Intelligent frost intelligent monitoring method and system

Abstract

The invention discloses an intelligent frost intelligent monitoring method and system. The method comprises the steps of meteorological data acquisition, baseline standardization, frost factor correction, factor difference distance construction, initial risk center identification, frost risk grade iterative optimization and frost intelligent monitoring. The invention belongs to the field of data processing, and particularly relates to an intelligent frost monitoring method and system.According to the scheme, by monitoring unit individual meteorological baseline standardization, false alarms caused by characteristic differences are reduced; the time decline characteristic is introduced, so that abnormal meteorological changes are more striking; a monitoring unit factor difference span is constructed through a variational degree coefficient, and noise interference of low variation factors is reduced; introducing a weighting function to construct a risk level overlapping degree, and iteratively re-clustering the monitoring units judged to be wrongly allocated; and a monitoring unit factor comprehensive difference scale is constructed, and an adaptive mahalanobis distance is introduced for correction, so that the frost monitoring reliability is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to an intelligent frost monitoring method and system. Background Technology

[0002] Intelligent frost monitoring methods are technical means to analyze key influencing factors of frost occurrence and identify potential frost risks based on meteorological monitoring data and intelligent technologies such as machine learning and data mining. However, general intelligent frost monitoring methods do not consider the regional climate and crop characteristics differences among different monitoring units, and ignore the timeliness of meteorological factor changes, leading to frequent false monitoring and inaccurate monitoring results. Furthermore, general intelligent frost monitoring methods cannot adaptively adjust to the complexity of meteorological factors and the overlap of frost risk levels, and do not properly consider the correlation between meteorological factors, resulting in low reliability of frost monitoring. Summary of the Invention

[0003] To address the above issues and overcome the shortcomings of existing technologies, this invention provides an intelligent frost monitoring method and system. Addressing the problem that general frost monitoring methods fail to consider regional climate and crop characteristics differences among monitoring units and neglect the timeliness of meteorological factor changes, leading to frequent false alarms and inaccurate results, this solution standardizes individual meteorological baselines for each monitoring unit, establishing a dedicated climate reference system for each unit. This system is sensitive only to meteorological abrupt changes within the monitoring unit itself, reducing false alarms caused by regional climate / crop characteristic differences. It introduces time decay characteristics, giving higher weight to recent meteorological data and assigning faster decay to highly volatile meteorological factors, making abnormal meteorological changes more noticeable. Furthermore, it constructs the factor difference span of monitoring units through variational degree coefficients, more sensitively distinguishing monitoring units with different risk levels and reducing noise interference from low-variability factors. This further improves... To improve the accuracy of frost monitoring, this scheme addresses the problem that general intelligent frost monitoring methods cannot adaptively adjust to the complexity of meteorological factors and the overlap of frost risk levels, and that inadequate consideration of the correlation between meteorological factors leads to low reliability of frost monitoring. This scheme weights and mixes the local risk difference amplitude and the global risk difference span of monitoring units based on the risk level overlap to obtain an adaptive similarity metric. A weighting function is introduced to construct the risk level overlap, accurately assessing the confusion of monitoring units at level boundaries. The similarity metric is adaptively adjusted according to the overlap of different frost risk levels, retaining sensitivity to any group when level distinctions are significant. Monitoring units judged as misassigned are iteratively re-clustered. A comprehensive factor difference scale for monitoring units is constructed, and an adaptive Mahalanobis distance is introduced for correction, improving the reliability of frost monitoring.

[0004] The technical solution adopted by this invention is as follows: This invention provides an intelligent frost monitoring method, which includes the following steps:

[0005] Step S1: Meteorological data collection;

[0006] Step S2: Baseline standardization;

[0007] Step S3: Frost factor correction;

[0008] Step S4: Constructing factor difference distances;

[0009] Step S5: Initial risk center identification;

[0010] Step S6: Iterative optimization of frost risk level;

[0011] Step S7: Intelligent Frost Monitoring.

[0012] Furthermore, in step S1, the meteorological data acquisition involves obtaining historical meteorological monitoring data, marking frost risk levels, and encoding the non-numerical characteristics of the acquired data.

[0013] Furthermore, in step S2, the baseline standardization involves maintaining a rolling baseline for the influencing factors of each monitoring unit, referencing only its own historical data; and normalizing the factor values ​​to obtain the initial meteorological monitoring dataset.

