Multi-sensor fusion greenhouse environment data monitoring method and platform

Abstract

The invention relates to the technical field of greenhouse monitoring, in particular to a greenhouse environment data monitoring method and platform based on multi-sensor fusion. Receiving soil environment monitoring time sequence information, air environment monitoring time sequence information and crop growth information of the target area; extracting initial time soil environment information of the air environment monitoring time sequence information from the environment monitoring historical information; taking the soil environment information at the initial moment as an initial soil state constraint, taking the air environment monitoring time sequence information and the crop growth information as soil state influence constraints, collecting a soil state monitoring sample set, and performing statistics on soil environment prediction time sequence information; and when the soil environment prediction time sequence information and the soil environment monitoring time sequence information are inconsistent, performing suspicious identification on the soil environment monitoring time sequence information and the air environment monitoring time sequence information, and sending the information to a user side. The technical effect of improving the stability and reliability of greenhouse environment data monitoring is achieved.

Description

Technical Field

[0001] The present invention relates to the field of greenhouse monitoring technology, and in particular to a greenhouse environment data monitoring method and platform using multi-sensor fusion. Background Art

[0002] Traditional greenhouse environmental data detection involves collecting data multiple times at the same location and taking the average of these data as the final result to avoid errors introduced by a single measurement. This approach fails to consider the potential for errors in sensor hardware, regardless of how many times data is collected. Furthermore, sensor accuracy errors are difficult to monitor, making it difficult to ensure the stability of greenhouse environmental data detection. Summary of the Invention

[0003] The present invention aims to solve the technical problem that the stability of greenhouse environmental data detection is difficult to ensure in the existing technology, and provides a greenhouse environmental data monitoring method and platform with multi-sensor fusion to solve the problem.

[0004] The technical solution of the present invention to solve the above technical problems is as follows: In a first aspect, the present invention provides a greenhouse environmental data monitoring method with multi-sensor fusion, comprising: receiving soil environment monitoring time series information, air environment monitoring time series information and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth in the environmental monitoring time window; extracting the soil environment information at the initial moment of the air environment monitoring time series information from the environmental monitoring historical information; using the soil environment information at the initial moment as the initial soil state constraint, and using the air environment monitoring time series information and the crop growth information as the soil state impact constraints, collecting a soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information; when the soil environment prediction time series information is inconsistent with the soil environment monitoring time series information, marking the soil environment monitoring time series information and the air environment monitoring time series information as suspicious and sending them to a user end.

[0005] Optionally, taking the soil environment information at the initial moment as the initial soil state constraint, taking the air environment monitoring time series information and the crop growth information as the soil state impact constraint, collecting a soil state monitoring sample set, and counting the soil environment prediction time series information, including: performing spatial clustering on the soil environment information at the initial moment to obtain the soil environment information of the first region at the initial moment up to the soil environment information of the Nth region at the initial moment; partitioning the air environment monitoring time series information based on the first region up to the Nth region to obtain the air environment monitoring time series information of the first region up to the air environment monitoring time series information of the Nth region; traversing the air environment monitoring time series information of the first region up to the air environment monitoring time series information of the Nth region to perform spatiotemporal clustering to obtain the spatiotemporal clustering result of the air environment of the first region up to the spatiotemporal clustering result of the air environment of the Nth region; taking the soil environment information of the first region at the initial moment as the initial soil state constraint, taking the crop growth information and the spatiotemporal clustering result of the air environment of the first region as the soil state impact constraint, collecting the first soil state monitoring sample set, and counting the soil environment prediction time series information of the first region; The soil environment information of the Nth region at the initial moment is used as the initial soil state constraint, the crop growth information and the spatiotemporal clustering result of the air environment of the Nth region are used as the soil state impact constraint, the Nth soil state monitoring sample set is collected, and the soil environment prediction time series information of the Nth region is counted; the soil environment prediction time series information of the first region is added to the soil environment prediction time series information until the soil environment prediction time series information of the Nth region.

[0006] Among them, traversing the first area air environment monitoring time series information until the Nth area air environment monitoring time series information is subjected to spatiotemporal clustering to obtain the first area air environment spatiotemporal clustering result until the Nth area air environment spatiotemporal clustering result, including: performing spatial domain clustering on the first area air environment monitoring time series information to obtain the first sub-area air environment monitoring time series information until the Mth sub-area air environment monitoring time series information; performing time domain clustering on the first sub-area air environment monitoring time series information to obtain the first sub-area first time zone air environment information until the first sub-area Yth time zone air environment information; until performing time domain clustering on the Mth sub-area air environment monitoring time series information to obtain the Mth sub-area first time zone air environment information until the Mth sub-area Zth time zone air environment information; adding the first sub-area first time zone air environment information until the first sub-area Yth time zone air environment information until the Mth sub-area first time zone air environment information until the Mth sub-area Zth time zone air environment information into the first area air environment spatiotemporal clustering result.

[0007] Among them, the soil environment information of the first area at the initial moment is used as the initial soil state constraint, the crop growth information and the spatiotemporal clustering result of the air environment of the first area are used as the soil state impact constraint, the first soil state monitoring sample set is collected, and the soil environment prediction time series information of the first area is counted, including: from the spatiotemporal clustering result of the air environment of the first area, the air environment information of the first time zone of the first sub-area is extracted until the air environment information of the Yth time zone of the first sub-area, until the air environment information of the first time zone of the Mth sub-area is extracted until the air environment information of the Zth time zone of the Mth sub-area; the crop planting time and crop variety are extracted from the crop growth information; the soil environment information of the first area at the initial moment is used as the initial soil state constraint, and the crop planting time, the crop variety and the air environment information of the first time zone of the first sub-area are used as constraints, the first sub-soil state monitoring sample set is collected, and the end time of the first time zone of the first sub-area is counted. soil environment prediction information; until the soil environment prediction information of the first sub-region at the end moment of the Y-1 time zone is used as the initial soil state constraint, and the crop planting time, the crop variety and the air environment information of the first sub-region at the Y time zone are used as constraints, the Y-th sub-soil state monitoring sample set is collected, and the soil environment prediction information of the first sub-region at the end moment of the Y time zone is counted; the soil environment prediction information of the first sub-region at the end moment of the first time zone to the soil environment prediction information of the first sub-region at the end moment of the Y time zone is added to the soil environment prediction time series information of the first sub-region; until the soil environment prediction information of the M-th sub-region at the end moment of the first time zone to the soil environment prediction information of the M-th sub-region at the end moment of the Z time zone is obtained, and added to the soil environment prediction time series information of the M-th sub-region; the soil environment prediction time series information of the first sub-region to the soil environment prediction time series information of the M sub-region is added to the soil environment prediction time series information of the first region.

[0008] Optionally, when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious, including: extracting the first area soil environment prediction time series information from the soil environment prediction time series information, wherein the first area soil environment prediction time series information includes the first sub-area soil environment prediction time series information to the Mth sub-area soil environment prediction time series information; extracting the first area soil environment monitoring time series information from the soil environment monitoring time series information, wherein the first area soil environment monitoring time series information includes the first sub-area soil environment monitoring time series information to the Mth sub-area soil environment monitoring time series information; marking the first sub-area soil environment prediction time series information and the first A similarity analysis is performed on the sub-region soil environment monitoring time series information to obtain the soil environment similarity of the first sub-region; until a similarity analysis is performed on the soil environment prediction time series information of the Mth sub-region and the soil environment monitoring time series information of the Mth sub-region to obtain the soil environment similarity of the Mth sub-region; a consistency comparison rule is constructed: when the soil environment similarity is greater than the similarity threshold, the soil environment is deemed consistent, otherwise it is deemed inconsistent; based on the consistency comparison rule, the soil environment similarity of the first sub-region is processed until the soil environment similarity of the Mth sub-region is reached to obtain an inconsistent sub-region; based on the inconsistent sub-region, the soil environment monitoring time series information and the air environment monitoring time series information are suspiciously marked to obtain the first region suspicious marking data, and added to the full region suspicious marking data.

