Intelligent maintenance and management method and system for garden plant growth based on multi-source perception

Abstract

The application discloses a kind of based on multi-source perception's garden plant growth intelligent maintenance management method and system, belong to environmental perception data processing technical field, comprising: diurnal period acquisition light radiation and leaf air temperature difference data, generate light heat lag feature data and accumulate as diurnal period cumulative data, when exceeding first threshold, output first intervention instruction, again by recovery slope adjustment time window;Night acquisition spectral ratio and feel night angle data, after attitude correction and trap wave suppression synthesis photosensitive imbalance feature data, when exceeding second threshold, output second intervention instruction. With quantifiable dynamic characteristic, the lag and recovery response of plant to light and heat change are described, the transient disturbance of building shadow and the flicker of artificial lighting power frequency are inhibited, diurnal period cumulative data corrects night attitude, night recovery slope back to first threshold, form cross diurnal self-adaptive closed loop, improve the accuracy and reliability of maintenance under complex environment.

Description

Technical Field

[0001] This invention relates to the field of environmental sensing data processing technology, and in particular to a method and system for intelligent maintenance and management of garden plant growth based on multi-source sensing. Background Technology

[0002] As urban landscaping extends to high-density commercial areas, building facades, and rooftops, the microenvironment of garden plants exhibits drastic and unnatural changes in light and heat conditions during both day and night. During the day, the shadows cast by high-rise buildings and glass curtain walls frequently sweep across the leaves, accompanied by reflections from the walls, causing the leaves to experience transient changes in light intensity and heat shock. At night, continuous artificial lighting from building facades, advertising light boxes, and streetlights disrupts the natural photoperiod upon which plants depend. Intelligent maintenance of garden plant growth under such conditions places corresponding demands on perception and regulation.

[0003] Currently, there are multiple ways to perceive and maintain the growth of garden plants, including methods for monitoring and identifying the growth status of plants and methods for adjusting the light environment of plants.

[0004] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0005] Plants exhibit a lag in their response to changes in light and heat. Leaf temperature cannot immediately follow changes in light intensity due to the obstruction of physical processes such as stomatal regulation and transpiration. Furthermore, the recovery of the plant's condition after intervention is a process that changes over time. Existing methods lack quantifiable representations and utilization of this dynamic response, making it difficult to reflect the accumulating heat stress and the plant's real-time recovery capacity. The plant's state changes with the environment and its own recovery status. However, the judgment criteria used to trigger maintenance actions in existing methods are mostly pre-set and relatively fixed, lacking a mechanism to adjust the judgment criteria across periods based on the plant's real-time recovery performance, making it difficult to adjust as the plant's state evolves. In open urban environments, daytime building shadows and curtain wall reflections introduce transient interference that jumps and falls rapidly in light intensity data, while nighttime artificial lighting introduces periodic power frequency flicker in spectral data, making the judgment criteria susceptible to environmental interference. Summary of the Invention

[0006] To address the technical problems existing in the prior art, embodiments of the present invention provide a method and system for intelligent maintenance and management of garden plant growth based on multi-source sensing. The technical solution is as follows:

[0007] Acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle. Analyze and calculate the light radiation data and leaf-temperature temperature difference data within a preset first time window to generate photothermal hysteresis characteristic data, and accumulate them during the daytime operation cycle to obtain daytime cumulative data.

[0008] When the photothermal hysteresis characteristic data is greater than the preset first threshold, the first physical intervention command is generated and output.

[0009] The temporal change trajectory of the leaf-temperature temperature difference data in the first time window after the execution of the first physical intervention command is obtained and the first recovery slope data is calculated. The length of the first time window of the next daytime operation cycle is adjusted using the first recovery slope data.

[0010] Data on the red-to-red spectral ratio and leaf night-sensing tilt angle of the target plant during the nighttime operating cycle were obtained. Based on the cumulative data during the daytime, the spatial bias of the leaf night-sensing tilt angle data was corrected to obtain the daytime corrected tilt angle data.

[0011] The red-far-red spectral ratio data is analyzed within a preset second time window to obtain spectral integral data, and the spectral integral data is calculated with the daytime correction tilt angle data to obtain photosensitivity imbalance characteristic data.

[0012] When the photosensitive imbalance characteristic data exceeds the preset second threshold, a second physical intervention command is generated and output.

[0013] The temporal change trajectory of the blade night-sensing tilt angle data in the second time window after the execution of the second physical intervention command is obtained, and the second recovery slope data is calculated. The value of the first threshold of the next daytime operation cycle is updated according to the preset mapping function using the second recovery slope data.

[0014] Intelligent maintenance and management methods for garden plant growth based on multi-source sensing include:

[0015] In addition, a smart maintenance management system for garden plant growth based on multi-source sensing is also provided, specifically including:

[0016] The daytime feature extraction module is used to acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle. It analyzes and calculates the light radiation data and leaf-temperature temperature difference data within a preset first time window to generate photothermal hysteresis feature data, and accumulates it during the daytime operation cycle to obtain daytime cumulative data.

[0017] The daytime intervention triggering module is used to generate and output the first physical intervention command when the photothermal hysteresis characteristic data is greater than the preset first threshold.

[0018] The daytime feedback module is used to acquire the temporal change trajectory of the leaf-temperature difference data in the first time window after the execution of the first physical intervention command and to calculate the first recovery slope data. The first recovery slope data is used to adjust the length of the first time window of the next daytime operation cycle.

[0019] The attitude correction module is used to acquire the red-to-red spectral ratio data and leaf night-sensing tilt angle data of the target plant during the nighttime operating cycle. Based on the daytime cumulative data, the spatial deviation correction of the leaf night-sensing tilt angle data is performed to obtain the daytime corrected tilt angle data.