[0014] Furthermore, in step S3, the frost factor correction is to perform frost factor correction by introducing time decay characteristics into the normalized factor obtained in step S2 to obtain the corrected value; thereby obtaining the final meteorological monitoring dataset.

[0015] Further, in step S4, the construction of the factor difference distance involves calculating the factor variation degree, using the variation coefficient to measure the factor's volatility across all monitoring units, and constructing the factor difference distance.

[0016] Further, in step S5, the initial risk center identification involves calculating the local risk density to measure the degree of risk clustering around the monitoring unit; calculating the nearest neighbor risk distance to measure the distance between the monitoring unit and risk points with higher density than itself; and determining the initial risk center.

[0017] Furthermore, in step S6, the iterative optimization of the frost risk level specifically includes:

[0018] Step S61: Calculate the overlap of risk levels; calculate the nearest distance from the monitoring unit to other risk levels; calculate the average distance threshold, construct the overlap index, and introduce a weighting function;

[0019] Step S62: Construct the global risk difference span; after each iteration, update the risk center based on the monitoring unit data within the current risk level; construct the global risk difference span of the monitoring units; and design the measurement and adjustment coefficient;

[0020] Step S63: Construct a comprehensive difference scale for monitoring unit factors and introduce adaptive Mahalanobis distance for correction;

[0021] Step S64: Iterative reassignment; calculate the global risk difference span of all monitoring units, and mark those that do not match the initial clustering labels as misassigned; recalculate the risk level overlap, and for each misassigned monitoring unit, reassign the frost risk group label according to the comprehensive difference scale of monitoring unit factors; the frost risk group label selection rule is: select the label with the largest amount of actual frost record data in the current group as the label of the group; if the misassigned set or the risk level overlap no longer changes, stop, and the meteorological monitoring dataset clustering is completed.

[0022] Furthermore, in step S7, the frost intelligent monitoring is based on the clustering results of the meteorological monitoring dataset, real-time collection of meteorological monitoring data, participation in a clustering iteration of historical meteorological monitoring data, and assignment of labels based on the corresponding frost risk groups; frost intelligent monitoring is then performed based on the assigned labels.

[0023] The present invention provides an intelligent frost monitoring method and system, comprising a meteorological data acquisition module, a baseline standardization module, a frost factor correction module, a factor difference distance construction module, an initial risk center identification module, a frost risk level iterative optimization module, and a frost intelligent monitoring module;

[0024] The meteorological data acquisition module obtains historical meteorological monitoring data and marks the frost risk level;

[0025] The baseline standardization module maintains a rolling baseline for each monitoring unit and generates an initial meteorological monitoring dataset.

[0026] The frost factor correction module corrects the normalization factor to obtain the final meteorological monitoring dataset;

[0027] The factor difference distance construction module constructs a highly variational factor-weighted factor difference distance to measure the comprehensive differences between monitoring units;

[0028] The initial risk center identification module filters initial risk centers based on factor difference distance;

[0029] The frost risk level iterative optimization module defines the risk level overlap and the global risk difference span, iteratively reassigns misassigned monitoring units, and completes the clustering processing of the final meteorological monitoring dataset.

[0030] The frost intelligent monitoring module performs frost intelligent monitoring based on clustering results and real-time meteorological monitoring data.

[0031] The beneficial effects achieved by the present invention using the above solution are as follows:

[0032] (1) In view of the fact that general frost intelligent monitoring methods do not consider the regional climate and crop characteristics differences of different monitoring units and ignore the timeliness of meteorological factor changes, which leads to frequent false monitoring and inaccurate monitoring results, this scheme establishes an exclusive climate reference system for each monitoring unit by standardizing the individual meteorological baseline of the monitoring unit. It is only sensitive to meteorological changes of the monitoring unit itself, reducing false alarms caused by regional climate / crop characteristics differences; introduces time decay characteristics to give higher weight to recent meteorological data and give high-fluctuation meteorological factors faster decay, making abnormal meteorological changes more noticeable; constructs the factor difference span of monitoring units through variation degree coefficient, more sensitively distinguishes monitoring units of different risk levels, and reduces noise interference from low-variable factors; thereby improving the accuracy of frost monitoring.

