Intelligent maintenance method and system for three-dimensional greening

Abstract

The invention relates to the technical field of intelligent maintenance, in particular to an intelligent maintenance method and system for three-dimensional greening, and the method comprises the following steps: extracting partition water content tracks based on physiological characterization and conductive characteristics, recognizing abnormal attributes, analyzing the consistency of demand gradient and microclimate vector to form a cross-region data set, and obtaining a cross-region data set; and quantifying the physiological performance supply and demand deviation matching maintenance compensation model, outputting a regulation and control strategy set, calling the strategy set to compare the risk level with the plant protection sequence, and outputting a maintenance adjustment partition list. According to the method, the sensing sensitivity of environment sudden change is effectively improved by identifying a synchronous abnormal region, cross-region maintenance requirements are accurately positioned by utilizing microclimate vector comparison, linkage identification of a defect region is realized, regulation and control parameters are automatically matched through multi-dimensional coupling analysis, the pertinence of a maintenance strategy is remarkably enhanced, and the coverage range is dynamically adjusted according to the risk level. The resource allocation efficiency is optimized, the three-dimensional greening growth balance and ecological stability are ensured, and the problem of maintenance lag is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent maintenance technology, and in particular to an intelligent maintenance method and system for vertical greening. Background Technology

[0002] The field of intelligent maintenance technology mainly involves the intelligent monitoring, management, and maintenance of green spaces through modern information technologies such as the Internet of Things, remote sensing, artificial intelligence, and data analysis. With the continuous development of urban green space construction, especially the promotion of innovative technologies such as "pocket parks" and vertical greening, intelligent maintenance has become crucial for the continuous optimization of green spaces. The core aspects of intelligent maintenance technology include green space monitoring based on environmental data, real-time data acquisition and processing, environmental change analysis, intelligent decision support, and automatic control. It utilizes intelligent technologies to accurately analyze and scientifically manage the ecological status of green spaces. This technology not only covers the maintenance and management of green spaces but also involves monitoring plant growth status, soil and water monitoring, and pest and disease control, in order to achieve refined and dynamic management of ecological green spaces.

[0003] Traditional intelligent maintenance methods for vertical greening refer to the real-time monitoring and precise control of vertical greening spaces through technologies such as remote sensing, the Internet of Things, and artificial intelligence. This method collects information on plant growth status, environmental parameters, and soil data within the green space, uses an intelligent control system to determine the plants' maintenance needs, and automatically adjusts water and fertilizer management, environmental regulation, and pest and disease control based on data analysis. Traditional methods rely on pre-set programs or rules for periodic environmental checks and manual adjustments, lacking real-time dynamic adjustment and intelligent feedback capabilities, making it difficult to cope with rapidly changing environmental conditions and unforeseen circumstances.

[0004] Existing technologies rely on preset programs or rules to periodically inspect and manually adjust vertical greening spaces. The static management model makes it difficult to capture the instantaneous changes in the plant microenvironment in the vertical space in real time. As a result, the system cannot respond in time to the real physiological needs of plants when faced with complex and changeable climatic conditions or sudden environmental pressures. It lacks the ability to dynamically perceive the connectivity and microclimate gradient of different areas in the vertical space. When the water and heat fields in a local area are unbalanced, it cannot form a coordinated replenishment, resulting in uneven or delayed allocation of maintenance resources. It is difficult to maintain the long-term stability and growth consistency of the ecosystem, and it is very easy to cause local plant growth to be hindered or the overall landscape effect to degrade. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose an intelligent maintenance method and system for vertical greening.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent maintenance method for vertical greening, comprising the following steps:

[0007] S1: Based on the distribution of plant physiological characteristics on the surface of the three-dimensional greening structure and the conductivity of the root matrix, the detection zones are divided, the water content trajectory and response curve of the zones are extracted, the synchronous abnormal areas of physiological fluctuations and environmental changes are identified, the numbers and coordinates of the synchronous abnormal areas are extracted, and a set of abnormal maintenance attributes of the greening zones is generated.

[0008] S2: Based on the set of abnormal attributes of greening zone maintenance, identify the demand gradient and microclimate vector of the zone, analyze the cosine similarity of the angle between the demand gradient vector and the microclimate vector, filter the abnormal connectivity zones in combination with connectivity, and form a cross-zone maintenance connectivity data group based on the abnormal connectivity zones.

[0009] S3: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporation heat field characteristics within the partition, analyze the uniformity of moisture content and the consistency of heat field, screen and match maintenance gap areas, and obtain maintenance defect linkage area groups.

[0010] S4: Based on the maintenance defect linkage area group, quantitatively analyze the parameter differences between the physiological performance of the plants in the partition and the preset standard growth benchmark. Based on the degree of deviation, match the preset maintenance compensation model, calculate the water and fertilizer regulation parameters and environmental intervention intensity of the partition, and output the self-matching maintenance regulation strategy set of the three-dimensional greening partition.

[0011] As a further embodiment of the present invention, the set of abnormal attributes for greening zoning maintenance includes physiological mutation zoning numbers, abnormal environmental response points, zoning coordinate markers, and time series mutation identifiers; the cross-zone maintenance connectivity data set includes growth gradient abnormal zoning identifiers, zoning boundary connectivity units, and zoning boundary consistency blocks; the maintenance defect linkage area group includes continuous zoning zones with abnormal water content, irregular transpiration heat fields, overlapping areas of matrix performance mutations, and linkage abnormal zoning numbers; and the set of self-matching maintenance control strategies for three-dimensional greening zoning includes risk level labels, zoning response deviation values, local physiological abnormality indicators, and health status deviation levels.

[0012] As a further aspect of the present invention, the step of obtaining the abnormal attribute set of greening zone maintenance specifically includes:

[0013] S111: Based on the distribution of plant physiological characteristics on the surface of three-dimensional greening structure and the conductivity of root matrix, detection zones are divided. The water content change trajectory and environmental response curve of the zone are extracted by a multi-dimensional sensor array preset on the three-dimensional greening support. The two types of data in the same zone are dimensionless normalized, the sequence deviation of the normalized data is calculated, and the trend value of the difference between physiological and environmental response is obtained.

[0014] S112: Based on the trend value of the difference between physiological and environmental responses, identify the fluctuation range in the moisture content change curve and the abrupt change point of the environmental response trajectory, perform time alignment processing on the two types of values, extract the time interval where the fluctuation exceeds the preset benchmark value and the abrupt change point is within the set range, and generate a set of high-frequency abnormal interval time periods.

[0015] S113: For the set of high-frequency abnormal interval time periods, retrieve the corresponding partition number and three-dimensional spatial coordinate information, extract the partition location where the abnormal signal occurred, and generate a set of abnormal attributes for greening partition maintenance.

[0016] As a further aspect of the present invention, the step of acquiring the cross-regional maintenance connectivity data set specifically includes:

[0017] S211: Identify the maintenance demand gradient direction and microclimate response vector of the greening zone in the set of abnormal maintenance attributes of the greening zone, extract the projection trajectory of the two at the boundary of the zone, identify the distribution quantity and aggregation degree of the boundary point in the spatial topology of the zone, and obtain the boundary consistency data matrix of the zone.