[0020] The nighttime feature synthesis module is used to analyze the red-far-red spectral ratio data within a preset second time window to obtain spectral integral data, and calculate the photosensitivity imbalance feature data by combining the spectral integral data with the daytime correction tilt angle data.

[0021] The nighttime intervention trigger module is used to generate and output a second physical intervention command when the photosensitivity imbalance characteristic data is greater than a preset second threshold.

[0022] The nighttime feedback module is used to acquire the temporal change trajectory of the blade nighttime tilt angle data in the second time window after the execution of the second physical intervention command, calculate the second recovery slope data, and use the second recovery slope data to update the value of the first threshold of the next daytime operating cycle according to the preset mapping function.

[0023] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0024] 1. The intelligent maintenance and management method for garden plant growth based on multi-source sensing provided by this invention overcomes the limitation that instantaneous environmental data alone cannot reflect the real-time state of plants by characterizing the rapid physiological response of plants to changes in light and heat with quantifiable dynamic features. Plants exhibit a lag in their response to changes in light and heat, and the recovery of their physical state after maintenance intervention is a process that changes over time. Transforming this lag and recovery rate into directly usable feature data for judgment and control allows maintenance decisions to be based on the plant's true dynamic physiological state, rather than isolated environmental readings at a particular moment, thereby significantly improving the accuracy and timeliness of growth sensing.

[0025] 2. This invention uses the recovery observed by the plant after intervention as a basis to make cross-cycle adjustments to the judgment criteria for subsequent cycles. In this process, unreliable and invalid responses are eliminated, and fluctuations in the regulatory process are suppressed, making adaptive regulation both sensitive and stable. Therefore, the maintenance system can continuously self-adjust as the plant's condition evolves, ensuring that the regulation remains consistent with the plant's actual condition over the long term.

[0026] 3. This invention can distinguish and suppress transient interferences introduced by daytime building shadows and curtain wall reflections, as well as periodic interferences introduced by nighttime artificial lighting, from real environmental changes. At the same time, it preserves real environmental information to the maximum extent, so that the characteristic data used for judgment truly reflects the light environment and physiological state of plants, thereby improving the reliability of the system in complex urban environments.

[0027] 4. This invention overcomes the shortcomings of independent day-night regulation, which is difficult to adapt to the continuous evolution of plant states, by establishing cross-cycle state transmission and coupling between day and night. Plant stress tolerance and physiological state are continuous between day and night. Based on this, the state accumulated during the day is transmitted to the night for judgment, and the night's recovery performance is used to adjust the daytime judgment benchmark. This makes daytime treatment and nighttime treatment no longer isolated from each other, but rather interconnected and mutually corrected through data, forming a continuously adaptive day-night closed loop that adapts to environmental changes. This makes the overall maintenance management more in line with the actual needs of plants under continuous day-night rhythms. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of the intelligent maintenance and management method for garden plant growth based on multi-source sensing provided in the embodiments of this application.

[0030] Figure 2 A schematic diagram of a multi-source sensing-based intelligent maintenance management system module for garden plant growth provided in this application embodiment.

[0031] Figure 3 This is a schematic diagram of the intelligent maintenance and management logic flow of garden plants based on multi-source perception, provided in an embodiment of this application. Detailed Implementation

[0032] Embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present disclosure are shown in the drawings, it should be understood that embodiments of the present disclosure may be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure.

[0033] It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure. In the description of the embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "this embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects.

[0034] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0035] A smart maintenance and management method for garden plant growth based on multi-source sensing specifically includes:

[0036] like Figure 3 The diagram illustrates the intelligent maintenance and management logic of garden plants based on multi-source sensing in this embodiment, describing a closed-loop maintenance process with two links during the day and night, coupled through three feedback loops. The daytime link includes: collecting light radiation data and leaf-temperature temperature difference data to generate photothermal hysteresis characteristic data, which is then accumulated during the daytime as daytime accumulated data; subsequently, determining whether the photothermal hysteresis characteristic data exceeds a first threshold; if so, outputting a first physical intervention command; after intervention, the system calculates a first recovery slope data and adjusts the length of the first time window accordingly.

[0037] The nighttime link includes: collecting red-far-red spectral ratio data and leaf night-sensing tilt angle data; first, correcting the attitude space deviation of the night-sensing tilt angle; then synthesizing photosensitivity imbalance characteristic data; subsequently, determining whether the photosensitivity imbalance characteristic data is greater than a second threshold; if so, outputting a second physical intervention command; after intervention, the system calculates the second recovery slope data.

[0038] Combination Figure 1 The method flow shown provides a detailed explanation of each step in this embodiment.

[0039] Acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle. Analyze and calculate the light radiation data and leaf-temperature temperature difference data within a preset first time window to generate photothermal hysteresis characteristic data, and accumulate them during the daytime operation cycle to obtain daytime cumulative data.

[0040] Generate photothermal hysteresis feature data, specifically including:

[0041] Photothermal hysteresis characteristic data is used to characterize the degree of lag in leaf temperature change when incident light intensity changes abruptly: In high-density commercial areas of cities, the frequent alternation of building shadows causes the leaves to experience transient light intensity changes. However, due to the obstruction of physical processes such as stomatal regulation and transpiration heat dissipation, the leaf temperature cannot keep up with the light intensity change in real time. The severity of this hysteresis reflects whether there is mechanical blockage in the heat dissipation channels of the leaf surface.

