Intelligent lamp post emergency system based on multi-source data fusion

Abstract

The invention discloses an intelligent lamp pole emergency system based on multi-source data fusion, and relates to the field of intelligent lamp poles. Comprising a sensing layer, an edge calculation layer and an application layer. The sensing layer is installed on a lamp pole and specifically comprises an optical sensor array, a millimeter wave radar matrix, an electric power metering module, a meteorological monitoring module and an acoustic sensor module. The edge calculation layer is installed in a lamp pole and specifically comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine. The application layer is connected to an emergency linkage system, a traffic management center, a city management scheduling platform and an environment monitoring platform, the optical sensor array is a wide-angle camera array, and a wide-angle camera supports HDR imaging. According to the invention, the whole system is respectively provided with front-end sensing, middle-end calculation decision and terminal emergency linkage based on an intelligent lamp pole structure, so that the phenomenon of'data island 'of a traditional intelligent lamp pole is broken, and automatic decision linkage after multi-source data fusion is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent lamp poles, in particular to an intelligent lamp pole emergency system based on multi-source data fusion. BACKGROUND

[0002] The intelligent lamp pole (Smart Street Light Pole) is a new type of smart city infrastructure taking city road lighting facilities as the carrier, integrating information sensing, communication transmission, edge computing and other modules, and realizing the intensive function integration of "multi-pole integration and one pole with multiple functions" through the pole-mounted Internet of Things system.

[0003] The city emergency system (Emergency Response System) is a public safety protection system with a closed loop of pre-warning, disposal and post-assessment, which realizes the rapid and coordinated disposal of emergencies (traffic accidents, natural disasters and public safety events) by integrating multi-department data resources.

[0004] In the prior art, the intelligent lamp pole can perform environmental monitoring, road traffic monitoring and other functions, but its functions are relatively independent, forming a form similar to "data island", so it cannot perform emergency linkage according to multi-source data.

[0005] Therefore, the present application proposes an intelligent lamp pole emergency system based on multi-source data fusion. SUMMARY

[0006] The purpose of the present application is to solve the problems in the prior art and propose an intelligent lamp pole emergency system based on multi-source data fusion.

[0007] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0008] An intelligent lamp pole emergency system based on multi-source data fusion includes a perception layer, an edge computing layer and an application layer.

[0009] The perception layer is installed on the lamp pole and specifically includes an optical sensor array, a millimeter wave radar matrix, a power metering module, a weather monitoring module and an acoustic sensor module.

[0010] The edge computing layer is installed in the lamp pole and specifically includes a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine.

[0011] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0012] Preferably: the optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the weather monitoring module includes a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor, and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0013] Preferably: the working steps of the intelligent lamp post emergency system are:

[0014] S1: The perception layer collects data around the lamp post in real time;

[0015] S2: The edge computing layer sequentially performs data preprocessing, spatio-temporal alignment, and multi-modal data fusion on the collected real-time data;

[0016] S3: The edge computing layer performs decision-making on the fused multi-modal data based on federated learning-based algorithms;

[0017] S4: The application layer links the emergency linkage system, the traffic management center, the urban management dispatch platform, and the environmental monitoring platform according to the type of the decision event.

[0018] Preferably: in the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the weather monitoring module.

[0019] Preferably: in the S1 step, the spatio-temporal alignment step is:

[0020] S11: Time synchronization is performed using a hierarchical synchronization architecture;

[0021] S12: Radar-visual coordinate transformation is performed for spatial registration, and the transformation model is: where K is the internal parameter of the wide-angle camera, R is the rotation matrix, t is the installation offset, t = [a, b, c] T , a, b, and c are installation offset parameters, and x, y, and z are coordinates.

[0022] Preferably: the multi-modal data fusion in the S2 step uses a weight-based fusion method, wherein the weight uses dynamic weight distribution.

[0023] Preferably: the weight-based fusion method includes the following steps:

[0024] S21: Data fusion, coordinate transformation matrix calculation based on collinearity equation, and then mask generation using a dynamic ROI mask generation algorithm;

[0025] S22: Feature fusion, feature fusion using cross-modal attention mechanism;

[0026] S23: decision fusion, D-S evidence theory is used for decision fusion.

[0027] Preferably, the formula of the dynamic weight distribution is: wherein η (t) is a time attenuation factor, σ 2 is the instantaneous variance of the sensor.

[0028] Preferably, in the dynamic weight distribution, η (t) = e -γΔt , and γ is in the range of (0.04, 0.06).

