Farmland environment intelligent monitoring system based on multi-source data fusion

Abstract

The invention relates to the technical field of environment monitoring, in particular to a farmland environment intelligent monitoring system based on multi-source data fusion, which comprises a multi-source data acquisition module for synchronously acquiring original data; the dynamic compensation preprocessing module is used for receiving the original data and generating standard data with aligned time axes through a dynamic timestamp compensation mechanism; the spatial-temporal feature decoupling module is used for performing multi-scale feature extraction on the standardized data and separating spatial distribution features, time sequence features and environment coupling features; the multi-modal fusion analysis module is used for fusing the spatial distribution characteristics, the time sequence characteristics and the environment coupling characteristics by adopting an evidence chain fusion algorithm to generate farmland environment state evaluation parameters; and the edge decision optimization module is used for dynamically adjusting a resource allocation strategy of an edge computing node according to the farmland environment state evaluation parameters. Priority processing of key tasks is achieved under limited computing power, and the response efficiency and the resource utilization rate of the system are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and in particular to an intelligent farmland environment monitoring system based on multi-source data fusion. Background Art

[0002] Current agricultural production has put forward higher requirements for the accuracy and timeliness of environmental monitoring. Traditional farmland environmental monitoring systems mostly rely on a single sensor type or fixed-period sampling method, and cannot fully perceive the multidimensional environmental factors in complex agricultural ecosystems, such as the dynamic coupling relationship between soil properties, climate change and crop growth status; in addition, most existing systems rely on central servers for unified calculations in data processing, and have problems such as large data transmission delays, slow response speeds, and low resource utilization efficiency. It is especially difficult to achieve real-time, accurate, distributed perception and decision support in large-scale farmland scenarios.

[0003] Although edge computing and multimodal data fusion technologies have been gradually applied to the agricultural field in recent years, they still face challenges such as difficulty in unified alignment of multi-source heterogeneous data, insufficient feature decoupling, lack of semantic interpretation ability of fusion algorithms, and lack of intelligent perception in edge node resource scheduling. There is also a lack of a systematic method that can effectively extract key features and realize multimodal fusion analysis on the basis of ensuring the spatiotemporal consistency of monitoring data, and intelligently optimize edge computing resource allocation and result visualization according to the dynamic changes of farmland environment. Summary of the Invention

[0004] The present invention provides an intelligent farmland environment monitoring system based on multi-source data fusion to enhance the intelligence level and emergency response capability of agricultural ecological management.

[0005] An intelligent farmland environment monitoring system based on multi-source data fusion includes the following modules: Multi-source data acquisition module: includes soil parameter sensor groups, meteorological sensor groups, and crop growth sensor groups deployed at farmland monitoring nodes to synchronously collect raw data; Dynamic compensation preprocessing module: receives the raw data from the multi-source data acquisition module and generates time-axis aligned standardized data through a dynamic timestamp compensation mechanism; Spatiotemporal feature decoupling module: performs multi-scale feature extraction on the standardized data to separate spatial distribution features, time series features and environmental coupling features; Multimodal fusion analysis module: uses the evidence chain fusion algorithm to fuse the spatial distribution characteristics, time series characteristics and environmental coupling characteristics to generate farmland environmental status assessment parameters; Edge decision optimization module: dynamically adjusts the resource allocation strategy of edge computing nodes according to the farmland environmental status assessment parameters.

[0006] Optionally, the multi-source data acquisition module includes: Farmland area grid division and candidate deployment point selection: The farmland area is divided into several spatial cells. Based on the Voronoi region division method, a set of candidate deployment points with high edge stability and strong representativeness is selected; Calculation of optimal deployment locations based on coverage optimization model: Build a coverage optimization model to determine the optimal deployment locations for the soil parameter sensor group, meteorological sensor group, and crop growth sensor group; Optimizing the coverage plan under the minimum deployment amount: Using a genetic algorithm to solve the above optimization model, the maximum coverage combination under the minimum deployment amount is output to form a deployment coordinate diagram; Fixed-point deployment implementation of multiple types of sensor groups: soil parameter sensor groups, meteorological sensor groups, and crop growth sensor groups are installed at designated points according to the deployment coordinate map.