[0018] S212: Based on the partition boundary consistency data matrix, filter the boundary areas with a aggregation degree higher than the average level, compare the spatial boundaries of the overall structure map of the three-dimensional greening, identify the boundary clusters that have physical continuity and belong to the same detection partition, and obtain the abnormal growth gradient zoning within the greening partition.

[0019] S213: Based on the abnormal growth gradient zoning within the greening zone, an integrated analysis is performed on the consistency of zone boundaries, the dispersion of demand gradient, the uniformity of substrate moisture content, and the delay of physiological response. Based on the response blocks in the zone, a cross-zone maintenance connectivity data set is formed.

[0020] As a further aspect of the present invention, the step of obtaining the maintenance defect linkage area group specifically includes:

[0021] S311: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporative heat field characteristics of the numbered partitions within the partition, perform timestamp alignment processing on the heterogeneous data within the partition, identify the moisture content fluctuation value and the heat field consistency offset, and obtain the greening local maintenance defect response feature set.

[0022] S312: Based on the greening local maintenance defect response feature set, the uniformity of moisture content and the consistency of transpiration heat field within the partition are jointly analyzed. The moisture content value of each unit in the feature set and the average moisture content of all units are extracted. The square term of the difference between the two is calculated and summed by traversing all units to construct a spatial fluctuation term of moisture content that represents the overall uniformity of moisture. The temperature value of the unit at the corresponding position and the average temperature of all units are extracted simultaneously. The spatial fluctuation term of heat field that represents the consistency of heat field is generated by using the same difference square summation logic. The relative ratio of the spatial fluctuation term of moisture content is calculated based on the spatial fluctuation term of heat field. The coupling feature value of moisture content heat field that represents the synergistic evolution law of moisture and heat distribution is derived. Based on the extracted coupling feature value of moisture content heat field, partition units with abnormal coupling degree of moisture content heat field in the partition are screened, and a spatial distribution map of the synergistic response of moisture content heat field is established.

[0023] S313: Call the spatial distribution map of the coordinated response of the moisture content thermal field, perform spatial clustering analysis on the partitions in the coupled feature value partitions that exceed the coordinated identification benchmark, label the partition codes corresponding to the continuous abnormal areas, and obtain the maintenance defect linkage area group.

[0024] As a further aspect of the present invention, the specific steps for obtaining the self-matching maintenance and control strategy set for vertical greening zones are as follows:

[0025] S411: Based on the maintenance defect linkage area group, extract the physiological performance distribution curve of the partition under the specified number, perform time normalization processing, identify the physiological performance change per unit time, and obtain the partition growth element deficiency vector.

[0026] S412: Input the deficit vector of the growth elements in the region into the multi-objective maintenance decision optimization model, and combine the microclimate characteristics of the region to calculate the matrix replenishment flux and environmental compensation parameters required to reach the health restoration threshold, thus forming a set of maintenance and restoration elements in the region.

[0027] S413: Based on the set of maintenance and repair elements in the zoning, the actuators in the three-dimensional greening structure are spatiotemporally coordinated and arranged, the spatial coordinates of the zoning are bound to the actuator ID, and control instructions including start time, duration and flow intensity are obtained, and the set of self-matching maintenance and control strategies for the three-dimensional greening zoning is output.

[0028] As a further aspect of the present invention, the method further includes step S5:

[0029] S5: Call the self-matching maintenance and control strategy set of the vertical greening zone, identify the corresponding number of the zone in the vertical greening functional layout diagram, retrieve the maintenance execution unit list, compare the risk level with the plant protection priority sequence, filter the zone numbers that need to adjust the maintenance coverage, and output the list of vertical greening maintenance adjustment zone numbers.

[0030] The plant protection priority sequence refers to the ranking based on the rarity of the plant, its criticality in landscape function, and its growth vulnerability.

[0031] The list of adjustment zones for vertical greening maintenance includes the adjustment target zone number, maintenance intensity adjustment parameters, protection priority comparison items, and linkage maintenance trigger types.

[0032] As a further aspect of the present invention, the steps for obtaining the list of zoning numbers for vertical greening maintenance adjustments are as follows:

[0033] S511: Call the self-matching maintenance and control strategy set of the three-dimensional greening zone, extract the zone number in the three-dimensional greening functional layout map, map the zone risk level value with the regional spatial coordinate boundary, identify the zone information corresponding to the plant protection level, and generate a greening zone risk distribution map.

[0034] S512: Based on the risk distribution map of the greening zone, extract the maintenance execution unit number and the current maintenance level, match the zone risk level with the execution unit response level, identify the unit number with insufficient or redundant maintenance coverage, and obtain a list of mismatched maintenance risks in the greening zone.

[0035] S513: Based on the greening zoning maintenance risk mismatch list, retrieve the level number in the plant protection priority sequence, extract the key zoning number that needs to be improved in terms of maintenance coverage, output the adjustment control parameters linked with the original maintenance execution unit in sequence, and output the list of three-dimensional greening maintenance adjustment zoning number.

[0036] The intelligent maintenance system for vertical greening is used to execute the aforementioned intelligent maintenance method for vertical greening. The system includes:

[0037] The environmental monitoring module divides the detection zones based on the distribution of plant physiological characteristics on the surface of the three-dimensional greening structure and the conductivity of the root matrix. It compares the water content fluctuation range and the environmental response mutation point within the same time period, screens out the synchronous abnormal areas of the two, extracts the zone number and spatial coordinates, summarizes the abnormal time period and zone number, and generates a set of abnormal attributes of greening zone maintenance.

[0038] The zoning positioning module, based on the set of abnormal attributes of greening zoning maintenance, identifies the consistency between the direction of maintenance demand gradient and the microclimate response vector, marks the boundary number of the zoning, matches the overall structure map of three-dimensional greening, extracts the zoning number range of the abnormal growth gradient area, and establishes a cross-zone maintenance connectivity data group.

[0039] The linkage analysis module, based on the cross-regional maintenance connectivity data group, retrieves the matrix moisture content distribution matrix and the local evaporation heat field characteristics in the region as continuous data sequences, judges the consistency between the moisture content anomaly boundary connectivity and the heat field, marks the partition number that meets the linkage threshold of both, and outputs the maintenance defect linkage area group.

[0040] The scheduling and early warning module analyzes the distribution trend of physiological performance of the corresponding partition and the degree of deviation of the original health status curve based on the partition number of the maintenance defect linkage area group, extracts the partition number of the deviation trend, completes the level identification according to the risk classification standard, and generates a set of self-matching maintenance control strategies for the three-dimensional greening partition.

[0041] The strategy optimization module, based on the self-matching maintenance and control strategy set of the vertical greening zones, finds the corresponding position number of the risk level zone in the vertical greening functional layout diagram, retrieves the current maintenance execution unit configuration list, compares the matching situation between plant protection priority and the current maintenance level, filters the zones that need to be updated, and outputs a list of vertical greening maintenance adjustment zone numbers.

