location and dwell time in the building. Through comprehensive data collection and sophisticated data preprocessing of the sensor group, it can extract detailed personnel flow characteristics, establish a Kalman filter model for trajectory prediction, and obtain the abnormal evaluation index Dyc of personnel intersection and stay. The decision module further generates a matching state response strategy based on the abnormal evaluation index to ensure that timely and effective personnel, broadcast and equipment response measures are taken in abnormal situations, significantly improving the intelligent level of building security. It can also effectively solve the problems of data lag, untimely response and monitoring blind spots in traditional security systems, providing a more efficient and safer building environment management solution.

[0075] (2) The trajectory prediction is combined with the Kalman filter model to achieve accurate monitoring of the flow and stay of people in the building. By training the feature vector X, the position and speed of people are tracked and recorded in real time, and the predicted intersection point and stay time of people are obtained and corrected, thus forming a predicted trajectory. By calculating the abnormal evaluation index Dyc of the trajectory intersection point position and the stay time of people, abnormal situations are effectively identified. By iteratively correcting the predicted trajectory and combining the state vector transfer equation and observation equation of the Kalman filter model, the accuracy of the prediction and the timeliness of the response are significantly improved.

[0076] (3) By comparing the location point anomaly assessment index Dyc with the preset anomaly assessment matching threshold PG to evaluate the personnel flow and stay matching status results, and generating a matching status response strategy based on the personnel flow and stay matching status results, this multi-level response strategy can not only deal with potential safety hazards in a timely manner and improve the safety and efficiency of building management, but also provide rapid and effective early warning and emergency response when the problem is serious, ensuring the safety of people and property in the building. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is a block diagram and flow chart of an intelligent building security control system based on big data analysis of the present invention;

[0078] Figure 2This is a schematic diagram of the steps of an intelligent building security control method based on big data analysis of the present invention. DETAILED DESCRIPTION

[0079] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0080] Example 1

[0081] The present invention provides an intelligent building security control system and control method based on big data analysis. Figure 1 , including data acquisition module, data preprocessing module, trajectory prediction module and decision module;

[0082] The data acquisition module collects data on personnel flow, density, location and dwell time through a sensor group placed in the building to form a collection data set;

[0083] The data preprocessing module preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stay time features to form a feature vector X;

[0084] The trajectory prediction module uses the feature vector X to establish a Kalman filter model, obtains the number of personnel intersection points and the number of personnel intersection stay time points by training the Kalman filter model, and obtains the position point abnormality evaluation index Dyc after synchronous fitting;

[0085] The decision module evaluates the personnel flow and stay matching status results using the location point anomaly assessment index Dyc, and generates a matching status response strategy plan based on the personnel flow and stay matching status results, and simultaneously performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response, and device response.

[0086] This embodiment enables real-time monitoring and analysis of key data such as personnel flow, density, location, and dwell time within a building. Through comprehensive data acquisition and sophisticated data preprocessing by the sensor suite, detailed personnel flow characteristics are extracted, a Kalman filter model is established for trajectory prediction, and an anomaly assessment index Dyc for personnel intersections and dwell times is derived. The decision-making module further generates a matching state response strategy based on the anomaly assessment index, ensuring timely and effective personnel, broadcast, and equipment response measures in abnormal situations. This significantly enhances the intelligent level of building security and effectively addresses issues such as data lag, untimely response, and monitoring blind spots in traditional security systems, providing a more efficient and secure building environment management solution.

[0087] Example 2

[0088] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a data collection unit and a data transmission unit;

[0089] The data collection unit collects data through a sensor group placed in the building. The sensor group includes a camera sensor, an infrared sensor, a motion sensor, and a positioning sensor. The sensor group collects data on the flow of people, the density of people in a fixed area, the location of people, and the time people stay in the building in real time to form a collection data set.

[0090] The data transmission unit compresses and packages the collected data set by using the Internet of Things, wireless network and Ethernet, and simultaneously transmits the packaged and compressed collected data set to the data preprocessing module;

[0091] Compression includes compressing the collected data set using the data compression algorithm Zlib and the data compression algorithm Gzip, marking the compressed collected data set with a timestamp and a data source identifier, and then packaging and sending it.