[0033] (2) To address the problem that general intelligent frost monitoring methods cannot adaptively adjust to the complexity of meteorological factors and the overlap of frost risk levels, and that inappropriate consideration of the correlation between meteorological factors leads to low reliability of frost monitoring, this scheme obtains an adaptive similarity measure by weighting and mixing the magnitude of local risk differences and the span of global risk differences of monitoring units based on the overlap of risk levels; introduces a weight function to construct the overlap of risk levels to accurately assess the confusion of monitoring units at the level boundaries; adaptively adjusts the similarity measure according to the overlap of different frost risk levels, and retains the sensitivity to any group when the level distinction is obvious; iteratively re-clusters the monitoring units judged as misassigned; constructs a comprehensive difference scale of monitoring unit factors, and introduces an adaptive Mahalanobis distance for correction to improve the reliability of frost monitoring. Attached Figure Description

[0034] Figure 1 A flowchart illustrating an intelligent frost monitoring method provided by the present invention;

[0035] Figure 2 This is a schematic diagram of an intelligent frost monitoring system provided by the present invention.

[0036] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0037] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0038] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0039] Example 1, see Figure 1 The present invention provides an intelligent frost monitoring method, which includes the following steps:

[0040] Step S1: Meteorological data collection; acquire historical meteorological monitoring data and mark the frost risk level;

[0041] Step S2: Baseline standardization; maintain a rolling baseline for each monitoring unit and generate an initial meteorological monitoring dataset;

[0042] Step S3: Frost factor correction; Correct the normalization factor to obtain the final meteorological monitoring dataset;

[0043] Step S4: Constructing factor difference distances; Constructing highly variational factor-weighted factor difference distances to measure the overall differences between monitored units;

[0044] Step S5: Initial risk center identification; initial risk centers are screened based on factor difference distance;

[0045] Step S6: Iterative optimization of frost risk levels; define the overlap of risk levels and the span of global risk differences, iteratively redistribute misassigned monitoring units, and complete the clustering processing of the final meteorological monitoring dataset;

[0046] Step S7: Intelligent frost monitoring; Based on the clustering results, perform intelligent frost monitoring on real-time meteorological monitoring data.

[0047] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, meteorological data acquisition involves obtaining historical meteorological monitoring data, including minimum air temperature, surface temperature, air humidity, wind speed, radiative cooling rate, crop growth period, historical frost occurrence frequency, and crop frost damage severity; and labeling frost risk levels, including no risk, mild risk (minor frost damage to leaves), moderate risk (frost damage to some branches), and severe risk (plant death); and encoding the non-numerical features of the collected data.

[0048] Example 3, see Figure 1This embodiment is based on the above embodiment. In step S2, baseline standardization is performed because different regions and crops have different sensitivity baselines to frost. Therefore, a dedicated climate baseline is established for each monitoring unit, which is only sensitive to abrupt changes in its own historical fluctuations. Specifically, a rolling baseline is maintained for the impact factors (data dimensions) of each monitoring unit, referring only to its own historical data, as shown below: Normalize the factor values ​​to highlight abrupt changes relative to the baseline, as shown below: ;in, is the rolling baseline of the d-th dimension factor for the i-th monitoring unit; W is the size of the rolling window; It is the time offset; It is the factor value of the d-th dimension of the i-th monitoring unit at the corresponding time. It is the current point in time; It is a normalization factor; It is the standard deviation of the d-th dimension factor value; thus, the initial meteorological monitoring dataset is obtained.

[0049] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S3, frost factor correction is performed because frost occurrence is strongly correlated with recent meteorological changes. Historical data of high-fluctuation factors decay faster, while historical data of low-fluctuation factors are retained longer, ensuring that recent anomalies are more noticeable. Therefore, frost factor correction is performed to highlight the monitoring value of recent anomalies. Specifically, the normalized factor obtained in step S2 is given a time decay characteristic, and the corrected value is expressed as follows: ; ;in, is the correction value of the d-th dimension factor for the i-th monitoring unit; L is the window size for time decay; It is the reference standard deviation, which is the mean of the standard deviations of all factors; It is the time decay effect coefficient; thus, the final meteorological monitoring dataset is obtained.

