CN-121981847-A - Intelligent rural space-world multi-source data fusion method and system

Abstract

The invention belongs to the technical field of agricultural informatization and multi-source data processing, and particularly relates to an aerospace-ground multi-source data fusion method and system for an intelligent village, wherein the method comprises the steps of acquiring environmental data through satellite remote sensing, unmanned aerial vehicle and ground Internet of things, performing aerospace grid reconstruction, and establishing a standardized data set; the method comprises the steps of calculating algebraic belief by using a current observed value of each microscopic information source, a local space neighborhood mean value and a time difference value, then carrying out weighted calculation on the microscopic observed value, introducing a second-order difference value of macroscopic data as a compensation item to obtain a collaborative fusion value, calculating risk potential energy by combining the previous time fusion value and a crop physiological optimal target point value, and generating a nonlinear coupling execution action instruction. According to the invention, noise is suppressed and disaster mutation signals are reserved through an algebraic attenuation mechanism, so that the problem of balance between accuracy and robustness in heterogeneous data fusion is solved, and real-time accurate decision of risks in intelligent rural environments is realized.

Inventors

• XU JIN
• CHEN YIXIANG
• ZHANG WENJIE

Assignees

• 宁波工程学院

Dates

Publication Date

20260505

Application Date

20260401

Claims (10)

1. 1. An aerospace-earth multi-source data fusion method for intelligent villages is characterized by comprising the following steps of: Acquiring environment data through a satellite remote sensing platform, an unmanned aerial vehicle inspection system and a ground Internet of things base station, performing space-time gridding reconstruction and time resampling on the environment data, and acquiring a standardized data set; Acquiring current observation values of all the micro information sources, the observation mean value in the local space neighborhood and time difference values from the standardized data set, and calculating algebraic degrees based on the current observation values of the micro information sources, the observation mean value in the local space neighborhood and the time difference values; The method comprises the steps of obtaining macroscopic data from a standardized data set, calculating a second-order differential value of the macroscopic data, carrying out weighted calculation on a current observation value of a microscopic information source based on algebraic reliability, and compensating a weighted calculation result by utilizing the second-order differential value of the macroscopic data to obtain a collaborative fusion value; and calculating risk potential energy based on the collaborative fusion value, the collaborative fusion value at the last moment and a preset crop physiological optimal target point value, and acquiring an execution action instruction based on the risk potential energy.
2. 2. The method for intelligent village-oriented space-time multisource data fusion according to claim 1, wherein the performing space-time gridding reconstruction and time resampling on the environmental data to obtain the standardized data set comprises: The environment data comprises space-sky multi-source monitoring data, image pixels and sensing points in the environment data are mapped into a uniform geographic grid by adopting an inverse distance weighted interpolation method and a spline function interpolation technology, and a time sequence of the environment data is resampled to obtain a standardized data set.
3. 3. The intelligent village-oriented space-earth multi-source data fusion method according to claim 2, wherein the spatial resolution of the geographic grid is 8 meters, and the resampling frequency is once every 10 minutes.
4. 4. The intelligent village-oriented space-earth multi-source data fusion method as claimed in claim 1, wherein the algebraic degree of confidence satisfies the expression: ; In the formula, Represent the first Algebraic beliefs of the individual microscopic sources; represent the first Current observations of the individual microscopic sources; representing the observed mean value in the local spatial neighborhood; represent the first Time differential values of the individual microscopic sources; representing the spatial discrete sensitivity; representing the time-ripple sensitivity.
5. 5. The intelligent village-oriented space-sky multisource data fusion method according to claim 4, wherein the spatial discrete sensitivity and the time fluctuation sensitivity are obtained by least square fitting historical one year monitoring data.
6. 6. The intelligent village-oriented space-earth multi-source data fusion method as claimed in claim 1, wherein the collaborative fusion value satisfies the expression: ; In the formula, Representing a collaborative fusion value; representing the total number of effective microscopic sources in the local spatial neighborhood; represent the first Algebraic beliefs of the individual microscopic sources; represent the first Current observations of the individual microscopic sources; Representing curvature compensation gain; representing a sign function; A second order difference value representing macroscopic data; Representing the statistical variance of the current observed value of the microscopic information source in the local space neighborhood; representing an absolute value operation.
7. 7. The intelligent village-oriented space-earth multi-source data fusion method as defined in claim 6, wherein the curvature compensation gain is a proportionality constant calculated by comparing a proportionality relation between historical macroscopic data and microscopic true values.
8. 8. The intelligent village-oriented space-earth multi-source data fusion method according to claim 1, wherein the execution action instruction satisfies the expression: ; In the formula, Representing execution of an action instruction; representing a sign function; representing a collaborative fusion value; Representing the physiological optimal target point value of crops; representing the collaborative fusion value of the previous moment; Representing a trend sensitivity coefficient; representing natural constants.
9. 9. The intelligent village-oriented space-earth multi-source data fusion method as defined in claim 8, wherein the executing the action command based on the risk potential energy acquisition comprises: And the intelligent gateway of the Internet of things keeps a silence state in response to the executing action instruction being a negative value or zero.
10. 10. An intelligent country oriented space-to-earth multi-source data fusion system comprising a processor and a memory, the memory storing computer program instructions which when executed by the processor implement a intelligent country oriented space-to-earth multi-source data fusion method according to any one of claims 1-9.