[0009] Among them, it also includes: counting the proportion of the number of inconsistent sub-regions in the total number of sub-regions in the first region, setting it as the first region inconsistent proportion; when the first region inconsistent proportion is greater than the inconsistent proportion threshold, based on the inconsistent sub-regions, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious; when the first region inconsistent proportion is less than or equal to the inconsistent proportion threshold, the air environment monitoring time series information is marked as suspicious.

[0010] Among them, performing similarity analysis on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain the first sub-region soil environment similarity includes: extracting the first moment soil environment prediction value from the first sub-region soil environment prediction time series information; extracting the first moment soil environment monitoring value from the first sub-region soil environment monitoring time series information; performing same attribute standard deviation calculation on the first moment soil environment prediction value and the first moment soil environment monitoring value to obtain several attribute standard deviations; traversing the several attribute standard deviations to perform dimensionless processing to obtain several deviation eigenvalues; using the several deviation eigenvalues ​​as several dimensional distance parameters, calculating the Euclidean distance, and then taking the inverse to obtain the first sub-region soil environment similarity.

[0011] In a second aspect, the present invention provides a greenhouse environment data monitoring platform with multi-sensor fusion, comprising: A monitoring information acquisition module is configured to receive soil environment monitoring time series information, air environment monitoring time series information, and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth during the environmental monitoring time window; An environmental information extraction module is used to extract soil environmental information at the initial moment of the air environment monitoring time series information from environmental monitoring historical information; a prediction information statistics module, configured to collect a soil state monitoring sample set and to collect statistics on the soil environment prediction time series information, using the soil environment information at the initial moment as an initial soil state constraint, and using the air environment monitoring time series information and the crop growth information as soil state impact constraints; The monitoring information identification module is used to mark the soil environment monitoring time series information and the air environment monitoring time series information as suspicious when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, and send them to the user end.

[0012] By implementing this invention, it is possible to receive time-series information from soil and air environment monitoring, as well as crop growth information, for a target area. The crop growth information represents the extent of crop growth during the environmental monitoring time window, enabling comprehensive perception of the greenhouse environment and providing basic data support for subsequent data analysis and prediction. The integration of multi-dimensional data can more comprehensively reflect the environmental conditions for crop growth, avoiding the limitations of single data. By implementing the present invention, it is possible to extract soil environmental information at the initial moment of the air environmental monitoring time series information from historical environmental monitoring information, providing an initial reference point for soil environmental prediction, giving the prediction process a clear starting point, and improving the accuracy and rationality of the prediction. Extracting the initial state based on historical information can leverage past data accumulation and enhance the reliability of the prediction. By implementing the present invention, it is possible to use the initial soil environment information as the initial soil state constraint, the air environment monitoring time series information and the crop growth information as the soil state impact constraint, collect a soil state monitoring sample set, and statistically analyze the soil environment prediction time series information. Through the comprehensive consideration of multiple constraints, soil environment prediction can be made more scientific and accurate. Partitioning and clustering processing can adapt to the differences in the environments of different areas within the greenhouse, improve the pertinence and meticulousness of the prediction, and provide a reliable prediction basis for subsequent consistency judgment. By implementing the present invention, it can be achieved that when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious and sent to the user end, so that possible abnormal data can be discovered in time, and the user is prompted to check the relevant sensors or monitoring data, which helps to troubleshoot problems such as sensor hardware abnormalities as soon as possible and ensure the stability and accuracy of greenhouse environment data monitoring.

[0013] In summary, by implementing the present invention, the technical effect of improving the stability and reliability of greenhouse environment data detection can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic flow chart of a greenhouse environment data monitoring method using multi-sensor fusion provided by the present invention; Figure 2 This is a structural diagram of a greenhouse environment data monitoring platform with multi-sensor fusion provided by the present invention.

[0015] In the accompanying drawings, the components represented by the reference numerals are as follows: Monitoring information acquisition module 11, environmental information extraction module 12, prediction information statistics module 13, monitoring information identification module 14. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, features specified as "first" or "second" may explicitly or implicitly include one or more features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0018] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0019] Example 1, as Figure 1 As shown, the embodiment of the present invention provides a greenhouse environment data monitoring method and platform using multi-sensor fusion, including: S100: receiving soil environment monitoring time series information, air environment monitoring time series information, and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth in the environmental monitoring time window; S200: Extracting soil environment information at the initial moment of the air environment monitoring time series information from the environmental monitoring history information; S300: Using the soil environment information at the initial moment as an initial soil state constraint, and using the air environment monitoring time series information and the crop growth information as soil state impact constraints, collecting a soil state monitoring sample set, and compiling soil environment prediction time series information; S400: When the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious and sent to the user end.

[0020] In step S100 of this embodiment, the target area's soil and air environment monitoring time-series information, as well as crop growth information, is received. This provides comprehensive, multi-dimensional data support for greenhouse environmental data monitoring and analysis. By integrating dynamic information about soil, air, and crop growth, a complete environment-crop correlation data system is constructed, laying the data foundation for subsequent steps such as soil environment prediction and anomaly identification.

[0021] The soil environment monitoring time series information and the air environment monitoring time series information can be obtained through a certain type of intelligent observation instrument.

[0022] Among them, the collection of soil environmental monitoring time series information can be obtained using the soil sensor extended by the intelligent observer. The sensor can measure soil temperature (℃); volumetric water content (RH); conductivity (μS / cm), pH value and other parameters, and record time series data at a set frequency, such as once an hour.

[0023] The time series information of air environment monitoring can be collected by the sensors of the smart observer itself, including temperature (℃), humidity (RH), light intensity (Lux), CO2 concentration (ppm), etc., and transmitted to the cloud in real time through 4G or Star Flash network to form time series information.

[0024] Crop growth information can be manually recorded, such as by recording the planting date and crop variety, or acquired through device expansion modules, such as growth images captured by cameras or chlorophyll content analyzed by spectrometers, to characterize the crop's growth stage and growth potential within a specific time window. For example, a set of crop growth information might be {monitoring time window: 2025-07-20; crop variety: tomato; planting time: 45 days; plant height: 65.2; number of leaves: 18; fruit development stage: flowering}.

[0025] The soil environment monitoring time series information and air environment monitoring time series information collected by the sensor are then transmitted to the cloud platform via 4G Cat.1 or Star Flash communication, while receiving crop growth information manually entered or uploaded by the expansion module.

[0026] The cloud platform aligns timestamps on multi-source data to ensure that soil and air data are associated with crop growth information within the same time window, forming a structured data set and providing a unified data input format for subsequent steps.