[0050] Example 5, see Figure 1 This embodiment is based on the above embodiment. In step S4, the factor difference distance construction is based on the different distinguishing abilities of different meteorological factors for frost. By constructing the factor difference distance, the differences between two monitoring units on all factors are measured, with high variation factors contributing more. The specific operation is as follows: calculate the factor variation degree. The variational coefficient measures the variability of a factor across all monitored units, and is expressed as: ; ; ;in, It is the global mean of the d-th factor; It represents the global standard deviation; N is the total number of monitored units, and i is the index of the monitored unit; construct the factor difference distance. , is represented as: Where M is the total number of factors; It is the correction value of the d-th dimension factor for the j-th monitoring unit.

[0051] By performing the above operations, this solution addresses the problem that general intelligent frost monitoring methods fail to consider regional climate and crop characteristics differences among different monitoring units and neglect the timeliness of meteorological factor changes, leading to frequent false alarms and inaccurate monitoring results. It standardizes individual meteorological baselines for each monitoring unit, establishing a dedicated climate reference system for each unit. This system is sensitive only to meteorological abrupt changes within the monitoring unit itself, reducing false alarms caused by regional climate / crop characteristic differences. Furthermore, it introduces time decay characteristics, giving higher weight to recent meteorological data and assigning faster decay to highly volatile meteorological factors, making abnormal meteorological changes more noticeable. Finally, it constructs the factor difference span of monitoring units through variational degree coefficients, more sensitively distinguishing monitoring units with different risk levels and reducing noise interference from low-variability factors, thereby improving the accuracy of frost monitoring.

[0052] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the initial risk center is identified based on the factor difference distance, through local density and nearest neighbor distance; specifically, the local risk density is calculated. The degree of risk clustering around the monitoring unit is measured and expressed as: ; It is the local risk density of the i-th monitoring unit; calculate the nearest neighbor risk distance. The distance between a monitoring unit and risk points with a higher density than itself is expressed as: Determine the initial risk centers (cluster centers) and construct a two-dimensional coordinate system. Select K (4) items The largest point is designated as the initial risk center, and other monitoring units are assigned to the frost risk group corresponding to the nearest risk center with a lower density; among them... It is an indicator function. The function value is 1 when the condition inside the parentheses is met, and 0 otherwise. x is an exponential variable. It is the density threshold, taking all... the median; It is the local risk density of the j-th monitoring unit; It is the monitoring unit with the highest overall density.

[0053] Example 7, see Figure 1This embodiment is based on the above embodiment. In step S6, the iterative optimization of frost risk levels is affected by the ambiguity of the boundaries between different risk levels. Therefore, the weights are adjusted by the overlap of risk levels: high overlap enhances global stability (avoiding over-subdivision), while low overlap preserves local sensitivity (accurately capturing subtle differences), ultimately outputting a reliable risk cluster. The specific operation is as follows:

[0054] Step S61: Calculate the risk level overlap; calculate the nearest distance from the monitoring unit to other risk levels. , is represented as: ; Calculate the average distance threshold , is represented as: To construct an overlap index R, considering the differences in risk levels, since the overlap between different risk levels varies, a weighting function is introduced. The risk level overlap R is represented as: ; ;in, It is the risk level corresponding to the risk center that is closest to the i-th monitoring unit and has a lower density than itself; It represents the risk level of the i-th monitoring unit;

[0055] Step S62: Construct the global risk difference span; after each iteration, update the risk center based on the monitoring unit data within the current risk level, represented as: ; Construct the span of global risk differences among monitoring units , is represented as: ;in, It is the risk center of the k-th frost risk group on the d-th dimension factor; It is the total number of monitoring units in the kth frost risk group; It is the d-th dimension factor value belonging to the i-th monitoring unit of the k-th frost risk group; and a moderating coefficient is designed to measure it. and , respectively represented as: ; K is the initial number of risk centers;

[0056] Step S63: Construct a comprehensive difference scale for monitoring unit factors To address the correlation between monitoring unit factors and the fact that the covariance matrix of meteorological monitoring data changes with clustering iterations, a fixed covariance matrix cannot accurately reflect the distribution of meteorological monitoring data within the current frost risk group. Therefore, an adaptive Mahalanobis distance is introduced. After correction, the overall difference scale of the monitoring unit factors is expressed as follows: ; ; ; ;in, and These are the covariance matrices of meteorological monitoring data during the nth and (n-1)th iterations within the kth frost risk group, respectively. This is the weighting coefficient, initially set to 0.7; It is the mean vector of the kth frost risk group; It is the local difference scale between the i-th monitoring unit and the k-th frost risk group; It is a weighting factor; x i and x j These are the meteorological factor vectors for the i-th and j-th monitoring units, respectively;

