| CPC G06Q 10/0635 (2013.01) [G16Y 20/10 (2020.01); G16Y 40/10 (2020.01); G16Y 10/40 (2020.01)] | 17 Claims |
|
1. A method for predicting a water accumulation risk in a smart city implemented based on a management platform of an Internet of Things (IoT) system for predicting a water accumulation in a smart city, wherein the IoT system is configured as a part of a processing device or realized by the processing device, the IoT system further comprising a user platform, a service platform, a sensor network platform, and an object platform, wherein the management platform includes a general database and a plurality of management sub-platforms, and the sensor network platform includes a plurality of sensor network sub-platforms; the sensor network platform is connected with the management platform and the object platform to realize functions of perceptual information sensor communication and control information sensor communication; the method comprising:
obtaining area information of a target area based on the object platform and transmitting the area information to a management sub-platform through a sensor network sub-platform corresponding to the management sub-platform, and uploading the area information to the general database of the management platform by the management sub-platform;
predicting, based on the area information, a water accumulation risk in the target area using the management platform;
determining, based on the water accumulation risk, an adjustment scheme corresponding to the target area using the management platform, including:
in response to the water accumulation risk satisfying a first preset condition, determining, based on processing of the area information performed by a joint scheduling model, a first joint regulation scheme corresponding to the target area, the first joint regulation scheme including a joint scheduling scheme of the regulation and storage facilities in the target area, the joint scheduling model being a neural network model;
wherein the joint scheduling model is obtained through a training process including:
inputting, by the management platform, a plurality of first training samples with first labels into an initial joint scheduling model, wherein the first training samples include historical area information of a historical target area, and the first labels include a historical joint regulation scheme corresponding to the historical area information;
constructing a first loss function based on the first labels and outputs of the initial joint scheduling model, and
iteratively updating parameters of the initial joint scheduling model based on the first loss function; and
completing training of the initial joint scheduling model until the first loss function satisfies a first preset training condition, and obtaining the joint scheduling model; wherein the first preset training condition includes that the first loss function converges or a count of iterations reaches a first threshold; and
executing, by the general database of the management platform, an adjustment instruction corresponding to the adjustment scheme and transmitting the adjustment scheme to the user platform through the service platform.
|