[0027] In step S200 of this embodiment of the present application, soil environmental information at the initial moment of the air environment monitoring time series information is extracted from the environmental monitoring historical information. This is to provide an initial baseline state for soil environmental prediction. By extracting the soil environmental data corresponding to the starting moment of the air environment monitoring time series information, a time starting point anchor point for the prediction model is established. This initial state serves as the basis for subsequent analysis of soil environmental changes. It can eliminate the ambiguity of the prediction process without a baseline reference, ensure that the soil state prediction based on air environment and crop growth information has a clear starting point, and thus improve the scientificity and accuracy of the prediction.

[0028] Specifically, the first timestamp of the air environment monitoring data can be used as the benchmark. For example, the time when the smart observer starts collecting air data is assumed to be 2025-07-20 08:00, then this time point can be defined as the "initial moment".

[0029] Then, extract historical monitoring data related to the target greenhouse area from the cloud platform, including past soil environment monitoring time series information, air environment monitoring time series information, etc. This historical data must include timestamps and cover the monitoring period before the initial time.

[0030] Next, from the historical soil environment monitoring time series information, the soil environment monitoring time series information that is completely consistent with or closest to the initial time of the air environment, such as 2025-07-2008:00, is selected as the "initial time soil environment information". For example: If air environment monitoring starts at 2025-07-2008:00, then the soil temperature, humidity, conductivity, pH value and other parameters at that time point are extracted from the historical environment monitoring time series information, such as the soil environment monitoring time series information at 2025-07-2008:00 in the previous dataset example.

[0031] Step S200 aligns the air environment monitoring start point with the initial soil environment state through timestamp alignment, providing specific input values ​​for predicting the soil environment using the initial soil state constraint in step S300. For example, if subsequent air environment monitoring shows a sudden increase in light intensity, combining the soil moisture at the initial moment can more accurately predict the humidity trend caused by soil evaporation. In other words, the clarity of the initial state directly affects the credibility of such predictions.

[0032] In step S300 of the embodiment of the present application, the soil environment information at the initial moment is used as the initial soil state constraint, the air environment monitoring time series information and the crop growth information are used as soil state impact constraints, a soil state monitoring sample set is collected, and soil environment prediction time series information is statistically analyzed, including: Performing spatial clustering on the soil environment information at the initial moment to obtain soil environment information of a first region at the initial moment to soil environment information of an Nth region at the initial moment; Based on the first area to the Nth area, the air environment monitoring time series information is partitioned to obtain the air environment monitoring time series information of the first area to the air environment monitoring time series information of the Nth area; Traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information to perform spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering results until the N-th regional air environment spatiotemporal clustering results; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and the crop growth information and the spatiotemporal clustering result of the air environment in the first region as the soil state impact constraint, collecting a first soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information of the first region; The Nth soil state monitoring sample set is collected, and the soil environment prediction time series information of the Nth region is statistically analyzed, using the soil environment information of the Nth region at the initial moment as the initial soil state constraint, and using the crop growth information and the spatiotemporal clustering result of the air environment of the Nth region as the soil state impact constraint; The first region soil environment prediction time series information to the Nth region soil environment prediction time series information are added to the soil environment prediction time series information.

[0033] In step S300 of the embodiment of the present application, spatial clustering is performed on the soil environment information at the initial moment. The purpose is to refine the greenhouse target area into several sub-areas through spatial division and data matching, so that the soil environment prediction is upgraded from overall fuzzy estimation to precise analysis of partitions.

[0034] Specifically, spatial clustering is first performed on the initial soil environmental information to obtain soil environmental information for the first region up to the Nth region. Spatial clustering can be performed based on spatial distribution characteristics of the initial soil environmental information, such as spatial differences in soil moisture and electrical conductivity, using a clustering algorithm such as K-means to divide the target greenhouse area into N highly homogeneous sub-regions.

[0035] Specifically, soil environmental parameters such as soil temperature and air humidity are collected from multiple monitoring points in the greenhouse. Each point contains spatial coordinates, such as x=1m, y=2m, and specific parameter values, such as soil temperature 22°C.

[0036] Assume there are five monitoring points in the greenhouse, and their coordinates and soil temperatures are: A(1,1), 22°C; B(1,3), 23°C; C(3,1), 18°C; D(3,3), 19°C; E(2,2), 21°C. The above features are combined into feature vectors to calculate the Euclidean distance. For example, A(1,1), 22°C is combined into [1,1,22] The Euclidean distance between two points is then used to calculate the overall difference between them. The formula for this is: Dij = √[(Xi - Xj)² + (Yi - Yj)² + (Pi - Pj)²], where X and Y are spatial coordinates, and P is an environmental parameter, such as temperature. i and j represent the two sets of feature vectors to be calculated.

[0037] Using the above formula, we can calculate the Euclidean distance between A and B to be 2.24, the Euclidean distance between A and C to be 4.47, and so on. Then, using the K-means clustering algorithm, we divide the greenhouse target area into N highly homogeneous sub-regions based on the Euclidean distances between the feature vectors of the sub-regions within the target greenhouse area. The K-means clustering algorithm is a state-of-the-art technique and will not be further described here.

[0038] For example, if soil moisture in a greenhouse exhibits a spatial distribution with higher moisture levels in the east and lower moisture levels in the west, clustering can generate "Region 1 (East)" and "Region 2 (West)." Each region corresponds to a set of initial soil state data, ranging from "Soil environment information at the initial moment of Region 1" to "Soil environment information at the initial moment of Region N." For example, the initial soil moisture in Region 1, i.e., the initial soil temperature in the east of the greenhouse, could be 22°C, and so on. Spatial distribution data can be obtained by installing the aforementioned intelligent observation instruments at different locations in the greenhouse, providing raw samples for spatial clustering.

[0039] Furthermore, it is necessary to partition the air environment monitoring time series information based on the first area to the Nth area to obtain the air environment monitoring time series information of the first area to the Nth area.

[0040] Specifically, the N regions divided in the first step need to be used as spatial boundaries. The overall air environment monitoring time series information needs to be split by region, so that each region only corresponds to the air environment data within its range. For example, the air environment monitoring time series information for the first region (east side) only includes temperature, humidity, light, and other data monitored by sensors in that area. This forms air environment data that corresponds one-to-one with the soil partitions, namely "air environment monitoring time series information for the first region" through "air environment monitoring time series information for the Nth region."

[0041] In step S300 of the embodiment of the present application, traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information is subjected to spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering result until the N-th regional air environment spatiotemporal clustering result includes: Performing spatial clustering on the air environment monitoring time series information of the first region to obtain the air environment monitoring time series information of the first sub-region up to the air environment monitoring time series information of the Mth sub-region; Performing time-domain clustering on the air environment monitoring time series information of the first sub-region to obtain air environment information of the first time zone of the first sub-region up to air environment information of the Yth time zone of the first sub-region; Until the time domain clustering is performed on the air environment monitoring time series information of the M-th sub-region to obtain the air environment information of the first time zone of the M-th sub-region to the air environment information of the Z-th time zone of the M-th sub-region; The air environment information of the first sub-region in the first time zone to the air environment information of the Yth time zone of the first sub-region, to the air environment information of the first time zone of the Mth sub-region to the air environment information of the Zth time zone of the Mth sub-region, are added to the spatiotemporal clustering result of the air environment of the first region.