[0092] The data preprocessing module includes a data filtering unit and an extraction unit;

[0093] The data filtering unit preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, wherein the data cleaning preprocessing includes noise detection, outlier detection and filling missing values, specifically including using standard deviation method preprocessing, box plot method and isolation forest method, mean filling method, median filling method and nearest neighbor interpolation method for specific processing; data normalization processing includes using maximum and minimum normalization processing and Z-Score normalization method to convert data of different dimensions into the same range;

[0094] The extraction unit then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stop time features to form a feature vector X, specifically X={v, , P(x, y), Ts};

[0095] Among them, the personnel flow characteristics include calculating the personnel's moving speed v, through The formula is calculated and obtained, where v represents the moving speed of the personnel, Indicates the time interval The distance moved within Indicates a time interval;

[0096] The population density feature includes calculating the population density ,pass The formula is calculated and obtained, where represents the density of people, N represents the number of people in the detection area, and a represents the area of ​​the detection area;

[0097] Personnel position features include extracting the coordinate position P(x, y) of the person, where x and y represent the horizontal and vertical positions of the person in the detection area, respectively.

[0098] The personnel dwell time feature includes calculating the personnel dwell time Ts, through The formula is calculated and obtained, where Indicates the time when the personnel moves to the stopping position, Indicates the time when the person leaves the stopping position.

[0099] Example 3

[0100] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the trajectory prediction module includes a modeling unit and an evaluation unit;

[0101] The modeling unit establishes a Kalman filter model using the feature vector X, trains the Kalman filter model, and records the predicted position and speed of the personnel through marking and real-time tracking, obtains the predicted trajectory formed by the number of personnel intersection position points and the number of personnel intersection stay time position points, synchronously marks the predicted trajectory points as the total number of intersection position points n, and corrects the state vector Iteratively correct the predicted trajectory;

[0102] The evaluation unit trains the personnel flow features and personnel position features in the feature vector X through the Kalman filter model to obtain the intersection point of the personnel trajectory , and calculate the intersection position of each trajectory by The personnel location characteristics and personnel stop time characteristics are used to obtain the length of time personnel stay at each intersection. , synchronize the track intersection position Length of stay of people at the meeting point Fitting is performed to obtain the abnormality evaluation index Dyc of the personnel's location point.

[0103] The location point anomaly evaluation index Dyc is obtained by the following calculation formula:

[0104] ;

[0105] Where, represents the location of the i-th trajectory intersection point, Represents the fitting value of the intersection position, specifically by fitting all the trajectory intersection positions Perform fitting to obtain the center point trend and center representative position, Represents the standard deviation of the intersection position, specifically by counting the intersection positions of all trajectories The distribution of , calculate the standardization to obtain, Indicates the length of time people stay at the i-th intersection, It represents the fitting value of the length of time people stay at the intersection, by calculating the length of time people stay at all intersections Perform fitting to obtain the center point trend and representative value, It represents the standard deviation of the length of time people stay at the intersection, specifically by counting the length of time people stay at all intersections The distribution of is obtained by calculating the standard deviation, where n represents the total number of intersection locations.

[0106] Among them, the corrected state vector The predicted trajectory is iteratively corrected through the following training steps: Step 1: Define the state vector of the Kalman filter model :

[0107] ;

[0108] Where, and They represent the position coordinates of the personnel at time t, and They represent the speed of the person at time t;

[0109] Step 2: Create a state vector Transfer equation:

[0110] ;

[0111] Where A represents the state vector Transfer matrix, B represents the control matrix, represents the control input, represents the process noise,

[0112] State vector The transfer matrix A is as follows, which is used to describe the state vector The state change relationship from time t to time t+1 is:

[0113] ;

[0114] Where, Indicates a time interval;

[0115] Step 3: Establish the observation equation:

[0116] ;

[0117] Where H represents the observation matrix, Represents measurement noise, specifically the random error in the observation process. The observation matrix H is as follows: the observation matrix H transforms the state vector Mapped to the observation space,

[0118] ;

[0119] Step 4: Establish the prediction step, including establishing the prediction prior state vector Estimate and predict the a priori error covariance:

[0120] Predicted prior state vector estimate:

[0121] ;

[0122] Where, represents the prior state vector, represents the posterior state vector estimate at time t-1;

[0123] Forecast a priori error covariance:

[0124] ;

[0125] Where, Represents the prior error covariance matrix, specifically the prediction of the error covariance at time t, reflecting the predicted state vector uncertainty, represents the posterior error covariance matrix at time t-1, specifically the error covariance corrected after combining the observed data at time t-1, Q represents the process noise covariance matrix, specifically the covariance of the process noise, reflecting the system in the state vector uncertainty in the transfer process;

[0126] Step 5: Establish the update step and calculate the Kalman gain:

[0127] ;

[0128] Where, Represents the Kalman gain, specifically the weight value used to balance the predicted value and the observed value, represents the prior error covariance matrix, H represents the observation matrix, and R represents the measurement noise covariance matrix, which specifically represents the covariance of the observation noise and reflects the uncertainty in the observation process.