[0042] The derivatives of the light radiation data and the leaf temperature difference data were calculated to obtain the rate of change of light radiation and the rate of change of temperature difference, which respectively reflect how fast light intensity and leaf temperature change over time.

[0043] Light radiation data refers to the time series that characterizes the intensity of incident light, collected by light radiation sensors placed at the location of the blade. Leaf-air temperature difference data refers to the time series composed of the difference between the surface temperature of the blade and the surrounding air temperature collected simultaneously. It characterizes the instantaneous thermodynamic state of the blade during the heat exchange process with the environment.

[0044] In real-world scenarios, reflections from building glass curtain walls and momentary occlusion by pedestrians and vehicles can introduce false abrupt changes in light radiation data, causing rapid jumps and falls. These false abrupt changes are physically distinguishable from real light changes: real light changes remain at a new level after the abrupt change, while false abrupt changes quickly fall back to the original level after a brief deviation.

[0045] Monitor optical radiation data, and record the moment when the absolute value of the change in optical radiation data relative to the previous moment exceeds the preset jump variable as the jump start moment.

[0046] The preset jump variable is derived from the historical solar radiation data collected during the pre-operation period of the target site when the system platform is built. Solar radiation data of the site is collected first during a period of typical sunny days. In this historical data, the edge changes caused by building shadows and curtain wall reflections, and the changes caused by sensor noise and gradual changes in natural light, are at different levels in the statistical distribution of single-step changes. There is a range of changes between the two that do not overlap with either one, and the preset jump variable is set within this range.

[0047] From the start of the transition, the duration for which the statistical optical radiation data remains deviated from the baseline value before the start of the transition is recorded as the end of the transition when the optical radiation data falls back to within a preset neighborhood of the baseline value.

[0048] The jump start time refers to the moment when the absolute value of the single-step change in optical radiation data relative to the previous moment exceeds a preset jump variable, thus being determined as the start of a jump. The optical radiation baseline value refers to the representative level of optical radiation data within a short stable period before the jump start time, used to characterize the steady-state light intensity before the jump and as a reference for judging whether the light intensity subsequently returns to normal. The optical radiation baseline value is determined in real-time during each jump identification process by averaging the optical radiation data within a preset backtracking time before the jump start time, using the average to smooth out random noise within that stable period. The duration refers to the time elapsed from the jump start time until the optical radiation data returns to near the baseline value, characterizing the degree of deviation from steady state during this jump.

[0049] When the duration between the start and end of the transition is less than the preset duration, the data segment between the start and end of the transition is determined to be a pseudo-transition, and the rate of change of light radiation and the rate of change of temperature difference at the corresponding time in the data segment are removed.

[0050] False mutations refer to sudden changes in light intensity caused by external factors such as building shadows, curtain wall reflections, and momentary obstruction by pedestrians and vehicles. These mutations manifest as a brief deviation of light radiation data from the steady state followed by a rapid decline, and do not reflect the true and continuous changes in the light environment in which the plant is located.

[0051] The product of the rate of change of light radiation after removal and the rate of change of temperature difference is integrated within the first time window to obtain the photothermal hysteresis characteristic data.

[0052] The length of the first time window needs to match the thermal response time constant of the target plant itself. During the calibration phase before the system is put into operation, the leaves can be subjected to a step-by-step shading and unshading under typical sunny and hot conditions. The time taken for the leaf temperature difference to stabilize again after being disturbed is recorded, and 1.5 to 3 times this time is used as the initial value of the length of the first time window. Subsequently, this length will be adjusted online by the first recovery slope data so that the time scale of feature extraction always fits the real-time state of the plant.

[0053] Since the strength of photothermal hysteresis depends on whether changes in light intensity and leaf temperature are synchronous and in the same direction, the system does not consider either rate in isolation. Instead, it integrates the product of the rate of change in light radiation and the rate of change in air temperature difference within the first time window, using this integral result as the photothermal hysteresis characteristic data. The introduction of the product ensures that the characteristic values ​​accumulate significantly only when light intensity and leaf temperature change significantly simultaneously, thus accurately capturing the true thermal hysteresis coupling, rather than the isolated fluctuations of any single data stream.

[0054] The percentage of eliminated moments within the first time window is calculated relative to the total number of moments. When the percentage of eliminated moments exceeds the preset upper limit, the length of the first time window is extended proportionally to the percentage of eliminated moments, and then the calculation is repeated.

[0055] Accumulated daytime data is obtained, specifically including:

[0056] The daytime operating cycle is divided into a preset number of sub-periods, and the cumulative value of the photothermal hysteresis characteristic data is calculated for each sub-period.

[0057] A sub-period refers to several consecutive time segments obtained by equally dividing a complete daytime cycle along the time axis. The cumulative value of a sub-period refers to the cumulative amount of photothermal lag characteristic data within a certain sub-period, characterizing the intensity of heat stress experienced by the plant during that sub-period. The daytime cumulative data refers to a single scalar obtained by limiting and weighting the cumulative values ​​of each sub-period, characterizing the total amount of heat stress accumulated on that day, and is the only state carrier transmitted from day to night.

[0058] The preset number, i.e. the number of sub-time periods divided into daytime periods, is determined by the total duration of daytime and the system sampling rate.

[0059] The cumulative values ​​for each time period are compared with the preset feature upper limit, and the cumulative values ​​for time periods that exceed the feature upper limit are limited to the feature upper limit.