[0029] Preferably, in the S3 step, the type and index of the event decision are:

[0030] Traffic congestion, vehicle speed less than 5km / h and duration greater than 5min and traffic density ≥ 85%;

[0031] Group gathering, the number of people in the area within 10 square meters is greater than 8 and the number of people talking at the same time is greater than 3 and the displacement entropy is less than 0.5;

[0032] Illegal parking, vehicle stationary time greater than 3 minutes and parking area non-parking area;

[0033] Environmental anomaly, PM2.5 mutation greater than 50ug / m 3 or wind mutation greater than 3 levels or temperature mutation greater than 5 degrees Celsius.

[0034] The beneficial effects of the present application are:

[0035] 1. The present application breaks the "data island" phenomenon of traditional intelligent lamp poles by setting up front-end sensing, middle-end computing decision and terminal emergency linkage based on the intelligent lamp pole structure, thereby realizing automatic decision linkage after multi-source data fusion. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is an intelligent lamp pole emergency system architecture diagram based on multi-source data fusion proposed by the present application;

[0037] Figure 2 is an intelligent lamp pole emergency system work flow chart based on multi-source data fusion proposed by the present application. DETAILED DESCRIPTION

[0038] The technical solutions of the present application will be further described in detail below in combination with specific embodiments.

[0039] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting", "setting" should be understood broadly, for example, it can be fixedly connected, set, or detachably connected, set, or integrally connected, set. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0040] Embodiment 1:

[0041] An intelligent lamp pole emergency system based on multi-source data fusion includes a perception layer, an edge computing layer and an application layer;

[0042] The perception layer is installed on the lamp pole, and specifically includes an optical sensor array, a millimeter wave radar matrix, a power metering module, a weather monitoring module and an acoustic sensor module;

[0043] The edge computing layer is installed in the lamp pole, and specifically includes a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine;

[0044] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0045] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the weather monitoring module includes a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rain sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0046] The working steps of the intelligent lamp pole emergency system are:

[0047] S1: The perception layer collects data around the lamp pole in real time;

[0048] S2: The edge computing layer sequentially performs data preprocessing, space-time alignment and multi-modal data fusion on the collected real-time data;

[0049] S3: The edge computing layer makes decisions based on federated learning algorithm for the fused multi-modal data;

[0050] S4: The application layer links the emergency linkage system, the traffic management center, the city management dispatch platform and the environmental monitoring platform according to the type of decision event.

[0051] Embodiment 2:

[0052] An intelligent lamp pole emergency system based on multi-source data fusion includes a perception layer, an edge computing layer and an application layer;

[0053] The perception layer is installed on the lamp pole and specifically includes an optical sensor array, a millimeter wave radar matrix, a power metering module, a weather monitoring module and an acoustic sensor module.

[0054] The edge computing layer is installed in the lamp pole and specifically includes a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine.

[0055] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0056] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the weather monitoring module includes a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0057] The working steps of the intelligent lamp pole emergency system are as follows:

[0058] S1: The perception layer collects data around the lamp pole in real time;

[0059] S2: The edge computing layer sequentially performs data preprocessing, space-time alignment and multi-modal data fusion on the collected real-time data.

[0060] S3: The edge computing layer makes decisions based on the algorithm of federated learning on the fused multi-modal data.

[0061] S4: The application layer links the emergency linkage system, the traffic management center, the city management dispatch platform and the environmental monitoring platform according to the type of the decision event.

[0062] In the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the weather monitoring module.

[0063] In the S1 step, the space-time alignment step is as follows:

[0064] S11: Time synchronization is performed using a hierarchical synchronization architecture.

[0065] S12: Radar-visual coordinate transformation is performed for space registration, and the transformation model is as follows: where K is the internal parameter of the wide-angle camera, R is the rotation matrix, t is the installation offset, t = [a, b, c] T , a, b, c are installation offset parameters, and x, y, z are coordinates.

[0066] Example 3:

[0067] An intelligent lamp pole emergency system based on multi-source data fusion, comprising a perception layer, an edge computing layer and an application layer;

[0068] The perception layer is installed on the lamp pole and specifically comprises an optical sensor array, a millimeter wave radar matrix, a power metering module, a meteorological monitoring module and an acoustic sensor module.

[0069] The edge computing layer is installed in the lamp pole and specifically comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine.