[0007] Optionally, the dynamic compensation preprocessing module includes: Multi-source raw timestamp extraction: extract the timestamp sequences of the raw data from the soil parameter sensor group, meteorological sensor group, and crop growth sensor group respectively; Dynamic time offset correction function construction: Build a time offset correction function to dynamically compensate the data timestamps of each type of sensor; Multi-sensor timestamp alignment processing: compensate the original data timestamp based on the correction function to generate a time-aligned data sequence; Standardized data generation under a unified time axis: All sensor data sequences are interpolated or resampled to form a standardized data set under a unified time axis.

[0008] Optionally, the spatiotemporal feature decoupling module includes: Sliding window reconstruction and scale set construction: On the unified time axis, a sliding window with a length of The sliding window of is used to reconstruct the standardized data sequence and generate a multi-scale sample set; Time series feature extraction: Based on the obtained multi-scale sample set, a one-dimensional discrete wavelet transform is performed on each dimension of each sliding window sample matrix to extract multi-resolution time series features; Spatial distribution feature extraction: based on the data vector at each time point ,The principal component analysis method based on neighboring location perception is used to extract the spatial synergistic variation characteristics among observations from different sensors; Environmental coupling feature extraction: Based on the dynamic correlation between different modes, the environmental coupling coefficient matrix is ​​constructed and the eigenvalues ​​are extracted as environmental coupling features.

[0009] Optionally, the multimodal fusion analysis module includes: Modal feature normalization and weight assignment: The spatial distribution features, time series features, and environmental coupling features are normalized separately to the same scale range. The modal weights are dynamically calculated based on the correlation between each feature and the historical environmental state. Evidence fusion reasoning and state assessment generation: The normalized feature vector is multiplied by the corresponding modal weight to obtain the modal confidence distribution, and based on the weighted fusion strategy, the farmland environment state assessment parameters at the current time point are output.

[0010] Optionally, the modal feature normalization and weight allocation includes: Multimodal feature normalization: normalize the spatial distribution features, time series features, and environmental coupling features to construct a unified modal feature vector; Construct a modal evidence weight matrix: dynamically calculate the modal weight based on the correlation between various features and historical environmental states.

[0011] Optionally, the evidence fusion reasoning and state assessment generation includes: Calculate the modal evidence confidence distribution: map the fused feature vector to the modal confidence space to form the modal evidence distribution function; Generate farmland environmental status assessment parameters: Calculate the comprehensive assessment function based on the modal confidence output and eigenvector.

[0012] Optionally, the edge decision optimization module includes: Priority calculation and resource allocation modeling: Calculate the corresponding scheduling priorities based on the farmland environmental status assessment parameters and build a resource allocation optimization model; Resource allocation result generation and execution: The optimal resource allocation plan at each time point is obtained by solving the resource allocation optimization model, and the optimal resource allocation plan is sent to the edge computing node.

[0013] Optionally, the priority calculation and resource allocation modeling includes: Resource scheduling priority function construction: Based on the environmental status evaluation parameters at each time point, define the edge node scheduling priority function; Edge resource allocation model construction and optimization: Build a resource allocation optimization model based on scheduling priorities.

[0014] Optionally, the resource allocation result generation and issuance includes: Resource allocation result generation: Use constrained least squares method to solve and output the optimal resource allocation solution vector; Resource distribution model expression: the optimal resource allocation vector obtained It is sent to each edge computing node through the edge control interface.

