CN-121301891-B - Method, device, equipment and medium for determining mycorrhiza restoration scheme

CN121301891BCN 121301891 BCN121301891 BCN 121301891BCN-121301891-B

Abstract

The application provides a method, a device, equipment and a medium for determining a mycorrhiza restoration scheme, which relate to the technical field of ecological restoration data analysis and comprise the steps of acquiring hyperspectral data and environmental data of a target soil area, obtaining predicted soil mechanical compositions based on the hyperspectral data, matching a plurality of alternative mycorrhiza corresponding to the predicted soil mechanical compositions of the target soil area from a configured association database of different soil mechanical compositions and different mycorrhiza suitability, calculating adaptation indexes of different alternative mycorrhiza according to the plurality of alternative mycorrhiza, the predicted soil mechanical compositions and the environmental data, determining restoration schemes and predicted restoration effects corresponding to different alternative mycorrhiza according to the adaptation indexes of different alternative mycorrhiza, the predicted soil mechanical compositions and the environmental data, and determining the mycorrhiza restoration scheme of the target soil area based on the adaptation indexes of different alternative mycorrhiza and the predicted restoration effects of corresponding restoration schemes. The application automatically generates a targeted mycorrhiza restoration scheme of the target area.

Inventors

JU XINGJUN
YANG HUIHUI
LI LIN
GAO SIHUA
GUO HAIQIAO
WANG CHANGJIAN
LI HONGQING
BI YINLI
ZHANG YANXU
MA SHAOPENG

Assignees

国能宝日希勒能源有限公司
中国矿业大学(北京)

Dates

Publication Date: 20260512
Application Date: 20251112

Claims (8)