[0042] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0043] In this invention, by dividing detection zones based on plant physiological characteristics and matrix conductivity and identifying synchronous abnormal areas, the sensitivity to environmental mutations in three-dimensional space is effectively improved. By using connectivity analysis and microclimate vector consistency comparison, cross-regional maintenance needs are accurately located, enabling the linkage identification of maintenance defect areas. By quantifying supply and demand deviations through moisture content and thermal field coupling analysis, water and fertilizer regulation parameters and environmental intervention intensity are automatically matched, significantly enhancing the pertinence and accuracy of maintenance strategies. The maintenance coverage is dynamically adjusted according to risk level and plant protection priority sequence, optimizing resource allocation efficiency and ensuring the growth balance and ecological stability of vertical greening in complex environments, effectively solving the problems of traditional maintenance lag and resource misallocation. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0045] Figure 2 This is a flowchart illustrating the process of obtaining the abnormal attribute set for greening zone maintenance in this invention.

[0046] Figure 3 This is a flowchart illustrating the acquisition of cross-regional maintenance connectivity data sets in this invention.

[0047] Figure 4 This is a flowchart illustrating the process of obtaining the maintenance defect linkage area group in this invention.

[0048] Figure 5 This is a flowchart illustrating the acquisition of the self-matching maintenance and control strategy set for three-dimensional greening zones in this invention.

[0049] Figure 6 This is a flowchart illustrating the process of obtaining the list of zoning numbers for vertical greening maintenance adjustments in this invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0051] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0052] Example 1

[0053] Please see Figure 1 This invention provides a technical solution, an intelligent maintenance method for vertical greening, comprising the following steps:

[0054] S1: Based on the distribution of plant physiological characteristics on the surface of three-dimensional greening structure and the conductivity of root matrix, detection zones are divided. The water content change trajectory and environmental response curve of the zones are extracted. On the time axis, synchronous abnormal areas of physiological state fluctuation range and environmental disturbance mutation point are identified. The corresponding zone number and spatial coordinates are extracted to generate a set of abnormal attributes of greening zone maintenance.

[0055] S2: Based on the set of abnormal attributes of greening zone maintenance, identify the direction of maintenance demand gradient and microclimate response vector of the zone, analyze the consistency of their directions at the boundary of the zone, and combine the zone boundary connectivity to screen the zones with connectivity and abnormal growth gradient to form a cross-zone maintenance connectivity data set.

[0056] S3: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporation heat field characteristics within the zone, analyze the uniformity of moisture content and the consistency of heat field, screen and match maintenance gap areas, and obtain maintenance defect linkage area groups.

[0057] S4: Based on the maintenance defect linkage area group, quantitatively analyze the parameter differences between the physiological performance of plants in the zone and the preset standard growth benchmark. Based on the degree of deviation, match the preset maintenance compensation model, calculate the water and fertilizer regulation parameters and environmental intervention intensity of the zone, and output the self-matching maintenance regulation strategy set of the vertical greening zone.

[0058] S5: Call the self-matching maintenance and control strategy set of vertical greening zones, identify the corresponding number of the zone in the vertical greening functional layout diagram, retrieve the maintenance execution unit list, compare the risk level with the plant protection priority sequence, filter the zone numbers that need to adjust the maintenance coverage, and output the list of vertical greening maintenance adjustment zone numbers.

[0059] Plant protection priority sequence refers to the ranking of plants based on their rarity, criticality to landscape function, and vulnerability to growth.

[0060] The set of abnormal attributes for greening zoning maintenance includes physiological mutation zoning numbers, abnormal environmental response points, zoning coordinate markers, and time series mutation identifiers. The cross-zone maintenance connectivity data set includes growth gradient abnormal zoning identifiers, zoning boundary connectivity units, and zoning boundary consistency blocks. The maintenance defect linkage area group includes continuous zoning with abnormal water content, irregular transpiration heat field areas, overlapping areas of matrix performance mutations, and linkage abnormal zoning numbers. The set of self-matching maintenance control strategies for vertical greening zoning includes risk level labels, zoning response deviation values, local physiological abnormality indicators, and health status deviation levels. The list of vertical greening maintenance adjustment zoning numbers includes adjustment target zoning numbers, maintenance intensity adjustment parameters, protection priority comparison items, and linkage maintenance trigger types.

[0061] Please see Figure 2 The specific steps for obtaining the abnormal attribute set of green space zoning maintenance are as follows:

[0062] S111: Based on the distribution of plant physiological characteristics on the surface of three-dimensional greening structure and the conductivity of root matrix, detection zones are divided. The water content change trajectory and environmental response curve of the zone are extracted by a multi-dimensional sensor array preset on the three-dimensional greening support. The two types of data in the same zone are dimensionless normalized, the sequence deviation of the normalized data is calculated, and the trend value of the difference between physiological and environmental response is obtained.

[0063] Based on the distribution of plant physiological characteristics on the surface of vertical greening structures and the conductivity of the root matrix, detection zones were defined. A multi-dimensional sensor array deployed on the vertical greening facade performed high-frequency data acquisition. A frequency-domain reflectometry soil moisture sensor utilized the frequency response and impedance differences of high-frequency electromagnetic waves in different media to determine the matrix dielectric constant and convert it into volumetric water content. A photosynthetically active radiation sensor and a leaf surface temperature and humidity sensor synchronously recorded microenvironmental parameters, with a sampling interval set to 600 seconds. For the acquired raw time-series data, Kalman filtering was performed to eliminate environmental electromagnetic noise interference. A state transition matrix was constructed to predict the state at the next moment. The predicted values ​​were corrected using the observation matrix to minimize the estimation error covariance. The maximum-minimum normalization method was used to map the matrix water content data to a dimensionless interval of 0 to 1, eliminating the differences in measurement ranges between different sensors. An ideal model of plant transpiration rate was constructed, using real-time acquired photosynthetically active radiation intensity, air temperature, and saturated vapor pressure difference as input variables. The theoretical water consumption trajectory within a specific time window was derived using a simplified form of the Penman-Montis equation. The normalized decrease trajectory of measured matrix moisture content within the same detection zone was aligned with the theoretical water consumption trajectory in the time domain, and the Euclidean distance between the two at corresponding timestamps was calculated point by point. This distance value characterizes the degree of deviation between the actual water use of plants and the theoretical environmental demand, i.e., the trend value of the difference between physiological and environmental responses. If light intensity increases and theoretical transpiration demand rises at a certain moment, but the measured moisture content changes slowly, the calculated difference value will increase significantly. Through continuous rolling calculation over 24 hours, a time series dataset reflecting the matching status of plant root water absorption function and environmental transpiration pull is formed, providing a quantitative basis for subsequent anomaly feature extraction.

[0064] S112: Based on the trend value of the difference between physiological and environmental responses, identify the fluctuation range in the moisture content change curve and the abrupt change point of the environmental response trajectory, perform time alignment processing on the two types of values, extract the time interval where the fluctuation exceeds the preset benchmark value and the abrupt change point is within the set range, and generate a set of high-frequency abnormal interval time periods.