[0070] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0071] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the meteorological monitoring module comprises a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0072] The working steps of the intelligent lamp pole emergency system are as follows:

[0073] S1: The perception layer collects data around the lamp pole in real time;

[0074] S2: The edge computing layer sequentially performs data preprocessing, space-time alignment and multi-modal data fusion on the collected real-time data;

[0075] S3: The edge computing layer makes decisions based on the algorithm of federated learning on the fused multi-modal data;

[0076] S4: The application layer links the emergency linkage system, the traffic management center, the city management dispatch platform and the environmental monitoring platform according to the type of the decision event.

[0077] In the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the meteorological monitoring module.

[0078] In the S1 step, the space-time alignment step is as follows:

[0079] S11: Time synchronization is performed using a hierarchical synchronization architecture;

[0080] S12: Radar-vision coordinate transformation is performed for space registration, and the transformation model is as follows: where K is the internal parameter of the wide-angle camera, R is the rotation matrix, t is the installation offset, t = [a, b, c] T , a, b, c are installation offset parameters, and x, y, z are coordinates.

[0081] The S2 multi-modal data fusion adopts a weight-based fusion method, wherein the weight adopts dynamic weight distribution.

[0082] The weight-based fusion method comprises the following steps:

[0083] S21: data fusion, coordinate transformation matrix calculation based on collinear equation, and then mask generation using a dynamic ROI mask generation algorithm;

[0084] S22: feature fusion, feature fusion using cross-modal attention mechanism;

[0085] S23: decision fusion, using D-S evidence theory for decision fusion.

[0086] The formula of the dynamic weight distribution is: wherein η (t) is a time decay factor, and σ 2 is the instantaneous variance of the sensor.

[0087] In the dynamic weight distribution, η (t) = e -γΔt , and γ = 0.04.

[0088] Embodiment 4:

[0089] An intelligent lamp pole emergency system based on multi-source data fusion, comprising a perception layer, an edge computing layer and an application layer;

[0090] The perception layer is installed on the lamp pole and specifically comprises an optical sensor array, a millimeter wave radar matrix, a power metering module, a weather monitoring module and an acoustic sensor module;

[0091] The edge computing layer is installed in the lamp pole and specifically comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine;

[0092] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0093] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the weather monitoring module comprises a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rain sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0094] The working steps of the intelligent lamp pole emergency system are:

[0095] S1: the perception layer collects data around the lamp pole in real time;

[0096] S2: the edge computing layer sequentially performs data preprocessing, space-time alignment, and multi-modal data fusion on the collected real-time data;

[0097] S3: the edge computing layer performs decision-making on the fused multi-modal data based on federated learning algorithm;

[0098] S4: the application layer links the emergency linkage system, the traffic management center, the urban management dispatch platform, and the environmental monitoring platform according to the type of the decision event.

[0099] In the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the meteorological monitoring module.

[0100] In the S1 step, the space-time alignment step is as follows:

[0101] S11: time synchronization is performed using a hierarchical synchronization architecture;

[0102] S12: radar-vision coordinate transformation is performed for space registration, and the transformation model is as follows: where K is the wide-angle camera internal parameter, R is the rotation matrix, t is the installation offset, t = [a, b, c] T , a, b, and c are installation offset parameters, and x, y, and z are coordinates.

[0103] The multi-modal data fusion in the S2 step uses a weight-based fusion method, wherein the weight uses dynamic weight distribution.

[0104] The weight-based fusion method includes the following steps:

[0105] S21: data fusion, coordinate transformation matrix calculation is performed based on the collinearity equation, and then a dynamic ROI mask generation algorithm is used to generate a mask;

[0106] S22: feature fusion, cross-modal attention mechanism is used for feature fusion;

[0107] S23: decision fusion, D-S evidence theory is used for decision fusion.

[0108] The formula for dynamic weight distribution is as follows: where η (t) is the time decay factor, and σ 2 is the instantaneous variance of the sensor.

[0109] In the dynamic weight distribution, η (t) = e -γΔt , and γ = 0.05.

[0110] Example 5:

[0111] An intelligent lamp pole emergency system based on multi-source data fusion, comprising a perception layer, an edge computing layer and an application layer;

[0112] The perception layer is installed on the lamp pole and specifically comprises an optical sensor array, a millimeter wave radar matrix, a power metering module, a meteorological monitoring module and an acoustic sensor module.

[0113] The edge computing layer is installed in the lamp pole and specifically comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine.

[0114] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform.

[0115] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the meteorological monitoring module comprises a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter.

[0116] The working steps of the intelligent lamp pole emergency system are as follows:

[0117] S1: The perception layer collects data around the lamp pole in real time;

[0118] S2: The edge computing layer sequentially performs data preprocessing, space-time alignment and multi-modal data fusion on the collected real-time data.