1. A method of determining a mycorrhiza restoration regimen, the method comprising: acquiring hyperspectral data and environmental data of a target soil area; inputting the hyperspectral data into a pre-trained soil organic matter and water content prediction model to obtain the predicted organic matter content and the predicted water content of a target soil area; Inputting the hyperspectral data, the predicted organic matter content and the predicted water content into a pre-trained soil mechanical composition prediction model to obtain a predicted soil mechanical composition of the target soil region; Matching a plurality of alternative mycorrhizas corresponding to the predicted soil mechanical composition of the target soil area from a configured association database of different soil mechanical compositions and different mycorrhizal suitability; Calculating the adaptation indexes of different alternative mycorrhiza according to various alternative mycorrhiza, predicted soil mechanical composition and environmental data, wherein the calculation comprises the steps of calculating the physical adaptation indexes and the functional adaptation indexes of different alternative mycorrhiza according to the environmental data and the predicted soil mechanical composition, taking the product of the physical adaptation indexes and the functional adaptation indexes of any alternative mycorrhiza as the adaptation index of the alternative mycorrhiza, taking the alternative mycorrhiza corresponding to the adaptation index higher than the configured adaptation threshold as a first mycorrhiza, cross-combining the different first mycorrhiza to obtain a plurality of first mycorrhiza combinations, and calculating the cooperative adaptation indexes of the different first mycorrhiza combinations; Inputting the adaptation indexes, the predicted soil mechanical compositions and the environmental data of different alternative mycorrhiza into a pre-trained mycorrhiza restoration effect prediction model to obtain restoration schemes and predicted restoration effects corresponding to different alternative mycorrhiza; The method comprises the steps of dynamically optimizing an agent by utilizing pre-trained inoculation parameters, deleting the first mycorrhiza or the first mycorrhiza combination with the adaptation index or the collaborative adaptation index not larger than a configured adaptation threshold value or predicting the first mycorrhiza or the first mycorrhiza combination with the restoration effect not larger than a configured restoration threshold value from different first mycorrhiza and different first mycorrhiza combinations, obtaining different second mycorrhiza or different second mycorrhiza combinations, dynamically optimizing a configured multidimensional comprehensive decision model of the agent according to the inoculation parameters, calculating comprehensive decision scores of the different second mycorrhiza or the different second mycorrhiza combinations, adopting a configured combination optimizing strategy, dividing the comprehensive decision into an adaptation function, determining the optimal mycorrhiza in a multi-round iteration mode, taking the restoration effect as a reward function, taking different inoculation parameters in the restoration scheme as action spaces, and adopting an optimizing algorithm to determine the target restoration scheme corresponding to the optimal mycorrhiza, thus obtaining the mycorrhiza restoration scheme of the target soil area.
2. The method of claim 1, wherein the method of acquiring hyperspectral data of the target soil region comprises: Acquiring original hyperspectral data of a target soil area; Preprocessing the original hyperspectral data to obtain first hyperspectral data; Fitting an upper envelope curve by adopting an envelope curve removing method, and dividing the first hyperspectral data by the envelope curve to obtain normalized second hyperspectral data; Intercepting third hyperspectral data positioned in a configured target band from the second hyperspectral data; And carrying out multiplication scattering correction and first derivative transformation on the third hyperspectral data to obtain fourth hyperspectral data, and outputting the fourth hyperspectral data as hyperspectral data of a target soil area.
3. The method of claim 1, wherein the soil organic matter and water content prediction model comprises: An input layer for inputting hyperspectral data of a target soil region; The physical attention embedding layer is used for multiplying the input hyperspectral data with the configured binary attention mask to obtain mask-optimized one-dimensional spectral data; The shared feature extraction layer is used for extracting features of the one-dimensional spectrum data subjected to mask optimization to obtain a multi-dimensional shared feature vector; The feature refinement extraction layer is used for performing refinement feature extraction on the multidimensional shared feature vectors respectively to obtain a first feature vector and a second feature vector; The double-target fusion prediction layer is used for generating predicted organic matter content and predicted water content based on the first feature vector and the second feature vector respectively; and the output layer is used for outputting the predicted organic matter content and the predicted water content.
4. The method of claim 1, wherein the soil mechanical composition prediction model comprises: the input layer is used for inputting hyperspectral data of a target soil area, the predicted organic matter content and the predicted water content; The cross-modal characteristic coding layer is used for respectively carrying out characteristic coding on the hyperspectral data, the predicted organic matter content and the predicted water content to obtain coding characteristics corresponding to the hyperspectral data, the predicted organic matter content and the predicted water content; The double-attention fusion module is used for carrying out weighted fusion on the hyperspectral data, the predicted organic matter content and the coding features corresponding to the predicted water content to obtain a global feature vector; The multi-scale particle feature extraction layer is used for respectively extracting sand features, powder features and sticky particle features and splicing the extracted sand features, powder features and sticky particle features into multi-scale particle feature vectors; the particle interaction calibration layer is used for calibrating the multi-scale particle feature vector to obtain a calibrated particle feature vector; The multi-target prediction head is used for obtaining predicted soil mechanical composition according to the calibrated particle characteristic vector, wherein the predicted soil mechanical composition comprises a sand content predicted value, a powder content predicted value and a clay content predicted value; and the output layer is used for outputting the predicted soil mechanical composition.
5. The method of claim 4, wherein the environmental data comprises soil pH, soil conductivity, soil fast acting nitrogen, phosphorus and potassium content, number of years of solar irradiation, annual average temperature precipitation, soil pollution status and soil water holding capacity.
6. A mycorrhiza restoration scheme determining apparatus, characterized in that the apparatus comprises: an acquisition unit for acquiring hyperspectral data and environmental data of a target soil region; the first prediction unit is used for inputting the hyperspectral data into a pre-trained soil organic matter and water content prediction model to obtain the predicted organic matter content and the predicted water content of the target soil area; the second prediction unit is used for inputting the hyperspectral data, the predicted organic matter content and the predicted water content into a pre-trained soil mechanical composition prediction model to obtain a predicted soil mechanical composition of the target soil region; the matching unit is used for matching a plurality of alternative mycorrhizas corresponding to the predicted soil mechanical composition of the target soil area from the configured association database of different soil mechanical compositions and different mycorrhizal suitability; The calculating unit is used for calculating the adaptation indexes of different alternative mycorrhiza according to various alternative mycorrhiza, the predicted soil mechanical composition and the environmental data, and comprises calculating the physical adaptation indexes and the functional adaptation indexes of different alternative mycorrhiza according to the environmental data and the predicted soil mechanical composition, taking the product of the physical adaptation indexes and the functional adaptation indexes of any alternative mycorrhiza as the adaptation index of the alternative mycorrhiza, taking the alternative mycorrhiza corresponding to the adaptation index higher than the configured adaptation threshold as a first mycorrhiza, cross-combining the different first mycorrhiza to obtain a plurality of first mycorrhiza combinations, and calculating the cooperative adaptation index of the different first mycorrhiza combinations; The third prediction unit is used for inputting the adaptation indexes of different alternative mycorrhiza, the predicted soil mechanical composition and the environmental data into a pre-trained mycorrhiza restoration effect prediction model to obtain restoration schemes and predicted restoration effects corresponding to different alternative mycorrhiza; The method comprises the steps of determining a mycorrhiza restoration scheme of a target soil area based on adaptive indexes of different alternative mycorrhiza and predicted restoration effects of corresponding restoration schemes, wherein the mycorrhiza restoration scheme comprises the steps of dynamically optimizing an agent by means of pre-trained inoculation parameters, determining optimal mycorrhiza by means of multiple rounds of iteration by means of dividing the comprehensive decision of the different second mycorrhiza or the different second mycorrhiza into fitness functions, deleting the first mycorrhiza or the first mycorrhiza combination with the adaptive indexes or the collaborative adaptive indexes not larger than configured adaptive thresholds or predicting the restoration effects not larger than configured restoration thresholds, obtaining different second mycorrhiza or different second mycorrhiza combinations, dynamically optimizing a configured multidimensional comprehensive decision model of the agent according to the inoculation parameters, obtaining the mycorrhiza restoration scheme of the target soil area by means of an optimizing algorithm by means of taking the different inoculation parameters in the restoration scheme as action spaces, and obtaining the mycorrhiza restoration scheme of the target soil area.
7. An electronic device, characterized in that the electronic device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are in communication with each other through the communication bus; a memory for storing a computer program; A processor for implementing the method of any of claims 1-5 when executing a program stored on a memory.
8. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a computer program which, when executed by a processor, implements the method of any of claims 1-5.