[0042] In the embodiments of this application, the above steps are the core of "spatiotemporal clustering." The core purpose is to decompose the air environment monitoring time series information of each region into smaller-scale spatiotemporal units through spatial and temporal segmentation, thereby achieving a "fine-grained depiction" of air environment changes. Specifically, the air environment data is first further divided into sub-regions in space and then divided into time zones in time. This converts the air environment data from a "region-level time series stream" into a "sub-region-time zone" gridded data block. This provides more detailed impact constraints for subsequent accurate prediction of soil environment changes and avoids prediction bias caused by coarse spatiotemporal scales.

[0043] First, it is necessary to perform spatial clustering on the air environment monitoring time series information of the first region to obtain the air environment monitoring time series information of the first sub-region up to the air environment monitoring time series information of the Mth sub-region. Specifically, within the first region, secondary spatial clustering is performed based on the spatial distribution of air environment parameters such as temperature, humidity, and light intensity. For example, a density clustering algorithm is used to subdivide the first region into M subregions of higher homogeneity. For example, if the first region has low humidity near vents and high humidity in corners, the region can be divided into a "first subregion" (i.e., the vicinity of the vents) and a "second subregion" (i.e., the corners), and so on.

[0044] By using the above method, the air environment monitoring time series information of the first sub-region to the Mth sub-region is obtained. Each sub-region corresponds to the change data of the air parameters within its range over time, such as the temperature and humidity records of the first sub-region every hour. The air environment information of the first sub-region in the first time zone is obtained until the air environment information of the first sub-region in the Yth time zone is obtained. Next, it is necessary to perform time-domain clustering on the air environment monitoring time series information of the first sub-region to obtain the air environment information of the first time zone of the first sub-region until the air environment information of the Yth time zone of the first sub-region. For the air environment monitoring time series information of each sub-region, time-domain clustering is performed according to the parameter change characteristics of the time dimension, such as the temperature fluctuation pattern between day and night, the day and night difference in light intensity and other parameter change characteristics. Sliding window clustering can be used to divide the continuous time into Z time zones. The specific time-domain clustering method is the same as the aforementioned spatial domain clustering method. It can also be implemented by quantizing the air environment monitoring time series information features of multiple sub-regions, and then calculating the Euclidean distance. The K-means time-domain clustering method based on the Euclidean distance is used. It will not be repeated here.

[0045] For example, the first sub-area can be divided into two time zones according to "daytime (8:00-18:00)" and "nighttime (18:00-8:00 the next day)", or divided into multiple time zones according to the parameter stability every 6 hours.

[0046] Through the above steps, the air environment monitoring time series information of each sub-region is split into air environment information of multiple time zones, such as the air environment information data corresponding to the first time zone (8:00-14:00) and the second time zone (14:00-20:00) of the first sub-region. Finally, the air environment information of the first time zone of the Mth sub-region is obtained until the air environment information of the Zth time zone of the Mth sub-region is obtained. Finally, it is necessary to add the air environment information of the first time zone of the first sub-region to the air environment information of the Yth time zone of the first sub-region, to the air environment information of the first time zone of the Mth sub-region to the air environment information of the Zth time zone of the Mth sub-region into the spatiotemporal clustering result of the air environment of the first region.

[0047] That is, the air environment information for all time zones from the first sub-region to the Mth sub-region is integrated to form the spatiotemporal clustering result of the air environment for the first region. For example, the spatiotemporal clustering result for the first region includes air environment information data blocks for "first sub-region - first time zone," "first sub-region - second time zone," ..., "Mth sub-region - Zth time zone."

[0048] Similarly, the same processing is completed for the second area to the Nth area, and finally the spatiotemporal clustering results of the air environment of all areas are obtained.

[0049] In step S300 of the embodiment of the present application, the soil environment information of the first region at the initial moment is used as the initial soil state constraint, the crop growth information and the spatiotemporal clustering result of the air environment of the first region are used as the soil state impact constraint, a first soil state monitoring sample set is collected, and the soil environment prediction time series information of the first region is statistically analyzed, including: Extracting the air environment information of the first sub-region in the first time zone up to the air environment information of the Yth time zone of the first sub-region from the spatiotemporal clustering result of the air environment of the first region, until extracting the air environment information of the first time zone of the Mth sub-region up to the air environment information of the Zth time zone of the Mth sub-region; Extracting crop planting time and crop varieties from the crop growth information; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and using the crop planting duration, the crop variety, and the air environment information of the first subregion in the first time zone as constraints, collecting a first sub-soil state monitoring sample set, and calculating soil environment prediction information of the first subregion at the end moment of the first time zone; The soil environment prediction information of the first sub-region at the end time of the Y-1 time zone is used as the initial soil state constraint, and the crop planting duration, the crop variety, and the air environment information of the first sub-region at the Y time zone are used as constraints. The Y-th sub-soil state monitoring sample set is collected, and the soil environment prediction information of the first sub-region at the end time of the Y time zone is calculated; Add the soil environment prediction information of the first sub-region at the end of the first time zone to the soil environment prediction information of the first sub-region at the end of the Yth time zone to the soil environment prediction time series information of the first sub-region; Until the soil environment prediction information of the end time of the first time zone of the Mth sub-region is obtained, until the soil environment prediction information of the end time of the Zth time zone of the Mth sub-region is obtained, and added to the soil environment prediction time series information of the Mth sub-region; The first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information are added to the first region soil environment prediction time series information.

[0050] In this embodiment, the purpose of statistically analyzing soil environmental prediction time series information is to predict soil environmental changes in each subregion in stages based on refined spatiotemporal constraints, ultimately integrating regional-level soil environmental prediction time series information. By combining the initial soil state, crop variety, and crop planting duration with the subdivided spatiotemporal characteristics of the air environment, dynamic predictions from the "initial state" to the "end state of each time zone" are achieved. This ensures that the prediction results accurately reflect the differences in soil environments across different subregions and time periods, providing a high-precision reference benchmark for comparison with actual monitoring data in the subsequent step S400.

[0051] First, based on the spatiotemporal clustering results of the air environment in the first region, extract the air environment information for the first time zone of the first subregion, through the Yth time zone of the first subregion, and then extract the air environment information for the first time zone of the Mth subregion, through the Zth time zone of the Mth subregion. This air environment information includes temperature, humidity, light intensity, CO2 concentration, and other information. For example, the air environment data for the first time zone of the first subregion (8:00-14:00) is "temperature 28-32°C, humidity 60%-65%, and light intensity 60,000-80,000 Lux."

[0052] Then, the crop planting time and crop variety need to be extracted from the crop growth information. For example, if the crop variety is tomato, the planting time is 45 days. These parameters directly affect the nutrient consumption and water demand of the soil.

[0053] Then, the soil environment information at the initial moment of the first region is used as the initial soil state constraint. For example, the initial soil environment information at that moment is: soil temperature 22.5°C, humidity 65.2%. Based on the crop planting duration, crop variety, and air environment information in the first time zone of the first sub-region, a first sub-soil state monitoring sample set is collected, and soil environment prediction information at the end time of the first sub-region and the first time zone is calculated. The first sub-soil state monitoring sample set is soil change data under similar historical conditions over the same period, and is used to predict the soil environment at the end time of the first sub-region and the first time zone, such as the predicted soil moisture value at 2:00 PM.

[0054] Next, a rolling time series forecast is required. For subsequent time zones in the same sub-region, the forecast result of the end point of the previous time zone is used as the new initial constraint, and the forecast is continued in combination with the air environment information and crop characteristics of the corresponding time zone.