[0119] S3: The edge computing layer makes decisions based on the algorithm of federated learning on the fused multi-modal data.

[0120] S4: The application layer links the emergency linkage system, the traffic management center, the city management dispatch platform and the environmental monitoring platform according to the type of the decision event.

[0121] In the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the meteorological monitoring module.

[0122] In the S1 step, the space-time alignment step is as follows:

[0123] S11: Time synchronization is performed using a hierarchical synchronization architecture.

[0124] S12: Radar-visual coordinate transformation is performed for space registration, and the transformation model is as follows: Where K is the internal parameter of the wide-angle camera, R is the rotation matrix, and t is the installation offset, t = [a, b, c] Ta, b, c are installation offset parameters, x, y, z are coordinates.

[0125] The S2 multi-modal data fusion adopts a weight-based fusion method, wherein the weight adopts dynamic weight distribution.

[0126] The weight-based fusion method comprises the following steps:

[0127] S21: data fusion, coordinate transformation matrix calculation based on collinear equation, and then mask generation using a dynamic ROI mask generation algorithm;

[0128] S22: feature fusion, feature fusion using cross-modal attention mechanism;

[0129] S23: decision fusion, using D-S evidence theory for decision fusion.

[0130] The formula of the dynamic weight distribution is: Wherein η (t) is a time decay factor, and σ 2 is the instantaneous variance of the sensor.

[0131] In the dynamic weight distribution, η (t) =e -γΔt , and γ=0.06.

[0132] Embodiment 6:

[0133] An intelligent lamp pole emergency system based on multi-source data fusion comprises a perception layer, an edge computing layer, and an application layer;

[0134] The perception layer is installed on the lamp pole and specifically comprises an optical sensor array, a millimeter wave radar matrix, a power metering module, a weather monitoring module, and an acoustic sensor module;

[0135] The edge computing layer is installed in the lamp pole and specifically comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit, and a time decision engine;

[0136] The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform, and an environmental monitoring platform.

[0137] The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the weather monitoring module comprises a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor, and a geomagnetic vehicle detector; and the power metering module is a smart meter.

[0138] The working steps of the intelligent lamp pole emergency system are as follows:

[0139] S1: The perception layer collects data near the lamp pole in real time;

[0140] S2: The edge computing layer sequentially performs data preprocessing, space-time alignment, and multi-modal data fusion on the collected real-time data;

[0141] S3: The edge computing layer performs decision-making on the fused multi-modal data based on federated learning algorithm;

[0142] S4: The application layer links the emergency response system, traffic management center, urban management dispatch platform, and environmental monitoring platform according to the type of decision event.

[0143] In the S1 step, data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for meteorological monitoring module.

[0144] In the S1 step, the space-time alignment step is:

[0145] S11: Time synchronization is performed using a hierarchical synchronization architecture;

[0146] S12: Radar-vision coordinate transformation is performed for spatial registration, and the transformation model is: where K is the wide-angle camera internal parameter, R is the rotation matrix, t is the installation offset, t = [a, b, c] T , a, b, c are installation offset parameters, and x, y, z are coordinates.

[0147] The multi-modal data fusion in S2 uses a weight-based fusion method, where the weight uses dynamic weight distribution.

[0148] The weight-based fusion method includes the following steps:

[0149] S21: Data fusion, coordinate transformation matrix calculation based on collinearity equation, and then using dynamic ROI mask generation algorithm to generate mask;

[0150] S22: Feature fusion, using cross-modal attention mechanism for feature fusion;

[0151] S23: Decision fusion, using D-S evidence theory for decision fusion.

[0152] The formula for dynamic weight distribution is: where η (t) is the time decay factor, and σ 2 is the instantaneous variance of the sensor.

[0153] In the dynamic weight distribution, η (t) = e -γΔt , γ ∈ (0.04, 0.06).

[0154] In the S3 step, the type and index of event decision are:

[0155] Traffic congestion, vehicle speed less than 5km / h and duration greater than 5min and traffic density ≥ 85%;

[0156] Group gathering, the number of people in the area within 10 square meters is greater than 8, the number of people talking at the same time is greater than 3, and the displacement entropy is less than 0.5;

[0157] Illegal parking, vehicle stationary time greater than 3 minutes and parking area non-parking area;

[0158] Environmental anomaly, PM2.5 mutation greater than 50μg / m 3 Or wind mutation greater than 3 levels or temperature mutation greater than 5 degrees Celsius.