[0065] Based on the trend values ​​of the differences between physiological and environmental responses, a sliding window statistical analysis technique is applied to identify fluctuation ranges in the moisture content change curve. The window length is set to 3600 seconds, and the sliding step size is 600 seconds. Within each time window, the statistical variance of the moisture content data is calculated, and the variance result is compared with a preset stability benchmark value of 0.05. Window values ​​exceeding this benchmark value are marked as fluctuation ranges. This benchmark value is set based on the sensor's static noise test results. Simultaneously, a first-order difference operation on the environmental response trajectory is performed to calculate the slope change rate between adjacent time points. Moments where the absolute value of the slope exceeds a set threshold of 1.5 are identified as environmental abrupt change points. This threshold corresponds to drastic changes in light and temperature caused by cloud cover or gusts of wind. A logical AND operation with dual conditions is performed to overlay the fluctuation ranges and abrupt change points on the time axis, filtering out time segments that are both within the non-steady-state fluctuation range of moisture content and contain at least one environmental response abrupt change point. A benchmark filtering mechanism was introduced, retaining only data segments with fluctuations exceeding the benchmark value by 20% and mutations occurring during the active daytime photosynthetic period, i.e., between 8:00 and 18:00. Continuous time segments satisfying all the above constraints were subjected to union processing to remove transient interferences shorter than 1800 seconds, generating a set of high-frequency anomaly intervals. Monitoring data revealed that the water content variance reached 0.08 between 13:20 and 14:40, and a sudden change in response caused by a rapid increase in ambient temperature was detected at 13:45. This time period was identified as a valid anomaly interval. By eliminating background noise and single environmental disturbances, it was ensured that the extracted time periods accurately corresponded to abnormal plant physiological response events.

[0066] S113: For high-frequency abnormal interval time sets, retrieve the corresponding partition number and three-dimensional spatial coordinate information, extract the partition location where the abnormal signal occurred, and generate a set of abnormal attributes for greening partition maintenance.

[0067] For high-frequency anomalous time intervals, a pre-constructed 3D spatial database of vertical greening is traversed. This database, indexed by the unique code of each spatial grid cell, stores 3D coordinate information including horizontal position, vertical height, and facade orientation vectors. Using the timestamps within the anomalous time intervals as query keys, the partition number of the sensor that generated the anomalous signal at the corresponding time is retrieved, and the geometric center coordinates of that partition are extracted. The anomalous time interval information, the partition's unique code, the 3D spatial coordinates, and the corresponding peak difference values ​​are associated and bound to construct a set of anomalous maintenance attributes for greening partitions, containing four-dimensional attribute features. This set is structured and encapsulated in JSON format to ensure data compatibility and transmission efficiency across different processing modules. During the extraction process, for continuous anomalous signals spanning multiple grid cells, their spatial centroid coordinates are calculated as the location identifier of the anomalous area. Combined with facade orientation vector information, partitions located on different light-receiving surfaces are classified and labeled to distinguish the differences in anomalous attributes between sunny and shady sides. By mapping the original time-only anomalous signals to the physical space, the specific location of the anomaly and its spatial distribution pattern were clarified. This provides a basic data source with accurate spatiotemporal labels for subsequent analysis of the hydraulic conduction characteristics on the vertical gradient and cross-regional connectivity, realizing the dimensional transformation from 1-dimensional time series anomalies to 3-dimensional spatial location anomalies.

[0068] Please see Figure 3 The specific steps for obtaining cross-regional maintenance connectivity data sets are as follows:

[0069] S211: Identify the gradient direction of maintenance demand and microclimate response vector of the greening zone in the set of abnormal maintenance attributes, extract the projection trajectory of the two at the boundary of the zone, identify the distribution quantity and aggregation degree of the boundary point in the spatial topology of the zone, and obtain the boundary consistency data matrix.

[0070] This study identifies the maintenance demand gradient direction and microclimate response vector of each green space zoning zone within anomaly attribute sets. Inverse distance weighted interpolation is used to process the spatial coordinate data of each anomalous zone, constructing a continuous maintenance demand field covering the entire green facade. Within this vector field, the demand gradient vector from the center point of each zone to its eight neighboring zones is calculated. The vector magnitude represents the intensity of demand differences, and the direction indicates the direction from high-potential areas to low-potential areas for water or nutrient transport. Simultaneously, wind speed vector field and light incidence angle data are integrated to construct a microclimate response vector, representing the driving direction of external environmental factors on plant physiological activities. The projection components of the demand gradient vector and microclimate response vector onto the physical boundary lines of the zones are extracted. Vector dot product operations are performed to quantify the synergy or conflict between the two in the transport direction. The distribution density of boundary points in the spatial topology of the zones is statistically analyzed. The K-nearest neighbor algorithm is applied to calculate the spatial aggregation degree of boundary points, with a search radius of 1.0 meter, calculating the total number of anomalous boundary points within this radius. A data matrix of boundary consistency is generated. This map records the aggregation degree values ​​at the boundaries of each spatial grid in the form of a two-dimensional matrix. The value directly reflects the intensity of the superposition effect of demand gradient and microclimate stress in the region. If five abnormal boundary points are detected per square meter in a certain region and the gradient direction is consistent and vertically downward, then the region shows a high aggregation characteristic in the map, indicating that there is a risk of continuous water deficit dominated by gravity flow.

[0071] S212: Based on the boundary consistency data matrix, filter the boundary areas with a higher degree of aggregation than the average level, compare the spatial boundaries of the overall structure map of vertical greening, identify the boundary clusters that have physical continuity and belong to the same detection zone, and obtain the abnormal growth gradient zoning within the greening zone.

[0072] Based on the boundary consistency data matrix, a spatial aggregation threshold of 3 boundary points per square meter was set. This value was derived from the average boundary density of contiguous dead areas in historical maintenance data. The map was binarized, and boundary areas with aggregation exceeding this threshold were selected to form high-risk connectivity candidate areas. The overall structural design drawings of the vertical greening system were used to extract physical boundary vector information such as the location of irrigation pipeline valves, building expansion joints, and waterproof layer partitions. The high-risk connectivity candidate areas were spatially overlaid with the physical boundary vector layer to identify concentrated boundary areas that cross artificial physical separations but logically exhibit continuous abnormal characteristics. A connected component labeling algorithm was applied to traverse all selected grid cells, merging spatially adjacent grids belonging to the same logical abnormality category, and removing isolated noise areas with an area less than 0.25 square meters. The resulting growth gradient anomaly zoning within the greening zones was then output. In this process, if two adjacent sub-regions physically separated by valves both exhibit high-aggregation water shortage characteristics at their boundary, and connectivity analysis confirms the presence of matrix capillary water conduction, then the physical boundary restrictions are ignored, and they are merged into a composite abnormal region spanning the physical boundary. This breaks through the limitations of traditional maintenance management based on physical zoning, and reconstructs the spatial unit of maintenance management based on the actual physiological response characteristics of plants, ensuring that maintenance measures can cover hidden contiguous problem areas that cross the limitations of physical facilities.