[0129] Among them, step 5: establish the update step, which also includes the updated posterior state vector Estimate and update the posterior error covariance:

[0130] Updated posterior state vector estimate:

[0131] ;

[0132] Where, Represents the updated posterior state vector estimate, specifically the state vector corrected after combining the observed data estimate, Represents the state vector at time t The prior prediction of the state vector at time t The predicted value of is not corrected with the observed data;

[0133] Update the posterior error covariance:

[0134] ;

[0135] Where, Represents the updated posterior error covariance matrix, specifically the error covariance corrected after combining the observed data, reflecting the uncertainty of the posterior state vector estimate.

[0136] In this embodiment, trajectory prediction is performed in conjunction with the Kalman filter model to achieve accurate monitoring of the flow and stay of people in a building. By training the characteristic vector X, the position and speed of people are tracked and recorded in real time, and the predicted intersection points and stay durations of people are obtained and corrected, thereby forming a predicted trajectory. By calculating the anomaly assessment index Dyc of the trajectory intersection point position and the stay duration of people, abnormal situations are effectively identified. By iteratively correcting the predicted trajectory and combining the state vector transfer equation and observation equation of the Kalman filter model, the accuracy of the prediction and the timeliness of the response are significantly improved.

[0137] Example 4

[0138] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the decision module includes a matching unit and an execution unit;

[0139] The matching unit evaluates the personnel flow and stay matching status results by comparing the location point abnormality evaluation index Dyc with the preset abnormality evaluation matching threshold PG, and generates a matching status response strategy scheme for the personnel flow and stay matching status results according to the matching status results;

[0140] The execution unit performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response and device response.

[0141] The matching status response strategy solution is obtained through the following matching method:

[0142] When the location point abnormality evaluation index Dyc is less than the abnormality evaluation matching threshold PG, the personnel flow and stay matching status is obtained without abnormal results, and the matching status is generated without executing the response strategy plan;

[0143] When the location point abnormality assessment index Dyc ≥ the abnormality assessment matching threshold PG, the abnormal results of the personnel flow and stay matching status are obtained, and a matching status execution response strategy plan is generated, including notifying relevant personnel to go for maintenance, adjusting the operating status of environmental equipment, and increasing the monitoring detection frequency of the abnormal location point to 150% of the original detection frequency;

[0144] When the location point anomaly assessment index Dyc ≥ twice the anomaly assessment matching threshold PG, obtain the personnel flow and stay matching status warning results, generate the matching status warning execution response strategy plan, including adjusting the access control status, turning on the broadcast notification, dispatching relevant personnel to deal with the situation, and adjusting the operating status of the environmental equipment.

[0145] In this embodiment, the personnel flow and stay matching status results are evaluated by comparing the location point anomaly assessment index Dyc with the preset anomaly assessment matching threshold PG, and a matching status response strategy is generated based on the matching status results. This multi-level response strategy can not only deal with potential safety hazards in a timely manner and improve the safety and efficiency of building management, but also provide rapid and effective early warning and emergency response when the problem is serious, thereby ensuring the safety of people and property in the building.

[0146] Example 5

[0147] An intelligent building security control system and control method based on big data analysis, please refer to Figure 2 , specifically: including the following steps:

[0148] Step 1: The data collection module collects data on personnel flow, density, location, and dwell time through a sensor group placed in the building to form a collection data set;

[0149] Step 2: The data preprocessing module preprocesses the collected data set, including data cleaning and normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features, and personnel dwell time features to form the feature vector X;

[0150] Step 3: The trajectory prediction module uses the feature vector X to establish a Kalman filter model. By training the Kalman filter model, the number of intersection points and the number of intersection duration points are obtained. After simultaneous fitting, the location point anomaly assessment index Dyc is obtained.

[0151] Step 4: The decision module evaluates the personnel flow and stay matching status results using the location point anomaly assessment index Dyc, and generates a matching status response strategy plan based on the personnel flow and stay matching status results. It also performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response, and device response.