[0060] The preset feature upper limit is determined by statistically analyzing the historical cumulative values ​​of healthy plants collected during the pre-operation period of the target site when the system platform is built. During platform construction, the cumulative values ​​of healthy plants for each sub-period across multiple historical sunny and hot days are collected to form their normal distribution. Cumulative values ​​exceeding the upper limit of this normal distribution are more likely due to sensor anomalies or extreme single-point mutations, rather than actual heat stress experienced by the plant. The feature upper limit is thus set based on the upper end of the normal distribution of the historical cumulative values ​​of the healthy plant.

[0061] The cumulative values ​​for each time period after the amplitude limit are weighted and summed according to their corresponding weights to obtain the daytime cumulative data.

[0062] Since the water content and turgor pressure in plants change continuously and continue until nightfall, the heat stress closer to nightfall has a greater impact on the initial physiological posture of plants after nightfall. Therefore, the weight of each sub-period increases as the sub-period is closer to the end of the day, and the sum of all weights is made uniform to ensure that the dimensions of the cumulative data during the day are consistent with the cumulative values ​​of individual periods.

[0063] When the number of cumulative values ​​in each time period that exceed the upper limit of the feature exceeds the preset number of anomalies, the current day period is determined to be an abnormal day, and the cumulative data of the day period is multiplied by the preset attenuation coefficient and then output.

[0064] If the cumulative value of an individual sub-period exceeds the feature limit, it can be considered an occasional local anomaly, which can be handled by the amplitude limiting itself. However, when the number of sub-periods exceeding the feature limit reaches a considerable proportion, it indicates that the anomaly is no longer occasional and the overall data for that day is questionable. The preset number of anomalies is set at a level that ensures that occasional local exceedances will not trigger an anomaly daytime judgment, while multiple overall exceedances will trigger a judgment. The preset attenuation coefficient is determined based on the typicality of the daytime cumulative data being overestimated under an anomaly daytime, and is taken as a proportion between zero and one.

[0065] When the photothermal hysteresis characteristic data is greater than the preset first threshold, the first physical intervention command is generated and output.

[0066] The first threshold is a criterion used to determine whether a plant is in a state of heat shock retardation. When the photothermal retardation characteristic data exceeds this threshold, the heat retardation experienced by the plant is considered to have reached a level requiring intervention. Heat stress is slowly applied to the target plant, and the critical point at which its photosynthetic rate observably decreases with increasing heat stress is observed. The photothermal retardation characteristic data corresponding to this heat shock critical point are recorded and used as the initial value of the first threshold.

[0067] The first physical intervention command is the control command output by the system when thermal shock inhibition is determined to be established. It is used to initiate physical intervention measures acting on the leaves. In this embodiment, it specifically initiates local micro-spray cooling and short-wavelength supplemental lighting to forcibly reduce the leaf boundary layer resistance and maintain photosynthetic electron transfer.

[0068] Heat shock retardation refers to the physiological state in which leaves, under strong light and heat stress, experience obstructed heat dissipation channels and limited photosynthesis due to impaired stomatal regulation and transpiration.

[0069] The temporal change trajectory of the leaf-temperature temperature difference data in the first time window after the execution of the first physical intervention command is obtained and the first recovery slope data is calculated. The length of the first time window of the next daytime operation cycle is adjusted using the first recovery slope data.

[0070] The leaf temperature difference data within the first time window after the execution of the first physical intervention command is obtained as the first recovery response trajectory.

[0071] The first recovery response trajectory refers to the sequence formed by the change of leaf temperature difference data over time within the selected observation interval after the execution of the first physical intervention command, reflecting the process of plant recovery from heat shock inhibition under the intervention.

[0072] Linear fitting is performed on the first recovery response trajectory to obtain the first recovery slope data and the first goodness of fit characterizing the fit quality. The difference between the first and last values ​​of the first recovery response trajectory is extracted as the first recovery amplitude.

[0073] The first recovery slope data refers to the slope of the straight line obtained by linearly fitting the first recovery response trajectory, characterizing the speed of plant recovery; the larger the absolute value, the faster the recovery. The first goodness of fit is a measure of the degree of goodness of the linear fit, reflecting how closely the recovery trajectory approximates a straight line, used to determine whether the trajectory constitutes a clearly identifiable recovery process. The first recovery amplitude is the difference between the values ​​at the beginning and end of the first recovery response trajectory, characterizing the actual change in leaf-temperature difference during the recovery process.

[0074] When the first goodness of fit is lower than the preset first goodness of fit lower limit, or the first recovery amplitude is less than the preset first recovery amplitude lower limit, the intervention is determined to be an invalid response, and the length of the first time window is not updated based on the first recovery slope data.

[0075] The determination process for the first lower limit of goodness of fit and the first lower limit of recovery amplitude is as follows: During debugging, two types of recovery trajectories are collected: one with actual recovery and the other with actual invalidity (such as weak triggering or noise spikes). The actual recovery trajectory is close to a clear straight line, with a high goodness of fit, and its first-to-last difference significantly exceeds the random fluctuation of the leaf-temperature difference without intervention. The invalid response trajectory is flat or chaotic, with a low goodness of fit, and its first-to-last difference consists only of random fluctuations. The first lower limit of goodness of fit is set between the high and low ends of the goodness of fit for the actual recovery and invalid response, respectively. The first lower limit of recovery amplitude is set at a level that distinguishes the obvious fluctuations of the actual recovery from the random fluctuations without intervention (this level of random fluctuation is given by the statistical dispersion of the leaf-temperature difference under the state of no intervention).

[0076] When the first goodness of fit is not lower than the first goodness of fit lower limit and the first recovery amplitude is not less than the first recovery amplitude lower limit, the intervention is determined to be an effective response, and the length of the first time window of the next day period is adjusted according to the absolute value of the first recovery slope data.