Description

Intelligent rural space-world multi-source data fusion method and system Technical Field The invention relates to the technical field of agricultural informatization and multi-source data processing. More particularly, the invention relates to an aerospace-earth multi-source data fusion method and system for intelligent villages. Background The timeliness and the accuracy of environmental monitoring are extremely high in requirements for accurate agriculture and disaster early warning, and complex actual requirements are difficult to meet by means of a traditional monitoring mode of a single microscopic information source. Satellite remote sensing can provide surface coverage information on a macro scale, but is generally limited by a longer revisit period and cloud cover, so that instantaneous change of the ground is difficult to capture. Although the unmanned aerial vehicle inspection system has flexible high-resolution imaging capability, the unmanned aerial vehicle inspection system is generally limited by battery endurance time and flight airspace control, and all-weather large-scale coverage is difficult to realize. Although the ground wireless sensor network can provide high-frequency microscopic point location data, the spatial distribution density of the ground wireless sensor network is limited by high deployment and maintenance cost, so that a monitoring blind area exists. To overcome the limitations of single monitoring means, multi-sensor data fusion techniques have been developed aimed at obtaining more accurate and reliable environmental descriptions than single microscopic sources by integrating air, day, and ground multi-source data. Early fusion technologies commonly adopt a Kalman filtering or Bayesian estimation framework, and the classical algorithms commonly assume that system noise is subjected to Gaussian distribution and state change presents linear characteristics, while the real rural natural environment is highly unstructured, and environment parameters such as soil humidity, pest and disease distribution and the like commonly present significant nonlinear mutation and non-Gao Sichang tail noise characteristics, so that the linear fusion model is very easy to generate larger estimation deviation when facing extreme weather or sensor faults. In recent years, with the intervention of artificial intelligence technology, a deep learning fusion model based on a convolutional neural network and a cyclic neural network gradually becomes a research hotspot, and although the data-driven method is excellent in feature extraction, massive annotation data are generally required for training, and high-quality annotation samples are difficult to acquire due to high dynamic change of the rural environment. In addition, complex deep neural network models generally consume huge computing resources, and are difficult to operate efficiently on country edge computing terminals with limited computational power, so that the problems of high delay and high energy consumption of data processing are caused. The existing multi-source data fusion technology is difficult to ensure calculation instantaneity and capture robustness of environmental mutation signals when facing heterogeneous data with huge space-time resolution difference and uneven data quality, and often has the phenomena of lag fusion results or insensitivity to abnormal risks, so that the severe requirements of real-time accurate decisions in intelligent rural scenes can not be met. Disclosure of Invention In order to solve the technical problems of space-time mismatch, poor robustness of data fusion and delay of disaster risk response of the multi-source heterogeneous data, the invention provides a scheme in the following aspects. In a first aspect, the invention provides an air-space-earth multi-source data fusion method for an intelligent village, which comprises the steps of obtaining environmental data through a satellite remote sensing platform, an unmanned aerial vehicle inspection system and a ground Internet of things base station, carrying out space-time gridding reconstruction and time resampling on the environmental data to obtain a standardized data set, obtaining current observation values of all micro information sources, observation mean values and time differential values in local space neighbors from the standardized data set, calculating algebraic beliefs based on the current observation values of the micro information sources and the observation mean values and time differential values in the local space neighbors, obtaining macroscopic data from the standardized data set, calculating second-order differential values of the macroscopic data, carrying out weighted calculation on the current observation values of the micro information sources based on the algebraic beliefs, compensating the weighted calculation result by utilizing second-order differential values of the macroscopic data to obtain collaborative fusion value