[0057] Step S64: Iterative reassignment; calculate the global risk difference span of all monitoring units, and mark those that do not match the initial clustering labels as misassigned; recalculate the risk level overlap, and for each misassigned monitoring unit, reassign the frost risk group label according to the comprehensive difference scale of monitoring unit factors (nearest match); the frost risk group label selection rule is: select the label with the largest amount of actual frost record data in the current group as the label of the group; if the misassigned set or the risk level overlap no longer changes, stop, and the meteorological monitoring dataset clustering is completed;

[0058] For the clustering parameters of historical meteorological monitoring data, grid search optimization was used, and the clustering results were evaluated using the risk level matching rate as the evaluation index to verify the clustering effect.

[0059] By performing the above operations, this scheme addresses the problem that general intelligent frost monitoring methods cannot adaptively adjust to the complexity of meteorological factors and the overlap of frost risk levels, and that inadequate consideration of the correlation between meteorological factors leads to low reliability of frost monitoring. Instead, it weights and mixes the magnitude of local risk differences and the span of global risk differences between monitoring units based on the overlap of risk levels to obtain an adaptive similarity metric. A weighting function is introduced to construct the risk level overlap, accurately assessing the confusion of monitoring units at level boundaries. The similarity metric is adaptively adjusted according to the overlap of different frost risk levels, retaining sensitivity to any group when level distinctions are significant. Monitoring units judged as misassigned are iteratively re-clustered. A comprehensive factor difference scale for monitoring units is constructed, and an adaptive Mahalanobis distance is introduced for correction, improving the reliability of frost monitoring.

[0060] Example 8, see Figure 1This embodiment is based on the above embodiment. In step S7, the intelligent frost monitoring is based on the clustering results of the meteorological monitoring dataset. Meteorological monitoring data is collected in real time and participates in a clustering iteration of historical meteorological monitoring data. Labels are assigned based on the corresponding frost risk groups. Intelligent frost monitoring is performed based on the assigned labels. A risk threshold is set. If a monitoring unit is assigned to a moderate or severe frost risk group and the factor comprehensive difference scale with the monitoring unit in the no-risk group is higher than the risk threshold, an early warning is triggered. If an early warning is triggered within three consecutive time windows, an early warning notification is pushed to the management personnel. Historical meteorological monitoring data is updated periodically for clustering iteration.

[0061] Example 9, see Figure 2 Based on the above embodiments, the present invention provides an intelligent frost monitoring method system, including a meteorological data acquisition module, a baseline standardization module, a frost factor correction module, a factor difference distance construction module, an initial risk center identification module, a frost risk level iterative optimization module, and a frost intelligent monitoring module.

[0062] The meteorological data acquisition module obtains historical meteorological monitoring data and marks the frost risk level;

[0063] The baseline standardization module maintains a rolling baseline for each monitoring unit and generates an initial meteorological monitoring dataset.

[0064] The frost factor correction module corrects the normalization factor to obtain the final meteorological monitoring dataset;

[0065] The factor difference distance construction module constructs a highly variational factor-weighted factor difference distance to measure the comprehensive differences between monitoring units;

[0066] The initial risk center identification module filters initial risk centers based on factor difference distance;

[0067] The frost risk level iterative optimization module defines the risk level overlap and the global risk difference span, iteratively reassigns misassigned monitoring units, and completes the clustering processing of the final meteorological monitoring dataset.

[0068] The frost intelligent monitoring module performs frost intelligent monitoring based on clustering results and real-time meteorological monitoring data.

