| CPC H04W 4/90 (2018.02) [G06Q 10/06313 (2013.01); G16Y 10/75 (2020.01); G16Y 40/50 (2020.01)] | 9 Claims |
|
1. A method for determining a fire rescue plan in a smart city applied to an Internet of Things (IoT) system for determining a fire rescue plan in a smart city, wherein the IoT system includes a user platform, a service platform, a management platform, a sensing network platform, and an object platform, comprising:
obtaining monitoring data of a disaster area in a first period of time, wherein the monitoring data includes at least one of a fire scene image or fire scene environment data, and the fire scene environment data includes at least one of a temperature, a humidity, a wind volume, or a wind direction;
predicting, based on the monitoring data and an arrival time of fire trucks, fire information in a second period of time via a fire prediction model, wherein
the fire prediction model is a machine learning model;
the fire prediction model includes a convolutional neural network model and a deep neural network model;
an input of the convolutional neural network model includes the fire scene image, and an output of the convolutional neural network model includes a fire scene image feature sequence;
an input of the deep neural network model includes the fire scene image feature sequence, a fire scene environment data sequence, and an arrival time sequence of the fire trucks, and an output of the deep neural network model includes a fire information sequence;
the fire prediction model is obtained through a joint training of the convolutional neural network model and the deep neural network model; and
the fire prediction mode is optimized based on known fire information, corresponding monitoring data, and the fire scene environment data;
determining, based on the monitoring data, flow information of the disaster area, via a flow prediction model, wherein
the flow information includes at least one of traffic flow information or human flow information;
the flow prediction model is a machine learning model;
the monitoring data includes a flow image of the disaster area;
the flow image includes at least one of a traffic flow image or a human flow image;
the flow prediction model includes a feature extraction layer and an identification layer; an input of the feature extraction layer includes the human flow image and the traffic flow image, and an output of the feature extraction layer includes a human flow image feature and a traffic flow image feature; an input of the identification layer includes the human flow image feature and the traffic flow image feature, and an output of the identification layer includes the human flow information and the traffic flow information;
the flow prediction model is obtained through a joint training of the feature extraction layer and the identification layer; the output of the feature extraction layer is used as the input of the identification layer;
wherein the joint training of the feature extraction layer and the identification layer includes:
inputting a sample human flow image and a sample traffic flow image into an initial feature extraction layer to obtain the human flow image feature and the traffic flow image feature output by the initial feature extraction layer, using the human flow image feature and the traffic flow image feature as an input of an initial flow prediction model to obtain the human flow information and the traffic flow information output by the initial flow prediction model; verifying the output of the initial feature extraction layer based on sample human flow information and sample traffic flow information to obtain verification data of an output structure of the initial feature extraction layer, and obtaining the flow prediction model by continuing training using the verification data until a trained feature extraction layer and a trained identification layer are obtained;
determining, based on the fire information and the flow information, a rescue plan, the rescue plan including at least one of a count of fire trucks or a count of ambulances;
feeding back, through the service platform, the fire information, the flow information, and the rescue plan to the user platform, wherein the user platform is configured for a user to check the fire information, the flow information, and the rescue plan; and
sending the rescue plan to the object platform through the sensing network platform, wherein the object platform is configured to implement the rescue plan; and the rescue plan includes:
sending the count of fire trucks and a count of firefighters required in the rescue plan to a fire department to implement rescue; and sending a count of ambulances required in the rescue plan to a medical system to implement the rescue.
|