[0073] S213: Based on the abnormal zoning of growth gradients within greening zones, an integrated analysis is conducted on the consistency of zone boundaries, the dispersion of demand gradients, the uniformity of substrate moisture content, and the delay of physiological response. Based on the response blocks in the zones, zone matching is performed to form a cross-zone maintenance connectivity data set.

[0074] Based on the anomaly zoning of growth gradients within greening zones, a multi-source feature evaluation matrix was constructed, encompassing four dimensions: boundary consistency, demand gradient dispersion, substrate moisture content uniformity, and physiological response delay. The standard deviation of the gradient vector at each sampling point within a zone was calculated as the demand gradient dispersion, quantifying the fluctuation of internal demand differences. The coefficient of variation of all moisture content sensor readings within a zone was calculated as the substrate moisture content uniformity index, reflecting the evenness of moisture distribution. The time difference between the moment of environmental factor mutation and the moment of plant physiological indicator response was calculated as the physiological response delay. The entropy weight method was used to determine the objective weights of each indicator, with boundary consistency weighted at 0.3, dispersion at 0.2, uniformity at 0.3, and response delay at 0.2. A weighted summation model was used to comprehensively quantify and score each anomaly zone. Based on the scoring results, a clustering matching operation was performed, grouping spatially non-adjacent blocks with a comprehensive score difference of less than 0.05 into the same response group, forming a cross-zone maintenance connectivity data group. Two separate areas located at the bottom and top of the wall, though physically far apart, both exhibited a matrix moisture content variation coefficient greater than 0.4 and a response delay exceeding 1800 seconds. These were therefore matched and grouped into the same hydraulic transport obstruction group. This data set provided a classification basis for subsequently developing homogeneous batch maintenance strategies, achieving grouping of maintenance objects based on pathological characteristics rather than geographical location.

[0075] Please see Figure 4 The specific steps for obtaining the maintenance defect linkage area group are as follows:

[0076] S311: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporative heat field characteristics of the numbered partitions within the partition, perform timestamp alignment processing on the heterogeneous data within the partition, identify the moisture content fluctuation value and the heat field consistency offset, and obtain the greening local maintenance defect response feature set;

[0077] Based on the cross-regional maintenance connectivity data set, a high-resolution infrared thermal imager was used to perform a full-coverage scan of the green facade, acquiring surface temperature distribution images. An affine transformation algorithm was applied to register the thermal imaging pixel coordinate system with the physical grid coordinate system of the moisture content sensor, extracting the average surface temperature within each grid cell as a local evaporative thermal field feature. Addressing the difference in sampling frequency between point-based moisture content data and area-based temperature data, timestamp alignment was performed, and a linear interpolation algorithm was used to fill the time gaps in low-frequency data, ensuring that each set of matrix moisture content data strictly corresponds to the temperature data at the same moment. The deviation between the current moisture content value and the moving average of the past 72 hours was calculated as the fluctuation value, and the offset between the current temperature and the neighborhood average temperature was simultaneously calculated as a thermal field consistency parameter. This constructed a local maintenance defect response feature set for greening, consisting of a series of time-synchronized and spatially aligned moisture content and temperature data pairs. During data processing, temperature anomalies caused by light and shadow or equipment obstruction were removed to ensure that the thermal field characteristics truly reflect the plant transpiration cooling effect. This completed the spatiotemporal fusion of heterogeneous data and laid the data foundation for revealing the deep coupling relationship between water and temperature.

[0078] S312: Based on the response characteristic set of local maintenance defects in greening, a joint analysis of the uniformity of moisture content and the consistency of transpiration heat field within the zone is conducted, using the following formula:

[0079] ;

[0080] Calculate the characteristic value of the coupling of the water content thermal field, screen the partition units with abnormal coupling degree of water content thermal field in the partition, and establish a spatial distribution map of the cooperative response of the water content thermal field;

[0081] in, Represents the characteristic value of the coupling between water content and thermal field. Representing the Moisture content of the unit Represents the average moisture content of all units. Representing the The temperature value of the unit, Represents the average temperature of all units. This represents the total number of units;

[0082] Based on the response feature set of local maintenance defects in greening, in order to quantify the coupling relationship between moisture distribution and thermal field distribution, features within the same logical partition were selected. Each monitoring unit (e.g.) ), to obtain the moisture content of each unit at the same time. and temperature First, calculate the average moisture content of all units. and average temperature Next, the formula is used: Calculate the coupling eigenvalues ​​of water content and thermal field. The formula is essentially the ratio of the spatial distribution variance of moisture to the spatial distribution variance of temperature. To verify the calculation process, data from five typical units within an anomaly zone were selected for verification. The specific data are shown in Table 1.

[0083] Table 1: Monitoring Data of Moisture Content and Temperature in Zonal Units

[0084]

[0085] Based on the data in Table 1, first calculate the average:

[0086] average moisture content ;

[0087] Average temperature ;

[0088] Substitute into the numerator of the formula (sum of squares of moisture content deviations): ;

[0089] Substituting into the denominator of the formula (sum of squares of temperature deviations): ;

[0090] Calculate the coupling eigenvalues: ;

[0091] The calculation result This indicates that within this partition, the spatial variation in substrate moisture content is extremely large (high variance), while the surface temperature remains relatively uniform (low variance). This suggests that although the moisture distribution is highly uneven, plants, through their own regulatory mechanisms or external environmental influences, have not yet shown significant differences in temperature, or indicates the existence of "latent drought" areas. Based on pre-defined collaborative identification criteria (e.g., (Base interval is [5, 50]), determine the partition. The value far exceeds the upper limit of the benchmark, which is an abnormal state of "severe imbalance of moisture conduction". Based on this, a spatial distribution map of the coordinated response of the moisture content thermal field is established, and the G value of each zone is mapped as a spatial color block.

[0092] S313: Call the spatial distribution map of the coordinated response of the moisture content thermal field, perform spatial clustering analysis on the partitions that exceed the coordinated identification benchmark in the coupled feature value partitions, label the partition codes corresponding to the continuous abnormal areas, and obtain the maintenance defect linkage area group.

[0093] The spatial distribution map of the coordinated response of the moisture content thermal field was invoked. Spatial clustering analysis was performed on partitions where the coupling characteristic values ​​exceeded the coordinated identification benchmark. A density-based noise-applied spatial clustering algorithm was adopted, with a neighborhood radius parameter of 1.5 meters and a minimum threshold of 4 points. The algorithm traversed each grid cell in the spatial distribution map, checking whether there were a sufficient number of similar anomalies in its neighborhood. If the condition was met, they were merged into core points and the cluster boundary was expanded outward. Grid cells that were spatially continuous and whose coupling characteristic values ​​were all in the anomalous range were clustered into independent anomalous clusters. The partition code corresponding to the continuous area covered by each cluster was labeled. The unique identifiers of all grids within the boundary range of each cluster were extracted and combined to construct maintenance defect linkage area groups. After clustering, 8 consecutively numbered grid cells in the left area of ​​the wall were identified as a connected defect group. The scattered single-point anomalous signals were integrated into a regional block with practical engineering significance, filtering out random noise interference, clarifying the physical range that needed centralized intervention, and providing a clear set of spatial objects for subsequent output of targeted regional control instructions.