[0015] Beneficial effects of the present invention: The present invention provides an intelligent farmland environment monitoring system based on multi-source data fusion. By deploying multiple types of sensors such as soil parameters, meteorological indicators, and crop growth status, it achieves comprehensive perception of the farmland environment; uses a dynamic timestamp compensation mechanism to uniformly align asynchronously collected data to improve data temporal consistency; and by constructing a multi-scale feature decoupling model, it extracts spatial distribution characteristics, time series characteristics, and environmental coupling characteristics respectively, effectively enhancing the system's ability to characterize complex agricultural ecological changes.

[0016] This invention uses environmental status assessment parameters as the driving basis for edge scheduling optimization and constructs a resource allocation model based on priority mapping, enabling edge computing nodes to dynamically adjust resource allocation strategies according to environmental changes, and achieve priority processing of critical tasks under limited computing power, effectively improving the system's response efficiency and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 is a system module diagram of an embodiment of the present invention; Figure 2 This is a data collection diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0020] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0021] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0022] like Figure 1-Figure 2 As shown in the figure, an intelligent farmland environment monitoring system based on multi-source data fusion includes the following modules: Multi-source data acquisition module: includes soil parameter sensor groups, meteorological sensor groups, and crop growth sensor groups deployed at farmland monitoring nodes to synchronously collect raw data; Dynamic compensation preprocessing module: receives the raw data from the multi-source data acquisition module and generates time-axis aligned standardized data through a dynamic timestamp compensation mechanism; Spatiotemporal feature decoupling module: extracts multi-scale features from standardized data to separate spatial distribution features, time series features, and environmental coupling features; Multimodal fusion analysis module: uses the evidence chain fusion algorithm to fuse spatial distribution characteristics, time series characteristics, and environmental coupling characteristics to generate farmland environmental status assessment parameters; Edge decision optimization module: Dynamically adjusts the resource allocation strategy of edge computing nodes based on farmland environmental status assessment parameters.

[0023] The multi-source data acquisition module includes: Farmland area grid division and candidate deployment point selection: The farmland area is divided into several spatial cells. Based on the Voronoi region division method, a set of candidate deployment points with high edge stability and strong representativeness is selected; Calculation of optimal deployment locations based on the coverage optimization model: A coverage optimization model is constructed to determine the optimal deployment locations for the soil parameter sensor group, meteorological sensor group, and crop growth sensor group. The objective function of the coverage optimization model is expressed as: ; in, represents the total number of candidate deployment points, Indicates the number of all space cells to be covered, For deployment variables, when When sensors are deployed at different locations, , otherwise 0, Indicates the The importance weight of each cell is calculated based on soil heterogeneity, crop growth activity and climate sensitivity. Deployment point For cells The degree of perceived coverage is modeled based on the distance attenuation function. A deployment cost adjustment factor to control the trade-off between perceived coverage quality and cost; Optimizing the coverage plan under the minimum deployment amount: Using a genetic algorithm to solve the above optimization model, the maximum coverage combination under the minimum deployment amount is output to form a deployment coordinate diagram; Fixed-point deployment of multiple sensor groups: soil parameter sensor groups (monitoring pH value, electrical conductivity, soil temperature and humidity), meteorological sensor groups (monitoring temperature, humidity, wind speed, precipitation), and crop growth sensor groups (monitoring chlorophyll content, stem height, and crown width) are installed at designated locations according to the deployment coordinate map.

[0024] The dynamic compensation preprocessing module includes: Multi-source raw timestamp extraction: extract the timestamp sequence of the raw data from the soil parameter sensor group, meteorological sensor group and crop growth sensor group respectively ,in, Corresponding to three types of sensor groups, Indicates the time sampling point sequence number; Dynamic time offset correction function construction: Constructing time offset correction function , which is used to dynamically compensate the data timestamp of each type of sensor, and is expressed as: ; in, is the time offset correction function, Indicates the a unified reference time point (e.g., the center time of the edge server aggregation window), For the Sensor group at sampling point The original timestamp of For the The time synchronization weight factor of the sensor group is dynamically estimated based on the average transmission delay and jitter of the sensor network; Timestamp alignment of multiple sensors: The original data timestamps are compensated based on the correction function to generate a time-aligned data sequence, which is expressed as: ; in, After compensation Sensors at sampling points Alignment timestamps; Standardized data generation under a unified time axis: All sensor data sequences are interpolated or resampled to form a standardized data set under a unified time axis, which is expressed as: ; in, Indicates the Class sensor groups at the alignment time point The observed value of For the A multi-source standardized data vector at a unified time point.