[0152] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent building security control system based on big data analysis, characterized by: It includes data acquisition module, data preprocessing module, trajectory prediction module and decision-making module; The data acquisition module collects data on personnel flow, density, location and dwell time through a sensor group placed in the building to form a collection data set; The data preprocessing module preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features, and personnel stay time features to form a feature vector X; The trajectory prediction module includes a modeling unit and an evaluation unit. The modeling unit uses the feature vector X to establish a Kalman filter model, obtains the number of personnel intersection points and the number of personnel intersection stay time points by training the Kalman filter model, and marks the predicted trajectory point as the total number of intersection points n, and corrects the state vector Iteratively correct the predicted trajectory; The evaluation unit trains the personnel flow features and personnel position features in the feature vector X through a Kalman filter model to obtain the positions of the personnel's trajectory intersection points, and obtains the personnel's stay time at each intersection point by calculating the personnel position features and personnel stop time features at each trajectory intersection point. The trajectory intersection point positions and the personnel's stay time at the intersection point are simultaneously fitted to obtain the position point anomaly evaluation index Dyc, wherein the position point anomaly evaluation index Dyc is obtained by the following calculation formula: ； Where, represents the location of the i-th trajectory intersection point, Represents the fitting value of the intersection position, specifically by fitting all the trajectory intersection positions Perform fitting to obtain the center point trend and center representative position, Represents the standard deviation of the intersection position, specifically by counting the intersection positions of all trajectories The distribution of , calculate the standardization to obtain, Indicates the length of time people stay at the i-th intersection, It represents the fitting value of the length of time people stay at the intersection, by calculating the length of time people stay at all intersections Perform fitting to obtain the center point trend and representative value, It represents the standard deviation of the length of time people stay at the intersection, specifically by counting the length of time people stay at all intersections The distribution of is obtained by calculating the standard deviation, where n represents the total number of intersection locations; The decision module evaluates the personnel flow and stay matching status results using the location point abnormality evaluation index Dyc, and generates a matching status response strategy scheme for the personnel flow and stay matching status results based on the matching status results, and simultaneously performs specific execution according to the content of the matching status response strategy scheme, including personnel response, broadcast response, and device response; wherein, the generating of the matching status response strategy scheme for the personnel flow and stay matching status results based on the matching status results includes: When the location point abnormality evaluation index Dyc is less than the abnormality evaluation matching threshold PG, it is determined that there is no abnormal result in the matching status of personnel flow and stay, and a matching status non-execution response strategy plan is generated; When the location point abnormality assessment index Dyc ≥ the abnormality assessment matching threshold PG, the personnel flow and stay matching status is determined to be abnormal, and a matching status execution response strategy is generated, including notifying relevant personnel to go for maintenance, adjusting the operating status of environmental equipment, and increasing the abnormal location point monitoring detection frequency to 150% of the original detection frequency; When the location point abnormality assessment index Dyc ≥ twice the abnormality assessment matching threshold PG, the personnel flow and stay matching status warning result is determined, and a matching status warning execution response strategy plan is generated, including adjusting the access control status, turning on broadcast notifications, dispatching relevant personnel to deal with the situation, and adjusting the operating status of environmental equipment.

2. The intelligent building security control system based on big data analysis according to claim 1 is characterized in that: The data acquisition module includes a data collection unit and a data transmission unit; The data collection unit collects data through the sensor group installed in the building. The sensor group includes a camera sensor, an infrared sensor, a motion sensor, and a positioning sensor. The data is collected in real time, including data on personnel flow, personnel density in a fixed area, personnel location, and personnel dwelling time, to form the collected data set; The data transmission unit compresses and packages the collected data set by using the Internet of Things, wireless network and Ethernet, and simultaneously transmits the packaged and compressed collected data set to the data preprocessing module; The compression includes compressing the collected data set using the data compression algorithm Zlib and the data compression algorithm Gzip, marking the compressed collected data set with a timestamp and a data source identifier, and then packaging and sending the data set.