[0077] Adjusting the first time window also includes:

[0078] Record the historical adjustment trajectory within the pre-set statistical number of daytime operation cycles for the length of the first time window.

[0079] Historical adjustment trajectory refers to the value sequence formed by the periodic adjustment of the first time window length over several consecutive days.

[0080] The preset number of statistical cycles is set during the system debugging phase. During debugging, on running segments where window length oscillations or drifts are known to exist, the number of daytime cycles under investigation is gradually increased to observe whether the oscillations and drifts can be stably detected without being accidentally triggered by one or two normal adjustments. The minimum number of cycles that can stably detect both is taken as the preset number of statistical cycles.

[0081] When the number of times the adjustment direction alternates between positive and negative during adjacent daytime operating cycles in the historical adjustment trajectory exceeds the preset number of oscillations, it is determined to be an oscillation, and the preset length range is reduced according to the preset tightening coefficient.

[0082] The preset number of oscillations is set during the system debugging phase. During debugging, the number of directional alternations within the preset statistical period is recorded for both the normal adaptive running segment and the artificially created oscillation running segment. The former has a lower number of alternations, while the latter is close to the period number. The number of alternations between the two is taken as the preset number of oscillations.

[0083] When the historical adjustment trajectory accumulates monotonically in the same direction within several daytime operating cycles of the preset statistical period, and the accumulated amount exceeds the preset drift limit, it is determined to be a drift, and the length of the first time window is reset to the preset reference length.

[0084] The preset tightening coefficient is set during the system debugging phase. During debugging, different ranges of tightening ratios are repeatedly applied to the identified oscillating segments, and the oscillations are observed to subside within several days thereafter. The tightening ratio that subsides the oscillations while still allowing for window adjustment is taken as the preset tightening coefficient.

[0085] The preset drift upper limit is set during the system debugging phase. During debugging, the accumulated amount of the window length is recorded unidirectionally from the reference length until it approaches the boundary of the preset length range. The accumulated amount when the window has not yet reached the boundary and still has room for adjustment is taken as the preset drift upper limit. The preset reference length is taken as the physically optimal initial value of the first time window based on the thermal response time scale.

[0086] Data on the red-to-red spectral ratio and leaf night-sensing tilt angle of the target plant during the nighttime operating cycle were obtained. Based on the cumulative data during the daytime, the spatial bias of the leaf night-sensing tilt angle data was corrected to obtain the daytime corrected tilt angle data.

[0087] The red-to-far-red spectral ratio data refers to the ratio of red light irradiance to far-red light irradiance collected at night, characterizing the light quality conditions that drive conformational changes in phytochromes within plants. Leaf night-sensing tilt angle data refers to the tilt angle of a leaf relative to a horizontal reference, collected at night, characterizing the plant's posture in accordance with its diurnal rhythm.

[0088] Based on the cumulative daytime data, spatial bias correction is performed on the leaf night-sensing tilt angle data to obtain the daytime corrected tilt angle data. The specific process includes:

[0089] Extract the spatial deviation amplitude of the blade night-sensing tilt angle data relative to the preset reference tilt angle.

[0090] The preset reference tilt angle refers to the stable tilt angle maintained by the drooping leaves of a healthy plant in its normal dormant state. It serves as a reference for measuring posture deviation, and the preset reference tilt angle is taken from the measured value of the healthy plant's normal dormant posture at night. The spatial deviation amplitude refers to the degree of deviation of the leaf's perceived nighttime tilt angle data from the preset reference tilt angle, characterizing the magnitude of the plant's current posture deviating from its normal dormant posture.

[0091] Extract the high-frequency jitter component from the blade night-sensing tilt angle data. When the amplitude of the high-frequency jitter component exceeds the preset jitter amplitude, determine that the corresponding data segment is an external force disturbance and remove it from the calculation of the spatial deviation amplitude.

[0092] High-frequency jitter refers to the rapid reciprocating component superimposed on the gradually changing night-sensing tilt angle, usually caused by external disturbances such as wind. External disturbances refer to external factors such as wind that cause the blades to oscillate rapidly, unrelated to their physiological posture. The preset jitter amplitude is determined from the nighttime posture data of the target site during the pre-operation period when the system platform is built. During platform construction, the blade night-sensing tilt angle data is collected on a calm night. At this time, its high-frequency component consists only of sensor noise, while external disturbances such as wind will superimpose a rapid reciprocating jitter that significantly exceeds this noise. The amplitude that can distinguish the high-frequency component on a calm night from wind-induced jitter is taken as the preset jitter amplitude. Data segments with high-frequency jitter components exceeding this amplitude are judged as external disturbances and are discarded.

[0093] The attitude correction offset is obtained by converting the accumulated daytime data using a preset ratio. The attitude correction offset increases as the accumulated daytime data increases.

[0094] The preset ratio is a conversion factor used to convert daytime cumulative data into attitude correction offset, which is determined by regression analysis of historical pairwise data during system platform construction. During platform construction, pairwise data from multiple historical sunny and hot days are collected. Each pair consists of the daytime cumulative data for that day and the initial attitude deviation at nightfall that can be attributed to daytime stress. Regression analysis of the pairwise data yields the preset ratio, ensuring that the attitude correction offset calculated from the daytime cumulative data is numerically equivalent to the measured initial attitude deviation.

[0095] The daytime correction tilt angle data is obtained by subtracting the attitude correction offset from the spatial deviation amplitude.

[0096] When the attitude correction offset is greater than the spatial deviation amplitude, the daytime correction tilt data is set to zero and the preset proportion of the daytime cumulative data participating in the correction of the next nighttime operating cycle is reduced.