[0069] It should be noted that, in this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0071] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An intelligent frost intelligent monitoring method, characterized in that: The method comprises the following steps: Step S1: meteorological data collection; obtain historical meteorological monitoring data, and label frost risk levels; Step S2: baseline standardization; maintain a rolling baseline for each monitoring unit, and generate an initial meteorological monitoring data set; Step S3: frost factor correction; correct the normalized factor to obtain a final meteorological monitoring data set; Step S4: factor difference distance construction; construct a high-variational factor weighted factor difference distance to measure the comprehensive difference between monitoring units; Step S5: initial risk center identification; select an initial risk center based on the factor difference distance; Step S6: iterative optimization of frost risk levels; define risk level overlap and global risk difference span, iteratively reassign misassigned monitoring units, and complete clustering processing of the final meteorological monitoring data set; Step S7: intelligent frost monitoring; based on the clustering results, perform intelligent frost monitoring on real-time meteorological monitoring data.

2. The intelligent frost intelligent monitoring method according to claim 1, characterized in that: In step S2, the baseline standardization is to maintain a rolling baseline for the influence factor of each monitoring unit, and only refer to its own historical data; normalize the factor value to obtain an initial meteorological monitoring data set.

3. The intelligent frost monitoring method according to claim 2, characterized in that: In step S3, the frost factor correction is to correct the frost factor, introduce the time decay characteristic to the normalized factor obtained in step S2, and obtain the corrected value; and then obtain the final meteorological monitoring data set.

4. The intelligent frost monitoring method according to claim 3, characterized in that: In step S4, the factor difference distance construction is to calculate the factor variation degree, measure the fluctuation ability of the factor in all monitoring units with a variation coefficient, and construct a factor difference distance.

5. The intelligent frost monitoring method according to claim 4, characterized in that: In step S5, the initial risk center identification is to calculate the local risk density to measure the risk aggregation degree around the monitoring unit, calculate the near-neighbor risk distance to measure the distance between the monitoring unit and the risk point with higher density than itself, and determine the initial risk center.

6. The intelligent frost monitoring method according to claim 5, characterized in that: In step S6, the iterative optimization of frost risk levels specifically includes: Step S61: calculate the risk level overlap; calculate the nearest distance of each monitoring unit to other risk levels; calculate the average distance threshold, construct an overlap index, and introduce a weight function; Step S62: construct a global risk difference span; Step S63: construct a monitoring unit factor comprehensive difference scale, and introduce a self-adaptive Mahalanobis distance for correction; Step S64: iterative reassignment; calculate the global risk difference span of all monitoring units, mark those not consistent with the initial clustering label as misassigned; recalculate the risk level overlap, and for each misassigned monitoring unit, reassign the frost risk group label according to the monitoring unit factor comprehensive difference scale; the frost risk group label selection rule is to select the label corresponding to the largest actual frost record data amount in the current group as the label of the group; if the misassigned set or the risk level overlap no longer changes, stop, and the meteorological monitoring data set clustering is completed.

7. The intelligent frost monitoring method according to claim 6, characterized in that: In step S6, the construction of the global risk difference span is to update the risk center based on the monitoring unit data in the current risk level after each iteration; construct the global risk difference span of the monitoring unit; and design a measure adjustment coefficient.

8. The intelligent frost monitoring method according to claim 7, characterized in that: In step S7, the frost intelligent monitoring is based on the clustering result of the meteorological monitoring data, real-time meteorological monitoring data is collected, a clustering iteration is performed on the historical meteorological monitoring data, and a label is assigned based on the corresponding frost risk group; The frost intelligent monitoring is based on the assigned label.

9. An intelligent frost intelligent monitoring method system for implementing the intelligent frost intelligent monitoring method of any one of claims 1-8, characterized in that: The system comprises a meteorological data acquisition module, a baseline standardization module, a frost factor correction module, a factor difference distance construction module, an initial risk center identification module, a frost risk level iterative optimization module and a frost intelligent monitoring module; The meteorological data acquisition module acquires historical meteorological monitoring data and labels the frost risk level; The baseline standardization module maintains a rolling baseline for each monitoring unit and generates an initial meteorological monitoring data set; The frost factor correction module corrects the normalized factor to obtain a final meteorological monitoring data set; The factor difference distance construction module constructs a high-variational factor weighted factor difference distance to measure the comprehensive difference between monitoring units; The initial risk center identification module screens the initial risk center based on the factor difference distance; The frost risk level iterative optimization module defines the risk level overlap degree and the global risk difference span, iteratively reassigns misassigned monitoring units, and completes the clustering processing of the final meteorological monitoring data set; The frost intelligent monitoring module performs frost intelligent monitoring on real-time meteorological monitoring data based on the clustering result.

Images