[0025] The spatiotemporal feature decoupling module includes: Sliding window reconstruction and scale set construction: On the unified time axis, a sliding window with a length of The sliding window of is used to reconstruct the standardized data sequence and generate a multi-scale sample set, which is expressed as: ; in, For the Normalized data vectors at time points, For the A sliding window sample matrix, For multi-source data dimensions, is the window length, is the total number of time points; Time series feature extraction: Based on the obtained multi-scale sample set, a one-dimensional discrete wavelet transform (DWT) is performed on each dimension of each sliding window sample matrix to extract multi-resolution time series features, which can be expressed as: ; in, Represents the discrete wavelet transform operation, using Daubechies basis wavelet, For the The time characteristic vector of each sample reflects the periodic fluctuation and mutation response characteristics; Spatial distribution feature extraction: based on the data vector at each time point , the principal component analysis (Spatial-PCA) method based on proximity perception is used to extract the spatial co-variation characteristics between observations of different sensors, which can be expressed as: ; in, Indicates a time point The corresponding spatial neighborhood sampling point set, Represents the principal component analysis operation, extracting the maximum covariance principal axis direction, For the The spatial feature vector corresponding to each time point; Environmental coupling feature extraction: Based on the dynamic correlation between different modes, the environmental coupling coefficient matrix is ​​constructed, and the eigenvalues ​​are extracted as environmental coupling features, which are expressed as: ; ; in, represents the covariance operation, For the Class data at time point The estimated standard deviation of For time point The environmental coupling coefficient matrix, represents the maximum eigenvalue of the matrix, which is used as a quantitative indicator of the coupling strength. For time point environmental coupling characteristics.

[0026] The multimodal fusion analysis module includes: Modal feature normalization and weight assignment: The spatial distribution features, time series features, and environmental coupling features are normalized separately to the same scale range. The modal weights are dynamically calculated based on the correlation between each feature and the historical environmental state. Evidence fusion reasoning and state assessment generation: The normalized feature vector is multiplied by the corresponding modal weight to obtain the modal confidence distribution, and based on the weighted fusion strategy, the farmland environment state assessment parameters at the current time point are output.

[0027] Modal feature normalization and weight allocation include: Multimodal feature normalization processing: normalize the spatial distribution features, time series features, and environmental coupling features to construct a unified modal feature vector, which is expressed as: ; in, Represents the maximum and minimum normalization function, which maps all feature distributions to the interval , For the The fusion input vector of each time point uniformly represents three types of features. is the total number of time points in the current cycle; Constructing the modal evidence weight matrix: Based on the correlation between various features and historical environmental states, the modal weight is dynamically calculated and expressed as: ; in, , For the The correlation coefficient between the class features and the historical environment state labels (such as the Pearson correlation coefficient), The modal evidence weight matrix that assigns the degree of trust to the three types of modalities satisfies .

[0028] Evidence fusion reasoning and state assessment generation include: Calculate the modal evidence confidence distribution: Map the fused feature vector to the modal confidence space to form the modal evidence distribution function, which is expressed as: ; in, is a standard normalized exponential function used to construct a probability confidence distribution. Represents the confidence output of the three modes at the current moment, is the total number of time points in the current cycle, It is The confidence vector of modal evidence at each moment; Generate farmland environmental status assessment parameters: Based on the modal confidence output and eigenvector, calculate the comprehensive assessment function, which is expressed as: ; in, For the The farmland environmental status assessment parameter at each time point represents the comprehensive environmental health status score, with a range of values. , the higher the value, the more stable or ideal the farmland environment is. After normalization, Class eigenvalues, is the modal assurance vector, It is Class modality at time point The confidence value of .