[0097] By setting the daytime correction tilt data to zero and adjusting the preset ratio night by night when excessive deduction occurs, it not only prevents the generation of physically meaningless negative values ​​when the correction amount exceeds the amount being corrected, but also allows the conversion relationship to self-correct and converge to an appropriate level during operation.

[0098] The red-far-red spectral ratio data is analyzed within a preset second time window to obtain spectral integral data, and the spectral integral data is calculated with the daytime correction tilt angle data to obtain photosensitivity imbalance characteristic data.

[0099] The red-far-red spectral ratio data is analyzed within a preset second time window to obtain spectral integral data. The specific process includes:

[0100] A discrete Fourier transform is performed on the time series of the far-infrared spectral ratio data within the second time window to obtain the amplitude spectrum with frequency as the independent variable.

[0101] The reason for converting to the frequency domain is that the power frequency flicker of artificial lighting at night is a narrow-band periodic component falling on a specific frequency, while the light quality changes related to plant rhythms are gradual. The two can be distinguished in the frequency domain, so we first convert to the frequency domain and then apply processing to the frequency where the flicker occurs.

[0102] Frequency components within a preset frequency range are selected from the amplitude spectrum as candidate interference components. The preset frequency range is set based on the power frequency and harmonics of artificial lighting.

[0103] Candidate interference components are frequency components that fall within a preset frequency range and are considered to be possibly caused by artificial lighting power frequency flicker. This preset frequency range is set based on the power frequency of artificial lighting and its harmonics: the flicker energy of artificial lighting is concentrated at the power frequency (such as 50 Hz or 60 Hz) and its integer multiples of harmonics. Therefore, a narrow band is set with the power frequency and each harmonic as the center to form the preset frequency range; the width of the narrow band is determined by the measured broadening of the artificial lighting spectrum during the system platform construction, so that it just covers the spectrum without affecting adjacent frequency bands, so that subsequent processing only targets the flicker spectrum and retains the true light quality information falling on other frequencies to the maximum extent.

[0104] For each candidate interference component, extract the amplitude of a preset number of adjacent frequency components that are outside the preset frequency range, and use the statistical value of the amplitude of the preset number of adjacent frequency components as the background amplitude of the candidate interference component.

[0105] The baseline amplitude represents the true signal level that the frequency should have in the absence of flicker. It is estimated using adjacent frequencies outside the preset frequency range because these frequencies are close to the flicker spectral lines but not contaminated by them, and can represent the true signal level near the spectral lines.

[0106] The preset quantity, which is the number of adjacent frequency points selected, is determined during the system debugging phase.

[0107] The difference between the amplitude of the candidate interference component and the background amplitude is calculated as the interference excess. When the interference excess is greater than zero, the interference excess is subtracted from the amplitude of the candidate interference component to bring it back to the background amplitude. When the interference excess is not greater than zero, the amplitude of the candidate interference component is kept unchanged, and the processed amplitude spectrum is obtained.

[0108] The amplitude of candidate interference components is usually not entirely composed of artificial lighting flicker; it also contains components that the real signal should have at that frequency. Suppressing the entire component would damage the real signal as well. Therefore, the baseline amplitude is used to characterize the level that the frequency should have in the absence of flicker, and the difference between the amplitude of the candidate interference component and the baseline amplitude is recorded as the interference excess. When the interference excess is greater than zero, it indicates that the amplitude at that frequency exceeds the true baseline, and the excess is caused by flicker. Therefore, only this part is subtracted to bring the amplitude back to the baseline amplitude, removing the flicker component while preserving the real signal at that frequency. When the interference excess is not greater than zero, it indicates that there is no identifiable flicker component exceeding the baseline at that frequency. Therefore, its amplitude is kept unchanged to avoid unnecessary modification of the real signal. Therefore, the suppression of flicker at the power frequency and its harmonics is an adaptive suppression that only subtracts the part exceeding the background. The amount of suppression is determined by the degree to which each frequency exceeds the background, rather than setting the candidate interference components to zero or attenuating them by a fixed ratio. This minimizes the loss of real light quality information while suppressing flicker. The resulting processed amplitude spectrum is purified at the flicker frequency and remains unchanged at other frequencies.

[0109] The processed amplitude spectrum was subjected to inverse discrete Fourier transform to obtain the suppressed red-far-red spectral ratio data.

[0110] The aforementioned amplitude subtraction is performed in the frequency domain, while subsequent processing requires a time-domain optical quality sequence. Therefore, an inverse discrete Fourier transform is performed on the processed amplitude spectrum to restore the purified spectrum to a time series. During the inverse transform, the phase of the original signal is retained, and only the processed amplitude is replaced with the original amplitude. This ensures that, except for the power frequency flicker component, the other time-domain characteristics of the signal are preserved. The result is a red-to-far-red spectral ratio time series that is synchronized with the original acquisition, of equal length, and has had its power frequency flicker fluctuations eliminated, providing flicker-free input for subsequent integration.

[0111] The suppressed red-to-red spectral ratio data is integrated within the second time window to obtain the spectral integral data.

[0112] Since the data used for integration has had its power frequency scintillation component removed in the aforementioned steps, the resulting spectral integration data reflects the actual cumulative level of nighttime light quality conditions, rather than the distorted result affected by artificial lighting scintillation. This ensures the reliability of the photosensitivity imbalance characteristic data synthesized from it and the nighttime judgment.