[0055] For example, the initial constraint for the second time zone (14:00-20:00) of the first sub-region is the soil prediction value at the end point of the first time zone. Combined with the air environment information data of this time period, such as temperature 30-26°C, humidity 50-65%, and light intensity 10,000-50,000 Lux, the soil environment at 20:00 is predicted. This is the process of using the soil environment prediction information at the end point of the Y-1th time zone of the first sub-region as the initial soil state constraint, the crop planting duration, the crop variety, and the air environment information of the Yth time zone of the first sub-region as constraints, collecting the Yth sub-soil state monitoring sample set, and statistically analyzing the soil environment prediction information at the end point of the Yth time zone of the first sub-region.

[0056] Repeat the above steps to complete the soil environment prediction for all sub-regions from the first to the Mth time zone in the first region, until the soil environment prediction information at the end moment of the first time zone of the Mth sub-region is obtained until the soil environment prediction information at the end moment of the Zth time zone of the Mth sub-region is obtained, and added to the soil environment prediction time series information of the Mth sub-region.

[0057] Finally, the soil environment prediction information for each subregion at the end of each time zone is arranged in chronological order to form a subregion-level soil environment prediction time series, such as the soil moisture prediction time series for the first subregion every 6 hours from 8:00 AM to 8:00 AM the next day. The soil environment prediction time series for all subregions are integrated to form the soil environment prediction time series for the first region, fully reflecting the dynamic changes in the soil environment over time within that region.

[0058] In step S400 of the embodiment of the present application, when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious, including: Extracting first-region soil environment prediction time series information from the soil environment prediction time series information, wherein the first-region soil environment prediction time series information includes the first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information; Extracting first-region soil environment monitoring time series information from the soil environment monitoring time series information, wherein the first-region soil environment monitoring time series information includes the first sub-region soil environment monitoring time series information to the M-th sub-region soil environment monitoring time series information; performing similarity analysis on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain a first sub-region soil environment similarity; until similarity analysis is performed on the soil environment prediction time series information of the M-th sub-region and the soil environment monitoring time series information of the M-th sub-region to obtain the soil environment similarity of the M-th sub-region; Construct consistency comparison rules: when the soil environment similarity is greater than the similarity threshold, the soil environment is considered consistent, otherwise it is considered inconsistent; Based on the consistency comparison rule, processing the soil environment similarity of the first sub-region until the soil environment similarity of the M-th sub-region to obtain inconsistent sub-regions; Based on the inconsistent sub-region, the soil environment monitoring time series information and the air environment monitoring time series information are suspiciously marked to obtain first region suspicious marking data, which is added to the full region suspicious marking data.

[0059] In the embodiments of this application, extracting the predicted and monitored soil environment time series information for the first region is the fundamental data processing step for identifying suspicious areas. The core purpose is to split the predicted and monitored soil environment data into a "region-subregion" hierarchy and accurately match them, providing a structured comparison unit for subsequent determination of their consistency. By focusing on the time series data of each subregion within the first region, anomaly identification can be localized to the smallest spatial unit, preventing differences in local subregions from being masked by overall regional comparisons, thereby improving the accuracy of anomaly detection.

[0060] Specifically, it is first necessary to extract the soil environment monitoring time series information of the first region from the soil environment monitoring time series information. That is, from the global soil environment prediction time series information generated in step S300, filter out the prediction time series belonging to the first region, such as the prediction data sequence of the east side of a greenhouse. Then further split it by sub-region, extract the prediction time series information of all sub-regions in the first region, such as the first sub-region is "near the vent" and the second sub-region is "corner", and obtain the measurement time series information of these sub-regions. These sub-region prediction time series information contains the predicted values ​​of soil parameters at the end points of each time zone, such as humidity, temperature, and conductivity, and corresponds one-to-one to the sub-regions divided in step S300.

[0061] For example, extract data such as "the predicted soil moisture value of the end point of the first sub-region 8:00-14:00 time zone is 63%" and "the predicted soil moisture value of the end point of the first sub-region 14:00-20:00 time zone is 61%" from the first region soil environment prediction time series. Use the same method to extract the soil environment prediction time series information of the first sub-region until the soil environment prediction time series information of the Mth sub-region. Next, it is necessary to extract the soil environment monitoring time series information of the first region from the soil environment monitoring time series information, that is, to filter out the soil environment monitoring time series information of the first region from the soil environment monitoring time series information actually monitored by the aforementioned intelligent observation instrument.

[0062] The soil environment monitoring time series information is split by sub-region, and the measured soil environment monitoring time series information of each sub-region is extracted. The data comes from the soil sensors in the smart observation instruments deployed in the sub-region, such as the measured values ​​of humidity and temperature uploaded every 6 hours by the sensors in the first sub-region.

[0063] For example, extract data such as "the measured soil moisture value at the end of the first sub-region between 8:00 and 14:00 is 58%" and "the measured soil moisture value at the end of the first sub-region between 14:00 and 20:00 is 56%" from the soil environment monitoring time series information of the first region. Use the same method to extract the soil environment monitoring time series information of the first sub-region until the soil environment monitoring time series information of the Mth sub-region. In step S400 of the embodiment of the present application, similarity analysis is performed on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain the first sub-region soil environment similarity, including: Extracting a soil environment prediction value at a first moment from the first sub-region soil environment prediction time series information; Extracting a soil environment monitoring value at a first moment from the soil environment monitoring time series information of the first sub-region; Calculating the same attribute standard deviation of the soil environment prediction value and the soil environment monitoring value at the first moment to obtain several attribute standard deviations; Traversing the plurality of attribute standard deviations and performing dimensionless processing to obtain a plurality of deviation eigenvalues; The plurality of deviation characteristic values ​​are respectively used as a plurality of dimensional distance parameters, and the Euclidean distance is calculated, and then the reciprocal is obtained to obtain the soil environment similarity of the first sub-region.

[0064] In the embodiment of the present application, the purpose of the above steps is to quantitatively evaluate the degree of matching between the predicted value and the measured value of the soil environment in the same sub-area, and to provide an objective basis for judging whether the two are consistent by calculating the soil environment similarity.

[0065] Specifically, it is first necessary to extract the soil environment prediction value at the first moment and the soil environment monitoring value at the first moment from the soil environment prediction time series information of the first sub-region.

[0066] For the soil environment prediction value at the first moment, a specific moment can be selected from the soil environment prediction time series information of the first sub-region, such as the soil environment prediction value at 14:00 in the previous example, which covers all monitoring parameters of the sub-region at that moment, such as soil temperature of 24.1°C, humidity of 63.8%, conductivity of 118.7 μS / cm, and pH value of 6.7.

[0067] For the soil environment monitoring values ​​at the first moment, the measured values ​​at the same moment can be extracted from the real-time monitored soil environment time series information of the first sub-area, such as soil temperature 23.5°C, humidity 59.2%, conductivity 125.3 μS / cm, and pH value 6.8, to ensure that the parameter type fully corresponds to the soil environment prediction value.

[0068] Then, it is necessary to calculate the same attribute standard deviation of the soil environment prediction value at the first moment and the soil environment monitoring value at the first moment to obtain several attribute standard deviations.

[0069] The standard deviation between the predicted and measured values ​​for each attribute, such as temperature and humidity, is calculated to reflect the degree of dispersion of the single-dimensional parameter. For example, the standard deviation between the predicted temperature value of 24.1°C and the measured value of 23.5°C is 0.6°C; the standard deviation between the predicted humidity value of 63.8% and the measured value of 59.2% is 4.6%. This is repeated for all attributes, resulting in the standard deviation of several attributes.