[0029] The edge decision optimization module includes: Priority calculation and resource allocation modeling: Calculate the corresponding scheduling priorities based on the farmland environmental status assessment parameters and build a resource allocation optimization model; Resource allocation result generation and execution: The optimal resource allocation plan at each time point is obtained by solving the resource allocation optimization model, and the optimal resource allocation plan is sent to the edge computing node.

[0030] Priority calculation and resource allocation modeling include: Resource scheduling priority function construction: Based on the environmental status evaluation parameters at each time point, the edge node scheduling priority function is defined, which is expressed as: ; in, is the aforementioned farmland environmental status assessment parameter. The lower the value, the higher the degree of environmental abnormality. For time point The scheduling priority coefficient on , the closer it is to 1, the more priority computing resources should be allocated; Edge resource allocation model construction and optimization: Based on the scheduling priority, a resource allocation optimization model is constructed, which is expressed as: ; in, Indicates the assignment to The amount of edge computing resources (such as CPU cycles, memory blocks) in a time window, Indicates the total amount of edge resources available for allocation in the current period. The optimization goal is to make the resource allocation ratio and priority The deviation between the minimum is the total number of time points in the current cycle.

[0031] The generation and distribution of resource allocation results include: Resource allocation result generation: The constrained least squares method is used to solve the problem and output the optimal resource allocation solution vector, which is expressed as: ; Among them, each element Indicates the assignment to The optimal amount of resources at a point in time; Resource distribution model expression: the optimal resource allocation vector obtained It is sent to each edge computing node through the edge control interface and is expressed as: ; in, Indicates that each time point Resource allocation value They are transmitted to the corresponding edge nodes respectively to guide the thread resource configuration of various monitoring, processing and response tasks.

[0032] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0033] The above are only preferred embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. An intelligent farmland environment monitoring system based on multi-source data fusion, characterized in that: Includes the following modules: Multi-source data acquisition module: includes soil parameter sensor groups, meteorological sensor groups, and crop growth sensor groups deployed at farmland monitoring nodes to synchronously collect raw data; Dynamic compensation preprocessing module: receives the raw data from the multi-source data acquisition module and generates time-axis aligned standardized data through a dynamic timestamp compensation mechanism; Spatiotemporal feature decoupling module: performs multi-scale feature extraction on the standardized data to separate spatial distribution features, time series features and environmental coupling features; Multimodal fusion analysis module: uses the evidence chain fusion algorithm to fuse the spatial distribution characteristics, time series characteristics and environmental coupling characteristics to generate farmland environmental status assessment parameters; Edge decision optimization module: dynamically adjusts the resource allocation strategy of edge computing nodes according to the farmland environmental status assessment parameters.

2. The farmland environment intelligent monitoring system based on multi-source data fusion according to claim 1 is characterized in that: The multi-source data acquisition module includes: Farmland area grid division and candidate deployment point selection: The farmland area is divided into several spatial cells. Based on the Voronoi region division method, a set of candidate deployment points with high edge stability and strong representativeness is selected; Calculation of optimal deployment locations based on coverage optimization model: Build a coverage optimization model to determine the optimal deployment locations for the soil parameter sensor group, meteorological sensor group, and crop growth sensor group; Optimizing the coverage plan under the minimum deployment amount: Using a genetic algorithm to solve the above optimization model, the maximum coverage combination under the minimum deployment amount is output to form a deployment coordinate diagram; Fixed-point deployment implementation of multiple types of sensor groups: soil parameter sensor groups, meteorological sensor groups, and crop growth sensor groups are installed at designated points according to the deployment coordinate map.