[0094] Please see Figure 5 The specific steps for obtaining the self-matching maintenance and control strategy set for vertical greening zones are as follows:

[0095] S411: Based on the maintenance defect linkage area group, extract the physiological performance distribution curve of the zone under the specified number, and perform time normalization processing, using the formula:

[0096] ;

[0097] Identify the changes in physiological performance per unit time and obtain the deficiency vector of growth elements in each region;

[0098] in, It represents the change in physiological performance per unit time. Representing the Each partition at time Physiological performance values Representing the Each partition at time Physiological performance values Represents the time step. Represents the total number of partitions;

[0099] Based on the maintenance defect linkage area group, the curve is composed of the time series of chlorophyll fluorescence parameters (Fv / Fm). The curve is time normalized, and the monitoring time of different zones is uniformly mapped to the standard time axis [0, 1]. Then, the formula is used: Identify the change in physiological performance per unit time, i.e., the deficiency vector of growth elements in different zones; assuming that for the aforementioned "maintenance defect linkage area group 1", which includes Three key monitoring points (three representative points are used for simplicity), time step The time frame is set to 1 hour (i.e., 1 unit of time). The monitored physiological performance values ​​(Fv / Fm, after normalization) are at time [time value missing]. and time The data is shown in Table 2;

[0100] Table 2: Monitoring Data of Regional Physiological Performance Changes

[0101]

[0102] Based on the data in Table 2, calculate the molecule (total change): ;

[0103] Substitute into the formula to calculate : ;

[0104] The calculation result This represents the rate of physiological function decline in this region per unit of time. The value was compared with a preset health threshold (e.g., 0.10), and it was found that... The significantly elevated values ​​indicate that plants in this region are experiencing rapid physiological stress. Value constructs a partition growth element deficiency vector ,in This represents the water deficit. The temperature stress is used as input for subsequent decision-making models.

[0105] S412: Input the deficiency vector of regional growth elements into the multi-objective maintenance decision optimization model, combine the regional microclimate characteristics, and invert the calculation of the matrix replenishment flux and environmental compensation parameters required to reach the health restoration threshold, forming a set of regional maintenance and restoration elements.

[0106] The deficiency vector of zonal growth elements is input into a multi-objective maintenance decision optimization model. This model is built on a non-dominated sorting genetic algorithm framework and sets two conflicting optimization objective functions: minimizing the time required for plants to recover to a healthy state and minimizing the total consumption of water and fertilizer resources. The input layer of the model receives the deficiency vector and current microclimate characteristic data such as light and wind speed. It performs feature transformation and inference through a three-layer fully connected neural network with 64, 32, and 16 nodes. The inter-layer activation function uses linear rectified units to enhance nonlinear expression. The model combines the current environmental transpiration potential parameters to inversely calculate the substrate replenishment flux and environmental compensation parameters required to reach the preset health recovery threshold. The output results clearly define the specific irrigation flow requirements (e.g., 2.5 liters per minute) and environmental adjustment measures (e.g., turning on the misting spray to reduce leaf surface temperature by 2 degrees Celsius), forming a set of zonal maintenance and restoration elements. Artificial intelligence algorithms are used to solve the problem of optimizing maintenance parameters under multivariate coupling, ensuring that the given restoration plan can both quickly alleviate plant stress and comply with the principle of resource conservation.

[0107] S413: Based on the set of maintenance and repair elements in the zoning, the actuators in the vertical greening structure are spatiotemporally coordinated and arranged. The spatial coordinates of the zoning are bound to the actuator IDs to obtain control instructions including start time, duration and flow intensity. The set of self-matching maintenance and control strategies for vertical greening zoning is output.

[0108] Based on the set of maintenance and repair elements for each zone, the system reads the equipment status lists of all solenoid valves, booster pumps, and atomizing nozzles, establishing a mapping relationship between the spatial coordinates of each zone and the unique hardware identifiers of the actuators. Based on the flow demand and environmental parameters in the set of repair elements, the system uses hydraulic formulas to calculate the opening duration and sequence of each valve. A staggered opening strategy is employed to prevent pressure drops in the main pipeline caused by simultaneous operation of multiple valves. A control command sequence containing precise opening times, durations, and flow intensity duty cycles is generated. This sequence is packaged and output as a self-matching maintenance and control strategy set for the vertical greening zone. This strategy set uses a standard industrial control protocol format and can be directly transmitted to a programmable logic controller (PLC) for execution. This achieves the transformation from algorithmic decision-making to hardware action, ensuring that the theoretically calculated optimal maintenance scheme can be executed with millisecond-level precision at the physical level, completing a crucial link in the closed-loop control of maintenance.

[0109] Please see Figure 6 The specific steps for obtaining the list of zoning numbers for vertical greening maintenance adjustments are as follows:

[0110] S511: Call the self-matching maintenance and control strategy set of vertical greening zones, extract the zone number in the vertical greening functional layout map, map the zone risk level value with the regional spatial coordinate boundary, identify the zone information corresponding to the plant protection level, and generate a risk distribution map of greening zones.

[0111] The system invokes a set of self-matching maintenance and control strategies for vertical greening zones, extracting the zone numbers within the functional layout map. Using Geographic Information System (GIS) technology, it maps the risk level values ​​of the zones, calculated based on physiological deficit, to the regional spatial coordinate boundaries. The risk level values ​​are determined based on the aforementioned physiological changes and coupling characteristic values, categorized into four levels, with level four representing the highest risk. The system identifies the zones corresponding to plant protection levels, marking areas for rare species as Level 1 protection and common herbaceous areas as Level 3 protection. The risk level layer and protection level layer are overlaid to generate a risk distribution map of the greening zones. In this map, red patches indicate high-risk, high-protection-level emergency treatment areas, while yellow patches indicate medium-risk areas. This transforms abstract numerical risk levels into an intuitive spatial visualization map, providing maintenance managers with a comprehensive risk situation awareness capability.

[0112] S512: Based on the risk distribution map of greening zones, extract the maintenance execution unit number and current maintenance level, match the zone risk level with the execution unit response level, identify unit numbers with insufficient or redundant maintenance coverage, and obtain a list of mismatched maintenance risks in greening zones.

[0113] Based on the risk distribution map of greening zones, the system extracts the maintenance execution unit numbers and ongoing maintenance levels associated with each zone. It then traverses the existing maintenance plan database to obtain the inspection frequency, irrigation cycle, and fertilization intensity set for each execution unit. A comparison matrix of risk levels and execution levels is constructed, comparing the actual needs of each zone with its current configuration. If a zone's risk level is Level 4 (extremely high risk), but the current execution unit's maintenance level is only Level 2 (routine inspection), it is determined to have insufficient maintenance coverage. Conversely, if a zone has extremely low risk but is configured with high-frequency, meticulous maintenance, it is determined to have redundant maintenance resources. All unit numbers and their deviation types that indicate supply-demand mismatch are identified, generating a greening zone maintenance risk mismatch list. This list details all nodes requiring adjustment, their current status, and the degree of deviation, accurately pinpointing weak links and resource waste points in maintenance management, providing precise targets for subsequent dynamic adjustments.