Description

Method, device, equipment and medium for determining mycorrhiza restoration scheme Technical Field The application relates to the technical field of ecological restoration data analysis, in particular to a method, a device, equipment and a medium for determining a mycorrhiza restoration scheme. Background The mining disturbed soil structure changes the mechanical composition of the soil, breaks the original proportion of powder particles, clay particles and sand particles of the soil, leads to the influence of the physical and chemical properties of the soil and is unfavorable for plant growth. The degradation of soil quality becomes a key problem for restricting ecological restoration and sustainable land utilization, and the traditional mycorrhiza restoration scheme is mainly dependent on manual experience to screen mycorrhizas and formulate restoration parameters, and has the defects of inaccurate soil parameter acquisition, one-sided matching of mycorrhiza suitability, difficult restoration effect prejudgement and the like, so that the mycorrhiza restoration scheme is insufficient in pertinence and low in restoration efficiency. Disclosure of Invention The embodiment of the application aims to provide a method, a device, equipment and a medium for determining a mycorrhiza restoration scheme, which are used for solving the problems in the prior art and automatically generating the mycorrhiza restoration scheme with targeted target areas. In a first aspect, a method for determining a mycorrhiza restoration scheme is provided, the method may include: acquiring hyperspectral data and environmental data of a target soil area; inputting the hyperspectral data into a pre-trained soil organic matter and water content prediction model to obtain the predicted organic matter content and the predicted water content of a target soil area; Inputting the hyperspectral data, the predicted organic matter content and the predicted water content into a pre-trained soil mechanical composition prediction model to obtain a predicted soil mechanical composition of the target soil region; Matching a plurality of alternative mycorrhizas corresponding to the predicted soil mechanical composition of the target soil area from a configured association database of different soil mechanical compositions and different mycorrhizal suitability; calculating the adaptation indexes of different alternative mycorrhiza according to the various alternative mycorrhiza, the predicted soil mechanical composition and the environmental data; Inputting the adaptation indexes, the predicted soil mechanical compositions and the environmental data of different alternative mycorrhiza into a pre-trained mycorrhiza restoration effect prediction model to obtain restoration schemes and predicted restoration effects corresponding to different alternative mycorrhiza; and determining the mycorrhiza restoration scheme of the target soil area based on the adaptation indexes of different alternative mycorrhiza and the prediction restoration effect of the corresponding restoration scheme. In an alternative implementation, the method for acquiring hyperspectral data of the target soil area includes: Acquiring original hyperspectral data of a target soil area; Preprocessing the original hyperspectral data to obtain first hyperspectral data; Fitting an upper envelope curve by adopting an envelope curve removing method, and dividing the first hyperspectral data by the envelope curve to obtain normalized second hyperspectral data; Intercepting third hyperspectral data positioned in a configured target band from the second hyperspectral data; And carrying out multiplication scattering correction and first derivative transformation on the third hyperspectral data to obtain fourth hyperspectral data, and outputting the fourth hyperspectral data as hyperspectral data of a target soil area. In an alternative implementation, the soil organic matter and water content prediction model includes: An input layer for inputting hyperspectral data of a target soil region; The physical attention embedding layer is used for multiplying the input hyperspectral data with the configured binary attention mask to obtain mask-optimized one-dimensional spectral data; The shared feature extraction layer is used for extracting features of the one-dimensional spectrum data subjected to mask optimization to obtain a multi-dimensional shared feature vector; The feature refinement extraction layer is used for performing refinement feature extraction on the multidimensional shared feature vectors respectively to obtain a first feature vector and a second feature vector; The double-target fusion prediction layer is used for generating predicted organic matter content and predicted water content based on the first feature vector and the second feature vector respectively; and the output layer is used for outputting the predicted organic matter content and the predicted water content. In an alternati