[0113] The calculation process for photosensitive imbalance characteristic data is as follows:

[0114] ;

[0115] F represents the photosensitive imbalance characteristic data; The spectral integral data is obtained by integrating the red-far-red spectral ratio data after suppressing power frequency flicker within the second time window, representing the cumulative level of real nighttime light quality conditions during this period; The daytime correction tilt angle data is obtained by subtracting the attitude correction offset from the cumulative daytime data from the spatial deviation amplitude of the leaf nighttime tilt angle, reflecting the attitude deviation that can be attributed to nighttime photoperiod disturbances. For the weights of the spectral integral data, The tilt angle data weights are used to correct for daytime conditions, and the two factors respectively characterize the contributions of light quality conditions and attitude deviations to abnormal residence of phytochrome. This is the reference value for spectral integral data, that is, the baseline level that spectral integral data should have under normal nighttime light quality conditions; This is a reference value for posture deviation, that is, the baseline for the permissible posture deviation when the plant is in a normal dormant posture.

[0116] and All measurements were obtained during system platform setup and calibration. The representative level of the spectral integral data obtained by processing the red-far-red spectral ratio data of the target plant under normal nighttime environment (i.e. no abnormal artificial lighting and the plant is in normal rhythm) using the same second time window and the same notch suppression procedure is used as a reference for normal light quality. The representative level of the posture deviation within the normal range of the leaf tilt angle relative to the preset reference tilt angle of a healthy plant under normal dormancy is taken as the reference for normal posture.

[0117] β and β were determined through controlled-condition calibration during system platform construction. During calibration, two types of controlled conditions were created: one with only nighttime spectral disturbances and normal plant posture, and the other with only posture deviations and normal nighttime spectra. Under each condition, the corresponding dimensionless single-channel data and whether and to what extent photosensitivity imbalance occurred in the plant were recorded. Since only one channel of evidence changed under each condition, the contribution of that channel of evidence to photosensitivity imbalance could be isolated, and regressions could be performed separately for the two conditions to obtain the correspondence between each channel of evidence and the degree of photosensitivity imbalance, thereby determining... The relative size of β.

[0118] When the photosensitive imbalance characteristic data is greater than the preset second threshold, a second physical intervention command is generated and output.

[0119] The second threshold serves as the criterion for determining whether a plant exhibits photoperiodic disorder. When the photosensitivity imbalance characteristic data exceeds this threshold, it is considered that the abnormal residence of phytochromes within the plant has reached a level requiring intervention. The second physical intervention command is the control command output by the system when photoperiodic disorder is determined to have occurred. It is used to activate the far-red light rhythm neutralizing lamp, forcibly converting the phytochromes within the plant into an inactive state at the photochemical level, thus establishing a physical dormancy dark period for the plant. Photoperiodic disorder refers to a state in which abnormal nighttime light (including artificial lighting) disrupts the normal conformational balance of phytochromes in plants, thereby hindering their normal dormancy physiological processes. Unlike the first threshold, which exhibits cross-period correction due to nighttime recovery during operation, the second threshold serves as a relatively stable criterion in this scheme. Nighttime adaptation is mainly reflected in the night-to-day correction and day-to-night attitude correction of the first threshold. Therefore, once the second threshold is set, it remains relatively fixed during operation and does not participate in cross-period regulation.

[0120] The second threshold is determined by actual measurements of the physiological critical state of photoperiod disorder in plants. During the system platform setup, abnormal nighttime light is gradually applied to the target plants, and the critical point at which observable dormancy disruption occurs (such as abnormal rhythms in posture or physiological indicators) is observed. The photosensitivity imbalance characteristic data corresponding to this critical point of photoperiod disorder are recorded and used as the second threshold. This second threshold corresponds to the plant's actual photoperiod disorder criticality from the beginning. Exceeding this threshold means that the plant has entered a state requiring rhythmic neutralization intervention.

[0121] The temporal change trajectory of the blade night-sensing tilt angle data in the second time window after the execution of the second physical intervention command is obtained, and the second recovery slope data is calculated. The value of the first threshold of the next daytime operation cycle is updated according to the preset mapping function using the second recovery slope data.

[0122] The value of the first threshold for the next day's operating cycle is updated using the second recovery slope data according to a preset mapping function, specifically including:

[0123] The temporal change trajectory of the blade night-sensing tilt angle data within the second time window after the execution of the second physical intervention command is obtained as the second recovery response trajectory.

[0124] The second recovery response trajectory is a sequence of leaf night-sensing tilt angle data over time within the selected observation interval after the execution of the second physical intervention command. It reflects the recovery process of the plant's posture tending towards normal dormancy and drooping under the influence of far-red light rhythm and intervention.

[0125] It should be noted that there is an effective lag between the physical intervention and the generation of an observable attitude response. Therefore, the observation interval is set after the intervention takes effect, and its starting point and duration are determined by actual measurements during system platform setup. During platform setup, the lag time from the application of the intervention to the appearance of an observable response, and the time from the start of the response to the return to stabilization are recorded. The starting point of the observation interval is set after this lag time, and its duration is set according to the attitude recovery and stabilization process time, ensuring that the observation interval precisely covers the main body of the attitude recovery process. The starting point is moved later to avoid including the window before the intervention takes effect in the trajectory, and the duration is not excessively extended to avoid including random fluctuations after stabilization in the fitting.

[0126] Linear fitting is performed on the second recovery response trajectory to obtain the second recovery slope data and the second goodness of fit characterizing the fit quality. The difference between the first and last values ​​of the second recovery response trajectory is extracted as the second recovery amplitude.