[0070] Then, the standard deviations of the several attributes are traversed and dimensionless processing is performed to obtain several deviation eigenvalues. Since the units and magnitudes of different attributes are different, such as temperature in ℃ and humidity in %, directly comparing the standard deviation will result in deviations. It is necessary to convert it into a dimensionless deviation eigenvalue through standardization. For example, if the standard deviation is divided by the reasonable value range of the attribute, the temperature deviation eigenvalue = 0.6 / (120-(-40)) = 0.00375, and the humidity deviation eigenvalue = 4.6 / (100-0) = 0.046), so that the differences in each dimension can be directly compared. Among them, the reasonable value range can be determined by the maximum range of the sensor, or it can be further determined by the numerical range that appears in the actual measurement process. The values ​​here are only examples and are not representative.

[0071] Then, the Euclidean distance is calculated using the several deviation eigenvalues ​​as the distance parameters for several dimensions, and the inverse is taken to obtain the soil environment similarity of the first sub-region. That is, the dimensionless deviation eigenvalues ​​are used as coordinates in a multidimensional space. For example, temperature, humidity, conductivity, and pH value correspond to four dimensions. The Euclidean distance between the predicted value and the measured value in this space is calculated to comprehensively reflect the overall difference in multiple dimensions. The larger the Euclidean distance, the more significant the overall difference. For example, if the deviation eigenvalues ​​of the four attributes in the above example are 0.00375, 0.046, 0.032, and 0.01, respectively, then the Euclidean distance is √(0.00375²+0.046²+0.032²+0.01²)≈0.057.

[0072] Then, the reciprocal of the Euclidean distance is taken, such as 1 / 0.057≈17.5, to obtain the soil environment similarity of the first subregion. The smaller the Euclidean distance, the larger the reciprocal, and the higher the soil environment similarity of the first subregion, which directly reflects the degree of match between the predicted value and the measured value.

[0073] Furthermore, it is necessary to perform a similarity analysis on the soil environment prediction time series information of the Mth sub-region and the soil environment monitoring time series information of the Mth sub-region according to the above method to obtain the soil environment similarity of the Mth sub-region.

[0074] Then, a consistency comparison rule is constructed: when the soil environment similarity is greater than the similarity threshold, the soil environment is considered consistent, otherwise it is considered inconsistent.

[0075] Specifically, a unified soil environment similarity threshold, such as 0.8, can be set. This threshold can be adjusted based on greenhouse crop type and sensor accuracy requirements. The consistency comparison rule is defined as follows: when the soil environment similarity of a subregion exceeds the threshold, the prediction and monitoring data for that subregion are considered "consistent," meaning the data is reliable. When the similarity is ≤ the threshold, the prediction is considered "inconsistent," indicating that the data may be abnormal.

[0076] The setting of the similarity threshold also needs to be combined with the characteristics of the monitoring equipment to ensure that the rule can not only identify real anomalies, such as deviations caused by sensor hardware failures, but also tolerate reasonable measurement errors, such as minor deviations caused by environmental fluctuations.

[0077] Furthermore, it is necessary to process the soil environment similarity of the first sub-region up to the soil environment similarity of the M-th sub-region based on the consistency comparison rule to obtain inconsistent sub-regions.

[0078] Specifically, the soil environment similarity of the first to Mth sub-regions can be compared with the threshold, for example: the soil environment similarity of the first sub-region is 0.85>0.8 (consistent), the soil environment similarity of the second sub-region is 0.72<0.8 (inconsistent), the soil environment similarity of the third sub-region is 0.83>0.8 (consistent)...the soil environment similarity of the Mth sub-region is 0.79<0.8 (inconsistent).

[0079] Summarize all "inconsistent" sub-areas, such as the second sub-area and the Mth sub-area, to form a list of abnormal sub-areas in the first area, and clearly mark their spatial locations, such as "corner sub-area" and "sub-area near the west fence."

[0080] In step S400 of the embodiment of the present application, the following is also included: Count the proportion of inconsistent sub-regions to the total number of sub-regions in the first region, and set it as the inconsistent proportion of the first region; When the inconsistency ratio of the first region is greater than the inconsistency ratio threshold, based on the inconsistent sub-regions, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious; When the inconsistency ratio of the first area is less than or equal to the inconsistency ratio threshold, the air environment monitoring time series information is marked as suspicious.

[0081] In the embodiment of the present application, the purpose of the above steps is to differentiate and judge the possible sources of data anomalies based on the distribution ratio of abnormal areas, so as to mark suspicious data in a targeted manner. By counting the proportion of inconsistent sub-areas in the area, that is, the aforementioned inconsistency ratio, and comparing it with the preset inconsistency ratio threshold, local anomalies and systemic anomalies can be distinguished. When the inconsistency ratio is too high, there may be problems with both the soil and air environment monitoring data; when the inconsistency ratio is too low, it is more likely to be caused by errors in the air environment monitoring data used as a prediction constraint, thereby improving the accuracy of suspicious identification and reducing invalid investigations.

[0082] First, we need to calculate the ratio of inconsistent subregions to the total number of subregions in the first region, which is the first region's inconsistent ratio. This means calculating the number of inconsistent subregions in the first region. For example, if there are two in the second and fifth subregions, then we calculate the ratio of the number of inconsistent subregions to the total number of subregions in that region. If the total number of subregions in that region is M = 5, then the inconsistent ratio = 2 / 5 = 40%.

[0083] Next, it is necessary to preset a threshold for the proportion of inconsistency, which can be set according to the stability requirements of the greenhouse environment. For example, it can be set to 30% as the boundary between local anomalies and systemic anomalies.

[0084] Then compare the non-consistent ratio of the first area with the non-consistent ratio threshold: if the non-consistent ratio of the first area is 40%, 40%>30%, it is judged as "high risk of systemic abnormality"; if the non-consistent ratio of the first area is 20%, 20%<30%, it is judged as "high risk of local abnormality".

[0085] When the inconsistency ratio is greater than the threshold, it can be considered that both the soil and air environment monitoring data may be abnormal. Based on the inconsistent sub-areas, the soil environment monitoring time series information and the air environment monitoring time series information of these sub-areas are marked as suspicious, such as marking "Data abnormality - sensor needs to be checked" on the cloud platform.

[0086] When the inconsistency ratio is less than or equal to the threshold, it can be considered that the anomaly is more likely to come from the air environment monitoring data, and only the air environment monitoring time series information of the first area is marked as suspicious, such as marking "air data may be distorted - affecting soil prediction", to reduce unnecessary checks on soil monitoring data.

[0087] By using the above method, the soil environment monitoring time series information and the air environment monitoring time series information are suspiciously marked, and the first area suspicious marking data is obtained and added to the whole area suspicious marking data.

[0088] Finally, the suspicious identification results are pushed to the user end through the cloud platform, clearly indicating the abnormal areas and the types of parameters that need to be checked, completing the greenhouse environment data monitoring with multi-sensor fusion.