3. The intelligent building security control system based on big data analysis according to claim 1 is characterized in that: The data preprocessing module includes a data filtering unit and an extraction unit; The data filtering unit preprocesses the collected data set, including data cleaning preprocessing and data normalization preprocessing, wherein the data cleaning preprocessing includes noise detection, outlier detection and filling missing values, specifically including using standard deviation method preprocessing, box plot method and isolation forest method, mean filling method, median filling method and nearest neighbor interpolation method for specific processing; the data normalization processing includes using maximum and minimum normalization processing and Z-Score normalization method to convert data of different dimensions into the same range; The extraction unit then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features and personnel stop time features to form a feature vector X, specifically X={v, , P(x, y), Ts}; The personnel flow characteristics include calculating the moving speed v of personnel, through The formula is calculated and obtained, where v represents the moving speed of the personnel, Indicates the time interval The distance moved within Indicates a time interval; The personnel density feature includes calculating the personnel density ,pass The formula is calculated and obtained, where represents the density of people, N represents the number of people in the detection area, and a represents the area of ​​the detection area; The personnel position feature includes extracting the coordinate position P(x, y) of the personnel, where x and y represent the horizontal axis position and vertical axis position of the personnel in the detection area respectively; The personnel dwell time feature includes calculating the personnel dwell time Ts, through The formula is calculated and obtained, where Indicates the time when the personnel moves to the stopping position, Indicates the time when the person leaves the stopping position.

4. The intelligent building security control system based on big data analysis according to claim 1, characterized in that: in, Corrected state vector The predicted trajectory is iteratively corrected through the following training steps: Step 1: Define the state vector of the Kalman filter model : ； Where, and They represent the position coordinates of the personnel at time t, and They represent the speed of the person at time t; Step 2: Create a state vector Transfer equation: ； Where A represents the state vector Transfer matrix, B represents the control matrix, represents the control input, represents process noise, and the state vector The transfer matrix A is as follows, which is used to describe the state vector The state change relationship from time t to time t+1 is: ； Where, Indicates a time interval; Step 3: Establish the observation equation: ； Where H represents the observation matrix, Represents measurement noise, specifically the random error in the observation process. The observation matrix H is as follows: the observation matrix H transforms the state vector Mapped to the observation space, ； Step 4: Establish the prediction step, including establishing the prediction prior state vector Estimate and predict the a priori error covariance: Predict the prior state vector estimate: ； Where, represents the prior state vector, represents the posterior state vector estimate at time t-1; Forecast a priori error covariance: ； Where, Represents the prior error covariance matrix, specifically the prediction of the error covariance at time t, reflecting the predicted state vector uncertainty, represents the posterior error covariance matrix at time t-1, specifically the error covariance corrected after combining the observed data at time t-1, Q represents the process noise covariance matrix, specifically the covariance of the process noise, reflecting the system in the state vector uncertainty in the transfer process; Step 5: Establish the update step and calculate the Kalman gain: ； Where, represents the Kalman gain, specifically the weight value used to balance the predicted value and the observed value, and R represents the measurement noise covariance matrix, specifically the covariance of the observation noise, reflecting the uncertainty in the observation process.

5. The intelligent building security control system based on big data analysis according to claim 4 is characterized in that: in, Step 5: Establish the update step, including the updated posterior state vector Estimate and update the posterior error covariance: Updated posterior state vector estimate: ； Where, Represents the updated posterior state vector estimate, specifically the state vector corrected after combining the observed data estimate, Represents the prior state vector, specifically the state vector at time t The predicted value of is not corrected with the observed data; Update the posterior error covariance: ； Where, Represents the updated posterior error covariance matrix, specifically the error covariance corrected after combining the observed data, reflecting the uncertainty of the posterior state vector estimate.

6. The intelligent building security control system based on big data analysis according to claim 1, characterized in that: The decision module includes a matching unit and an execution unit; The matching unit evaluates the personnel flow and stay matching status results by comparing the position point abnormality evaluation index Dyc with the preset abnormality evaluation matching threshold PG, and generates a matching status response strategy scheme for the personnel flow and stay matching status results according to the matching status results; The execution unit performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response and device response.

7. A smart building security control method based on big data analysis, applied to the smart building security control system based on big data analysis according to any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: The data collection module collects data on personnel flow, density, location, and dwell time through a sensor group placed in the building to form a collection data set; Step 2: The data preprocessing module preprocesses the collected data set, including data cleaning and normalization preprocessing, and then performs feature extraction to obtain personnel flow features, personnel density features, personnel location features, and personnel dwell time features to form the feature vector X; Step 3: The trajectory prediction module uses the feature vector X to establish a Kalman filter model. By training the Kalman filter model, the number of intersection points and the number of intersection duration points are obtained. After simultaneous fitting, the location point anomaly assessment index Dyc is obtained. Step 4: The decision module evaluates the personnel flow and stay matching status results using the location point anomaly assessment index Dyc, and generates a matching status response strategy plan based on the personnel flow and stay matching status results. It also performs specific execution according to the content of the matching status response strategy plan, including personnel response, broadcast response, and device response.

Images