[0114] S513: Based on the list of mismatched maintenance risks in greening zones, retrieve the level number in the priority sequence of plant protection, extract the key zone number that needs to be improved in terms of maintenance coverage, output the adjustment control parameters linked with the original maintenance execution unit in sequence, and output the list of adjustment zone numbers for vertical greening maintenance.

[0115] Based on the list of mismatches in the maintenance risks of greening zones, the priority order of adjustment operations is established. Key zones with high protection levels and insufficient maintenance coverage are prioritized for adjustment. Based on the replenishment flux and environmental compensation parameters calculated in the previous steps, adjustment control parameters linked to the original maintenance execution units are generated sequentially. The original timed irrigation mode is modified to responsive pulse irrigation, or the frequency of atomized cooling is increased. The adjusted parameter set is packaged and output as a list of vertical greening maintenance adjustment zones. This list clearly specifies the adjustment actions, target parameters, and execution priorities for each zone. Through this dynamic adjustment mechanism, maintenance resources are allocated in real time according to plant needs and risk status. Applying this adjustment list significantly improves the survival adaptability of vertical greening plants under extreme summer temperatures, while optimizing water resource utilization efficiency, verifying the effectiveness of this closed-loop control system.

[0116] Table 3: List of Zone Numbers for Vertical Greening Maintenance Adjustment

[0117]

[0118] As shown in Table 3, the list details the specific adjustment measures for different risk level zones, ensuring the accuracy and effectiveness of maintenance activities.

[0119] The intelligent maintenance system for vertical greening is used to implement the above-mentioned intelligent maintenance methods for vertical greening. The system includes:

[0120] The environmental monitoring module divides the detection zones based on the distribution of plant physiological characteristics on the surface of the three-dimensional greening structure and the conductivity of the root matrix. It compares the water content fluctuation range and the environmental response mutation point within the same time period, screens out the synchronous abnormal areas of the two, extracts the zone number and spatial coordinates, summarizes the abnormal time period and zone number, and generates a set of abnormal attributes of greening zone maintenance.

[0121] The zoning and positioning module, based on the set of abnormal attributes of greening zoning maintenance, identifies the consistency between the direction of maintenance demand gradient and the microclimate response vector, marks the boundary number of the zoning, matches the overall structure map of three-dimensional greening, extracts the zoning number range of the abnormal growth gradient area, and establishes a cross-zone maintenance connectivity data group.

[0122] The linkage analysis module, based on the cross-regional maintenance connectivity data group, retrieves the matrix moisture content distribution matrix and the local evaporation heat field characteristics in the region's continuous data sequence, judges the consistency between the moisture content anomaly boundary connectivity and the heat field, marks the partition number that meets the linkage threshold of both, and outputs the maintenance defect linkage area group.

[0123] The scheduling and early warning module analyzes the distribution trend of physiological performance of the corresponding partition and the degree of deviation of the original health status curve based on the partition number of the maintenance defect linkage area group, extracts the partition number of the deviation trend, completes the level identification according to the risk classification standard, and generates a set of self-matching maintenance control strategies for vertical greening partitions.

[0124] The strategy optimization module, based on the self-matching maintenance and control strategy set of vertical greening zones, finds the corresponding position number of the risk level zone in the vertical greening functional layout diagram, retrieves the current maintenance execution unit configuration list, compares the matching situation between plant protection priority and the current maintenance level, filters the zones that need to be updated, and outputs a list of vertical greening maintenance adjustment zone numbers.

[0125] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A smart maintenance method for vertical greening, characterized in that, Includes the following steps: S1: Based on the distribution of plant physiological characteristics on the surface of the three-dimensional greening structure and the conductivity of the root matrix, the detection zones are divided, the water content trajectory and response curve of the zones are extracted, the synchronous abnormal areas of physiological fluctuations and environmental changes are identified, the numbers and coordinates of the synchronous abnormal areas are extracted, and a set of abnormal maintenance attributes of the greening zones is generated. S2: Based on the set of abnormal attributes of greening zone maintenance, identify the demand gradient and microclimate vector of the zone, analyze the cosine similarity of the angle between the demand gradient vector and the microclimate vector, filter the abnormal connectivity zones in combination with connectivity, and form a cross-zone maintenance connectivity data group based on the abnormal connectivity zones. S3: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporation heat field characteristics within the partition, analyze the uniformity of moisture content and the consistency of heat field, screen and match maintenance gap areas, and obtain maintenance defect linkage area groups. S4: Based on the maintenance defect linkage area group, quantitatively analyze the parameter differences between the physiological performance of the plants in the partition and the preset standard growth benchmark. Based on the degree of deviation, match the preset maintenance compensation model, calculate the water and fertilizer regulation parameters and environmental intervention intensity of the partition, and output the self-matching maintenance regulation strategy set of the three-dimensional greening partition.

2. The intelligent maintenance method for vertical greening according to claim 1, characterized in that, The set of abnormal attributes for greening zone maintenance includes physiological mutation zone number, abnormal environmental response points, zone coordinate markers, and time series mutation identifiers. The cross-zone maintenance connectivity data set includes growth gradient abnormal zone identifiers, zone boundary connectivity units, and zone boundary consistency blocks. The maintenance defect linkage area group includes continuous zones with abnormal water content, irregular transpiration heat field zones, overlapping zones with matrix performance mutations, and linkage abnormal zone numbers. The set of self-matching maintenance control strategies for three-dimensional greening zones includes risk level labels, zone response deviation values, local physiological abnormality indicators, and health status deviation levels.

3. The intelligent maintenance method for vertical greening according to claim 1, characterized in that, The specific steps for obtaining the set of abnormal attributes for greening zone maintenance are as follows: S111: Based on the distribution of plant physiological characteristics on the surface of three-dimensional greening structure and the conductivity of root matrix, detection zones are divided. The water content change trajectory and environmental response curve of the zone are extracted by a multi-dimensional sensor array preset on the three-dimensional greening support. The two types of data in the same zone are dimensionless normalized, the sequence deviation of the normalized data is calculated, and the trend value of the difference between physiological and environmental response is obtained. S112: Based on the trend value of the difference between physiological and environmental responses, identify the fluctuation range in the moisture content change curve and the abrupt change point of the environmental response trajectory, perform time alignment processing on the two types of values, extract the time interval where the fluctuation exceeds the preset benchmark value and the abrupt change point is within the set range, and generate a set of high-frequency abnormal interval time periods. S113: For the set of high-frequency abnormal interval time periods, retrieve the corresponding partition number and three-dimensional spatial coordinate information, extract the partition location where the abnormal signal occurred, and generate a set of abnormal attributes for greening partition maintenance.