[0089] Example 2, as Figure 2 As shown, based on the same inventive concept as the greenhouse environment data monitoring method of multi-sensor fusion provided in Example 1, an embodiment of the present invention further provides a greenhouse environment data monitoring platform of multi-sensor fusion, including: A monitoring information acquisition module 11 is configured to receive soil environment monitoring time series information, air environment monitoring time series information, and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth during the environmental monitoring time window; An environmental information extraction module 12 is configured to extract soil environmental information at the initial moment of the air environment monitoring time series information from environmental monitoring historical information; The prediction information statistics module 13 is configured to collect a soil state monitoring sample set and calculate soil environment prediction time series information, using the soil environment information at the initial moment as an initial soil state constraint, the air environment monitoring time series information and the crop growth information as soil state impact constraints; The monitoring information identification module 14 is configured to, when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, mark the soil environment monitoring time series information and the air environment monitoring time series information as suspicious and send the information to the user end.

[0090] Furthermore, the prediction information statistics module 13 includes the following execution steps: Performing spatial clustering on the soil environment information at the initial moment to obtain soil environment information of a first region at the initial moment to soil environment information of an Nth region at the initial moment; Based on the first area to the Nth area, the air environment monitoring time series information is partitioned to obtain the air environment monitoring time series information of the first area to the air environment monitoring time series information of the Nth area; Traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information to perform spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering results until the N-th regional air environment spatiotemporal clustering results; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and the crop growth information and the spatiotemporal clustering result of the air environment in the first region as the soil state impact constraint, collecting a first soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information of the first region; The Nth soil state monitoring sample set is collected, and the soil environment prediction time series information of the Nth region is statistically analyzed, using the soil environment information of the Nth region at the initial moment as the initial soil state constraint, and using the crop growth information and the spatiotemporal clustering result of the air environment of the Nth region as the soil state impact constraint; The first region soil environment prediction time series information to the Nth region soil environment prediction time series information are added to the soil environment prediction time series information.

[0091] The step of traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information is subjected to spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering result until the N-th regional air environment spatiotemporal clustering result includes: Performing spatial clustering on the air environment monitoring time series information of the first region to obtain the air environment monitoring time series information of the first sub-region up to the air environment monitoring time series information of the Mth sub-region; Performing time-domain clustering on the air environment monitoring time series information of the first sub-region to obtain air environment information of the first time zone of the first sub-region up to air environment information of the Yth time zone of the first sub-region; Until the time domain clustering is performed on the air environment monitoring time series information of the M-th sub-region to obtain the air environment information of the first time zone of the M-th sub-region to the air environment information of the Z-th time zone of the M-th sub-region; The air environment information of the first sub-region in the first time zone to the air environment information of the Yth time zone of the first sub-region, to the air environment information of the first time zone of the Mth sub-region to the air environment information of the Zth time zone of the Mth sub-region, are added to the spatiotemporal clustering result of the air environment of the first region.

[0092] The soil environment information of the first region at the initial moment is used as the initial soil state constraint, the crop growth information and the spatiotemporal clustering result of the air environment of the first region are used as the soil state impact constraint, a first soil state monitoring sample set is collected, and the soil environment prediction time series information of the first region is statistically analyzed, including: Extracting the air environment information of the first sub-region in the first time zone up to the air environment information of the Yth time zone of the first sub-region from the spatiotemporal clustering result of the air environment of the first region, until extracting the air environment information of the first time zone of the Mth sub-region up to the air environment information of the Zth time zone of the Mth sub-region; Extracting crop planting time and crop varieties from the crop growth information; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and using the crop planting duration, the crop variety, and the air environment information of the first subregion in the first time zone as constraints, collecting a first sub-soil state monitoring sample set, and calculating soil environment prediction information of the first subregion at the end moment of the first time zone; The soil environment prediction information of the first sub-region at the end time of the Y-1 time zone is used as the initial soil state constraint, and the crop planting duration, the crop variety, and the air environment information of the first sub-region at the Y time zone are used as constraints. The Y-th sub-soil state monitoring sample set is collected, and the soil environment prediction information of the first sub-region at the end time of the Y time zone is calculated; Add the soil environment prediction information of the first sub-region at the end of the first time zone to the soil environment prediction information of the first sub-region at the end of the Yth time zone to the soil environment prediction time series information of the first sub-region; Until the soil environment prediction information of the end time of the first time zone of the Mth sub-region is obtained, until the soil environment prediction information of the end time of the Zth time zone of the Mth sub-region is obtained, and added to the soil environment prediction time series information of the Mth sub-region; The first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information are added to the first region soil environment prediction time series information.

[0093] Furthermore, the monitoring information identification module 14 includes the following execution steps: Extracting first-region soil environment prediction time series information from the soil environment prediction time series information, wherein the first-region soil environment prediction time series information includes the first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information; Extracting first-region soil environment monitoring time series information from the soil environment monitoring time series information, wherein the first-region soil environment monitoring time series information includes the first sub-region soil environment monitoring time series information to the M-th sub-region soil environment monitoring time series information; performing similarity analysis on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain a first sub-region soil environment similarity; until similarity analysis is performed on the soil environment prediction time series information of the M-th sub-region and the soil environment monitoring time series information of the M-th sub-region to obtain the soil environment similarity of the M-th sub-region; Construct consistency comparison rules: when the soil environment similarity is greater than the similarity threshold, the soil environment is considered consistent, otherwise it is considered inconsistent; Based on the consistency comparison rule, processing the soil environment similarity of the first sub-region until the soil environment similarity of the M-th sub-region to obtain inconsistent sub-regions; Based on the inconsistent sub-region, the soil environment monitoring time series information and the air environment monitoring time series information are suspiciously marked to obtain first region suspicious marking data, which is added to the full region suspicious marking data.

[0094] Among them, also include: Count the proportion of inconsistent sub-regions to the total number of sub-regions in the first region, and set it as the inconsistent proportion of the first region; When the inconsistency ratio of the first region is greater than the inconsistency ratio threshold, based on the inconsistent sub-regions, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious; When the inconsistency ratio of the first area is less than or equal to the inconsistency ratio threshold, the air environment monitoring time series information is marked as suspicious.

[0095] The similarity analysis of the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information is performed to obtain the first sub-region soil environment similarity, including: Extracting a soil environment prediction value at a first moment from the first sub-region soil environment prediction time series information; Extracting a soil environment monitoring value at a first moment from the soil environment monitoring time series information of the first sub-region; Calculating the same attribute standard deviation of the soil environment prediction value and the soil environment monitoring value at the first moment to obtain several attribute standard deviations; Traversing the plurality of attribute standard deviations and performing dimensionless processing to obtain a plurality of deviation eigenvalues; The plurality of deviation characteristic values ​​are respectively used as a plurality of dimensional distance parameters, and the Euclidean distance is calculated, and then the reciprocal is obtained to obtain the soil environment similarity of the first sub-region.

[0096] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0097] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0098] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0101] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0102] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A greenhouse environment data monitoring method based on multi-sensor fusion, characterized in that: include: receiving soil environment monitoring time series information, air environment monitoring time series information, and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth in the environmental monitoring time window; Extracting soil environmental information at the initial moment of the air environmental monitoring time series information from environmental monitoring historical information; Using the soil environment information at the initial moment as the initial soil state constraint, and the air environment monitoring time series information and the crop growth information as soil state impact constraints, collecting a soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information; When the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious and sent to the user end.