[0159] The present application, by setting up front-end sensing, middle-end computing decision and terminal emergency linkage based on the whole system of intelligent lamp pole structure, breaks the "data island" phenomenon of traditional intelligent lamp pole, and realizes automatic decision linkage after multi-source data fusion.

[0160] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this, any person skilled in the art in the technical range disclosed by the present application, according to the technical scheme and the inventive concept of the present application, equivalent replacement or change, should be covered in the protection scope of the present application.

Claims

1. An intelligent lamp pole emergency system based on multi-source data fusion, characterized in that, The system comprises a perception layer, an edge computing layer and an application layer. The perception layer is installed on a lamp pole and comprises an optical sensor array, a millimeter wave radar matrix, a power metering module, a meteorological monitoring module and an acoustic sensor module. The edge computing layer is installed in the lamp pole and comprises a data preprocessing module, a space-time alignment engine, a multi-modal fusion analysis unit and a time decision engine. The application layer is connected to an emergency linkage system, a traffic management center, a city management dispatch platform and an environmental monitoring platform. 2.The intelligent lamp pole emergency system based on multi-source data fusion according to claim 1, characterized in that, The optical sensor array is a wide-angle camera array, and the wide-angle camera supports HDR imaging; the meteorological monitoring module comprises a temperature sensor, a humidity sensor, a wind speed and direction sensor, a rainfall sensor, a PM2.5 sensor, a carbon dioxide sensor and a geomagnetic vehicle detector, and the power metering module is a smart meter. 3.The intelligent lamp pole emergency system based on multi-source data fusion according to claim 1 or 2, characterized in that, The working steps of the smart lamp pole emergency system are as follows: S1: The perception layer collects data around the lamp pole in real time; S2: The edge computing layer sequentially performs data preprocessing, space-time alignment and multi-modal data fusion on the collected real-time data; S3: The edge computing layer makes decisions based on the algorithm of federated learning on the fused multi-modal data; S4: The application layer links the emergency linkage system, the traffic management center, the city management dispatch platform and the environmental monitoring platform according to the type of the decision event.

4. The intelligent lamp pole emergency system based on multi-source data fusion according to claim 3, characterized in that, In the S1 step, the data preprocessing includes inter-frame difference noise reduction and illumination compensation for visual data, dynamic target clustering and motion trajectory prediction for millimeter wave radar, and temperature compensation for the meteorological monitoring module. 5.The intelligent lamp pole emergency system based on multi-source data fusion of claim 3, wherein, In the S1 step, the space-time alignment step is as follows: S11: Time synchronization is performed using a hierarchical synchronization architecture; S12: radar-vision coordinate transformation is performed, spatial registration is performed, and a transformation model thereof is: wherein K is a wide-angle camera internal parameter, R is a rotation matrix, t is an installation offset, t = [a, b, c] T , a, b, and c are installation offset parameters, and x, y, and z are coordinates. 6.The intelligent lamp pole emergency system based on multi-source data fusion of claim 3, wherein, The multi-modal data fusion in the S2 step adopts a weight-based fusion method, wherein the weight uses dynamic weight distribution. 7.The intelligent lamp pole emergency system based on multi-source data fusion of claim 6, wherein, The weight-based fusion method comprises the following steps: S21: Data fusion, coordinate transformation matrix calculation based on collinearity equation, and then mask generation using a dynamic ROI mask generation algorithm; S22: Feature fusion, feature fusion using cross-modal attention mechanism; S23: Decision fusion, decision fusion using D-S evidence theory. 8.The intelligent lamp pole emergency system based on multi-source data fusion of claim 6, wherein, The formula of the dynamic weight distribution is: where η (t) is the time decay factor, σ 2 is the instantaneous variance of the sensor. 9.The intelligent lamp pole emergency system based on multi-source data fusion of claim 8, wherein, In the dynamic weight distribution, η (t) = e -γΔt , γ ∈ (0.04, 0.06). 10.The intelligent lamp pole emergency system based on multi-source data fusion of claim 3, wherein, In the S3 step, the type and index of the event decision are as follows: Traffic congestion, vehicle speed less than 5 km / h, duration greater than 5 min and traffic flow density ≥ 85%; Group gathering, more than 8 people in a 10 square meter area, more than 3 people speaking at the same time, and displacement entropy < 0.5; Illegal parking, vehicle stationary time greater than 3 minutes and non-parking area in the parking area. Environmental anomaly, PM2.5 mutation greater than 50 μg / m 3 or wind mutation greater than 3 classes or temperature mutation greater than 5 degrees Celsius.

Images