4. The intelligent maintenance method for vertical greening according to claim 3, characterized in that, The specific steps for obtaining the cross-regional maintenance connectivity data set are as follows: S211: Identify the maintenance demand gradient direction and microclimate response vector of the greening zone in the set of abnormal maintenance attributes of the greening zone, extract the projection trajectory of the two at the boundary of the zone, identify the distribution quantity and aggregation degree of the boundary point in the spatial topology of the zone, and obtain the boundary consistency data matrix of the zone. S212: Based on the partition boundary consistency data matrix, filter the boundary areas with a aggregation degree higher than the average level, compare the spatial boundaries of the overall structure map of the three-dimensional greening, identify the boundary clusters that have physical continuity and belong to the same detection partition, and obtain the abnormal growth gradient zoning within the greening partition. S213: Based on the abnormal growth gradient zoning within the greening zone, an integrated analysis is performed on the consistency of zone boundaries, the dispersion of demand gradient, the uniformity of substrate moisture content, and the delay of physiological response. Based on the response blocks in the zone, a cross-zone maintenance connectivity data set is formed.

5. The intelligent maintenance method for vertical greening according to claim 4, characterized in that, The specific steps for obtaining the maintenance defect linkage area group are as follows: S311: Based on the cross-regional maintenance connectivity data set, extract the matrix moisture content distribution matrix and local evaporative heat field characteristics of the numbered partitions within the partition, perform timestamp alignment processing on the heterogeneous data within the partition, identify the moisture content fluctuation value and the heat field consistency offset, and obtain the greening local maintenance defect response feature set. S312: Based on the greening local maintenance defect response feature set, the uniformity of moisture content and the consistency of transpiration heat field within the partition are jointly analyzed. The moisture content value of each unit in the feature set and the average moisture content of all units are extracted. The square term of the difference between the two is calculated and summed by traversing all units to construct a spatial fluctuation term of moisture content that represents the overall uniformity of moisture. The temperature value of the unit at the corresponding position and the average temperature of all units are extracted simultaneously. The spatial fluctuation term of heat field that represents the consistency of heat field is generated by using the same difference square summation logic. The relative ratio of the spatial fluctuation term of moisture content is calculated based on the spatial fluctuation term of heat field. The coupling feature value of moisture content heat field that represents the synergistic evolution law of moisture and heat distribution is derived. Based on the extracted coupling feature value of moisture content heat field, partition units with abnormal coupling degree of moisture content heat field in the partition are screened, and a spatial distribution map of the synergistic response of moisture content heat field is established. S313: Call the spatial distribution map of the coordinated response of the moisture content thermal field, perform spatial clustering analysis on the partitions in the coupled feature value partitions that exceed the coordinated identification benchmark, label the partition codes corresponding to the continuous abnormal areas, and obtain the maintenance defect linkage area group.

6. The intelligent maintenance method for vertical greening according to claim 5, characterized in that, The specific steps for obtaining the self-matching maintenance and control strategy set for vertical greening zones are as follows: S411: Based on the maintenance defect linkage area group, extract the physiological performance distribution curve of the partition under the specified number, perform time normalization processing, identify the physiological performance change per unit time, and obtain the partition growth element deficiency vector. S412: Input the deficit vector of the growth elements in the region into the multi-objective maintenance decision optimization model, and combine the microclimate characteristics of the region to calculate the matrix replenishment flux and environmental compensation parameters required to reach the health restoration threshold, thus forming a set of maintenance and restoration elements in the region. S413: Based on the set of maintenance and repair elements in the zoning, the actuators in the three-dimensional greening structure are spatiotemporally coordinated and arranged, the spatial coordinates of the zoning are bound to the actuator ID, and control instructions including start time, duration and flow intensity are obtained, and the set of self-matching maintenance and control strategies for the three-dimensional greening zoning is output.

7. The intelligent maintenance method for vertical greening according to claim 1, characterized in that, The method also includes step S5: S5: Call the self-matching maintenance and control strategy set of the vertical greening zone, identify the corresponding number of the zone in the vertical greening functional layout diagram, retrieve the maintenance execution unit list, compare the risk level with the plant protection priority sequence, filter the zone numbers that need to adjust the maintenance coverage, and output the list of vertical greening maintenance adjustment zone numbers. The plant protection priority sequence refers to the ranking based on the rarity of the plant, its criticality in landscape function, and its growth vulnerability. The list of adjustment zones for vertical greening maintenance includes the adjustment target zone number, maintenance intensity adjustment parameters, protection priority comparison items, and linkage maintenance trigger types.

8. The intelligent maintenance method for vertical greening according to claim 7, characterized in that, The specific steps for obtaining the list of zoning numbers for vertical greening maintenance adjustments are as follows: S511: Call the self-matching maintenance and control strategy set of the three-dimensional greening zone, extract the zone number in the three-dimensional greening functional layout map, map the zone risk level value with the regional spatial coordinate boundary, identify the zone information corresponding to the plant protection level, and generate a greening zone risk distribution map. S512: Based on the risk distribution map of the greening zone, extract the maintenance execution unit number and the current maintenance level, match the zone risk level with the execution unit response level, identify the unit number with insufficient or redundant maintenance coverage, and obtain a list of mismatched maintenance risks in the greening zone. S513: Based on the greening zoning maintenance risk mismatch list, retrieve the level number in the plant protection priority sequence, extract the key zoning number that needs to be improved in terms of maintenance coverage, output the adjustment control parameters linked with the original maintenance execution unit in sequence, and output the list of three-dimensional greening maintenance adjustment zoning number.

9. An intelligent maintenance system for vertical greening, characterized in that, The system is used to implement the intelligent maintenance method for vertical greening as described in any one of claims 1-8, and the system includes: The environmental monitoring module divides the detection zones based on the distribution of plant physiological characteristics on the surface of the three-dimensional greening structure and the conductivity of the root matrix. It compares the water content fluctuation range and the environmental response mutation point within the same time period, screens out the synchronous abnormal areas of the two, extracts the zone number and spatial coordinates, summarizes the abnormal time period and zone number, and generates a set of abnormal attributes of greening zone maintenance. The zoning positioning module, based on the set of abnormal attributes of greening zoning maintenance, identifies the consistency between the direction of maintenance demand gradient and the microclimate response vector, marks the boundary number of the zoning, matches the overall structure map of three-dimensional greening, extracts the zoning number range of the abnormal growth gradient area, and establishes a cross-zone maintenance connectivity data group. The linkage analysis module, based on the cross-regional maintenance connectivity data group, retrieves the matrix moisture content distribution matrix and the local evaporation heat field characteristics in the region as continuous data sequences, judges the consistency between the moisture content anomaly boundary connectivity and the heat field, marks the partition number that meets the linkage threshold of both, and outputs the maintenance defect linkage area group. The scheduling and early warning module analyzes the distribution trend of physiological performance of the corresponding partition and the degree of deviation of the original health status curve based on the partition number of the maintenance defect linkage area group, extracts the partition number of the deviation trend, completes the level identification according to the risk classification standard, and generates a set of self-matching maintenance control strategies for the three-dimensional greening partition. The strategy optimization module, based on the self-matching maintenance and control strategy set of the vertical greening zones, finds the corresponding position number of the risk level zone in the vertical greening functional layout diagram, retrieves the current maintenance execution unit configuration list, compares the matching situation between plant protection priority and the current maintenance level, filters the zones that need to be updated, and outputs a list of vertical greening maintenance adjustment zone numbers.

Images