2. The method according to claim 1, wherein Taking the soil environment information at the initial moment as the initial soil state constraint, taking the air environment monitoring time series information and the crop growth information as the soil state impact constraint, collecting a soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information, including: Performing spatial clustering on the soil environment information at the initial moment to obtain soil environment information of a first region at the initial moment to soil environment information of an Nth region at the initial moment; Based on the first area to the Nth area, the air environment monitoring time series information is partitioned to obtain the air environment monitoring time series information of the first area to the air environment monitoring time series information of the Nth area; Traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information to perform spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering results until the N-th regional air environment spatiotemporal clustering results; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and the crop growth information and the spatiotemporal clustering result of the air environment in the first region as the soil state impact constraint, collecting a first soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information of the first region; The Nth soil state monitoring sample set is collected, and the soil environment prediction time series information of the Nth region is statistically analyzed, using the soil environment information of the Nth region at the initial moment as the initial soil state constraint, and using the crop growth information and the spatiotemporal clustering result of the air environment of the Nth region as the soil state impact constraint; The first region soil environment prediction time series information to the Nth region soil environment prediction time series information are added to the soil environment prediction time series information.

3. The method according to claim 2, wherein Traversing the first regional air environment monitoring time series information until the N-th regional air environment monitoring time series information performs spatiotemporal clustering, and obtaining the first regional air environment spatiotemporal clustering result until the N-th regional air environment spatiotemporal clustering result, including: Performing spatial clustering on the air environment monitoring time series information of the first region to obtain the air environment monitoring time series information of the first sub-region up to the air environment monitoring time series information of the Mth sub-region; Performing time-domain clustering on the air environment monitoring time series information of the first sub-region to obtain air environment information of the first time zone of the first sub-region up to air environment information of the Yth time zone of the first sub-region; Until the time domain clustering is performed on the air environment monitoring time series information of the M-th sub-region to obtain the air environment information of the first time zone of the M-th sub-region to the air environment information of the Z-th time zone of the M-th sub-region; The air environment information of the first sub-region in the first time zone to the air environment information of the Yth time zone of the first sub-region, to the air environment information of the first time zone of the Mth sub-region to the air environment information of the Zth time zone of the Mth sub-region, are added to the spatiotemporal clustering result of the air environment of the first region.

4. The method according to claim 3, wherein Taking the soil environment information of the first region at the initial moment as the initial soil state constraint, taking the crop growth information and the spatiotemporal clustering result of the air environment of the first region as the soil state impact constraint, collecting a first soil state monitoring sample set, and statistically analyzing the soil environment prediction time series information of the first region, including: Extracting the air environment information of the first sub-region in the first time zone up to the air environment information of the Yth time zone of the first sub-region from the spatiotemporal clustering result of the air environment of the first region, until extracting the air environment information of the first time zone of the Mth sub-region up to the air environment information of the Zth time zone of the Mth sub-region; Extracting crop planting time and crop varieties from the crop growth information; Using the soil environment information of the first region at the initial moment as the initial soil state constraint, and using the crop planting duration, the crop variety, and the air environment information of the first subregion in the first time zone as constraints, collecting a first sub-soil state monitoring sample set, and calculating soil environment prediction information of the first subregion at the end moment of the first time zone; The soil environment prediction information of the first sub-region at the end time of the Y-1 time zone is used as the initial soil state constraint, and the crop planting duration, the crop variety, and the air environment information of the first sub-region at the Y time zone are used as constraints. The Y-th sub-soil state monitoring sample set is collected, and the soil environment prediction information of the first sub-region at the end time of the Y time zone is calculated; Add the soil environment prediction information of the first sub-region at the end of the first time zone to the soil environment prediction information of the first sub-region at the end of the Yth time zone to the soil environment prediction time series information of the first sub-region; Until the soil environment prediction information of the end time of the first time zone of the Mth sub-region is obtained, until the soil environment prediction information of the end time of the Zth time zone of the Mth sub-region is obtained, and added to the soil environment prediction time series information of the Mth sub-region; The first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information are added to the first region soil environment prediction time series information.

5. The method according to claim 1, wherein When the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious, including: Extracting first-region soil environment prediction time series information from the soil environment prediction time series information, wherein the first-region soil environment prediction time series information includes the first sub-region soil environment prediction time series information to the M-th sub-region soil environment prediction time series information; Extracting first-region soil environment monitoring time series information from the soil environment monitoring time series information, wherein the first-region soil environment monitoring time series information includes the first sub-region soil environment monitoring time series information to the M-th sub-region soil environment monitoring time series information; performing similarity analysis on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain a first sub-region soil environment similarity; until similarity analysis is performed on the soil environment prediction time series information of the M-th sub-region and the soil environment monitoring time series information of the M-th sub-region to obtain the soil environment similarity of the M-th sub-region; Construct consistency comparison rules: when the soil environment similarity is greater than the similarity threshold, the soil environment is considered consistent, otherwise it is considered inconsistent; Based on the consistency comparison rule, processing the soil environment similarity of the first sub-region until the soil environment similarity of the M-th sub-region to obtain inconsistent sub-regions; Based on the inconsistent sub-region, the soil environment monitoring time series information and the air environment monitoring time series information are suspiciously marked to obtain first region suspicious marking data, which is added to the full region suspicious marking data.

6. The method according to claim 5, wherein Also includes: Count the proportion of inconsistent sub-regions to the total number of sub-regions in the first region, and set it as the inconsistent proportion of the first region; When the inconsistency ratio of the first region is greater than the inconsistency ratio threshold, based on the inconsistent sub-regions, the soil environment monitoring time series information and the air environment monitoring time series information are marked as suspicious; When the inconsistency ratio of the first area is less than or equal to the inconsistency ratio threshold, the air environment monitoring time series information is marked as suspicious.

7. The method according to claim 5, wherein Performing similarity analysis on the first sub-region soil environment prediction time series information and the first sub-region soil environment monitoring time series information to obtain the first sub-region soil environment similarity includes: Extracting a soil environment prediction value at a first moment from the first sub-region soil environment prediction time series information; Extracting a soil environment monitoring value at a first moment from the soil environment monitoring time series information of the first sub-region; Calculating the same attribute standard deviation of the soil environment prediction value and the soil environment monitoring value at the first moment to obtain several attribute standard deviations; Traversing the plurality of attribute standard deviations and performing dimensionless processing to obtain a plurality of deviation eigenvalues; The plurality of deviation characteristic values ​​are respectively used as a plurality of dimensional distance parameters, and the Euclidean distance is calculated, and then the reciprocal is obtained to obtain the soil environment similarity of the first sub-region.

8. A greenhouse environment data monitoring platform with multi-sensor fusion, characterized by: Used to implement the method according to any one of claims 1 to 7, comprising: A monitoring information acquisition module is configured to receive soil environment monitoring time series information, air environment monitoring time series information, and crop growth information of a target area, wherein the crop growth information represents the degree of crop growth during the environmental monitoring time window; An environmental information extraction module is used to extract soil environmental information at the initial moment of the air environment monitoring time series information from environmental monitoring historical information; a prediction information statistics module, configured to collect a soil state monitoring sample set and to collect statistics on the soil environment prediction time series information, using the soil environment information at the initial moment as an initial soil state constraint, and using the air environment monitoring time series information and the crop growth information as soil state impact constraints; The monitoring information identification module is used to mark the soil environment monitoring time series information and the air environment monitoring time series information as suspicious when the soil environment prediction time series information and the soil environment monitoring time series information are inconsistent, and send them to the user end.

Images