3. The intelligent farmland environment monitoring system based on multi-source data fusion according to claim 2 is characterized in that: The dynamic compensation preprocessing module includes: Multi-source raw timestamp extraction: extract the timestamp sequences of the raw data from the soil parameter sensor group, meteorological sensor group, and crop growth sensor group respectively; Dynamic time offset correction function construction: Build a time offset correction function to dynamically compensate the data timestamps of each type of sensor; Multi-sensor timestamp alignment processing: compensate the original data timestamp based on the correction function to generate a time-aligned data sequence; Standardized data generation under a unified time axis: All sensor data sequences are interpolated or resampled to form a standardized data set under a unified time axis.

4. The farmland environment intelligent monitoring system based on multi-source data fusion according to claim 3 is characterized in that: The spatiotemporal feature decoupling module includes: Sliding window reconstruction and scale set construction: On the unified time axis, a sliding window with a length of The sliding window of is used to reconstruct the standardized data sequence and generate a multi-scale sample set; Time series feature extraction: Based on the obtained multi-scale sample set, a one-dimensional discrete wavelet transform is performed on each dimension of each sliding window sample matrix to extract multi-resolution time series features; Spatial distribution feature extraction: based on the data vector at each time point ,The principal component analysis method based on neighboring location perception is used to extract the spatial synergistic variation characteristics among observations from different sensors; Environmental coupling feature extraction: Based on the dynamic correlation between different modes, the environmental coupling coefficient matrix is ​​constructed and the eigenvalues ​​are extracted as environmental coupling features.

5. The farmland environment intelligent monitoring system based on multi-source data fusion according to claim 4 is characterized in that: The multimodal fusion analysis module includes: Modal feature normalization and weight assignment: The spatial distribution features, time series features, and environmental coupling features are normalized separately to the same scale range. The modal weights are dynamically calculated based on the correlation between each feature and the historical environmental state. Evidence fusion reasoning and state assessment generation: The normalized feature vector is multiplied by the corresponding modal weight to obtain the modal confidence distribution, and based on the weighted fusion strategy, the farmland environment state assessment parameters at the current time point are output.

6. The intelligent farmland environment monitoring system based on multi-source data fusion according to claim 5 is characterized in that: The modal feature normalization and weight allocation include: Multimodal feature normalization: normalize the spatial distribution features, time series features, and environmental coupling features to construct a unified modal feature vector; Construct a modal evidence weight matrix: dynamically calculate the modal weight based on the correlation between various features and historical environmental states.

7. The farmland environment intelligent monitoring system based on multi-source data fusion according to claim 5 is characterized in that: The evidence fusion reasoning and state assessment generation include: Calculate the modal evidence confidence distribution: map the fused feature vector to the modal confidence space to form the modal evidence distribution function; Generate farmland environmental status assessment parameters: Calculate the comprehensive assessment function based on the modal confidence output and eigenvector.

8. The intelligent farmland environment monitoring system based on multi-source data fusion according to claim 7 is characterized in that: The edge decision optimization module includes: Priority calculation and resource allocation modeling: Calculate the corresponding scheduling priorities based on the farmland environmental status assessment parameters and build a resource allocation optimization model; Resource allocation result generation and execution: The optimal resource allocation plan at each time point is obtained by solving the resource allocation optimization model, and the optimal resource allocation plan is sent to the edge computing node.

9. The intelligent farmland environment monitoring system based on multi-source data fusion according to claim 8 is characterized in that: The priority calculation and resource allocation modeling include: Resource scheduling priority function construction: Based on the environmental status evaluation parameters at each time point, define the edge node scheduling priority function; Edge resource allocation model construction and optimization: Build a resource allocation optimization model based on scheduling priorities.

10. The farmland environment intelligent monitoring system based on multi-source data fusion according to claim 8, characterized in that: The resource allocation result generation and issuance execution includes: Resource allocation result generation: Use constrained least squares method to solve and output the optimal resource allocation solution vector; Resource distribution model expression: the optimal resource allocation vector obtained It is sent to each edge computing node through the edge control interface.

Images