[0127] A linear fit is performed on the second recovery response trajectory, and the slope of the fitted straight line is used as the second recovery slope data. The larger the absolute value, the faster the attitude recovery. At the same time, the second goodness of fit is obtained from the fitting, which characterizes the degree to which the recovery trajectory is close to a straight line, and is used to determine whether the trajectory constitutes a clear and identifiable recovery process. The difference between the values ​​at the beginning and end of the trajectory is taken as the second recovery amplitude, which characterizes the actual change in the night tilt angle during the recovery process.

[0128] When the second goodness of fit is lower than the preset lower limit of the second goodness of fit, or the second recovery magnitude is less than the preset lower limit of the second recovery magnitude, the intervention is determined to be an invalid response, and the first threshold remains unchanged.

[0129] When the second goodness of fit is lower than the preset lower limit of the second goodness of fit, it indicates that the recovered trajectory is flat or messy and does not constitute a clear and identifiable recovery process; when the second recovery amplitude is less than the preset lower limit of the second recovery amplitude, it indicates that the changes at the beginning and end of the trajectory are very small and the difference between the beginning and end is only composed of random fluctuations; if any of the above situations are true, it means that there is no identifiable attitude recovery after intervention, the obtained slope is unreliable, so it is judged as an invalid response and the first threshold is kept unchanged. The second lower limit of goodness of fit and the second lower limit of recovery amplitude were both determined during the system debugging phase. During debugging, two types of recovery trajectories were collected: one with actual recovery and the other with actual invalidity (such as weak triggering or noise spikes). The goodness of fit of the true recovery trajectory was high, and its first-to-last difference significantly exceeded the random fluctuation of the night-sensing tilt angle without intervention. The goodness of fit of the invalid response trajectory was low, and its first-to-last difference consisted only of random fluctuations. The second lower limit of goodness of fit was taken between the high and low ends of the goodness of fit of the true recovery and the invalid response. The second lower limit of recovery amplitude was taken at the level that made the obvious fluctuations of the true recovery distinguishable from the random fluctuations without intervention. This level of random fluctuation was given by the statistical dispersion of the night-sensing tilt angle under the state of no intervention.

[0130] When the second goodness of fit is not lower than the lower limit of the second goodness of fit and the second recovery amplitude is not less than the lower limit of the second recovery amplitude, the intervention is determined to be an effective response. The value of the first threshold for the next day's operating cycle is adjusted based on the absolute value of the second recovery slope data, specifically as follows:

[0131] ;

[0132] in, The first threshold is for d+1 daytime operating cycles. The first threshold for the d-th daytime operating cycle, For the second recovery slope data, This is the preset lower limit of the slope.

[0133] The absolute value of the second recovery slope data characterizes the speed of posture recovery at night. If its absolute value is small, it indicates a slow recovery towards normal drooping posture, meaning that the plant's energy reserves are depleted due to the daytime light and heat stress of the previous day, and its response to nighttime intervention has become dulled. Therefore, the first threshold for the next daytime period should be lowered so that even a small light and heat lag characteristic data can trigger the first physical intervention earlier, providing earlier heat shock protection for plants already in a depleted state. The preset mapping function is set so that the amount of reduction in the first threshold increases with the degree to which the absolute value of the second recovery slope data is lower than the preset slope lower limit, and is positive only when the absolute value of the second recovery slope data is less than the preset slope lower limit. The preset slope lower limit is obtained from the actual measurement of the normal nighttime recovery slope of healthy plants. Using the absolute value of the normal recovery slope of healthy plants after nighttime intervention as a reference, a slow recovery is judged only when the recovery speed is lower than this reference level.

[0134] Target plants, including multiple target plants, the specific process includes:

[0135] Without altering the existing process, this method is extended to multiple target plants, allowing each plant to operate independently according to the aforementioned process, and adding group-level coordination. In another embodiment, while maintaining the method of this example, corresponding preset first and second thresholds are configured for each target plant.

[0136] The initial values ​​of the first and second thresholds for each plant are determined by actual measurements of the plant's critical physiological state of heat shock and critical physiological state of photoperiod disorder, respectively, in the manner described above for single plants, allowing for different values ​​for each plant.

[0137] For each target plant, the photothermal hysteresis feature data are compared with its corresponding first threshold, and the photosensitive imbalance feature data are compared with its corresponding second threshold.

[0138] The first threshold is updated based on the second recovery slope data corresponding to each target plant.

[0139] The percentage of target plants whose first threshold was lowered within the same daytime operating cycle is counted. When the percentage exceeds the preset population percentage limit, the first threshold of all target plants is lowered according to the preset population coefficient.

[0140] Building upon the aforementioned individual plant-by-plant adjustments, this section adds a coordination mechanism at the population level. The system statistically analyzes the number of plants whose first threshold is lowered within the same daytime operating cycle, and its proportion among all plants. When this proportion exceeds the preset upper limit of the population proportion, it indicates that a large number of plants have had their first threshold lowered within the same cycle due to slow nighttime recovery. This is more likely due to a regional heat wave or other deterioration of the population environment, rather than local differences in individual plants. Therefore, the first threshold of all plants is uniformly lowered according to the preset population coefficient, providing earlier heat shock protection for the entire population.

[0141] The system includes a preset upper limit for the proportion of plants in the first threshold, used to distinguish between individual differences and collective stress: a small number of plants having their first threshold lowered can be attributed to individual differences or local microenvironments, requiring no collective action; when the lowered threshold is more than a significant proportion of plants, it is more likely due to a deterioration of the collective environment, so this upper limit is set at more than half, ensuring that collective coordination is only triggered when there is genuine collective stress; if this upper limit is too low, fluctuations in individual plants may cause erroneous global actions, while if it is too high, coordination will not be triggered even when collective stress has occurred. A preset population coefficient is the proportion by which the first threshold of all plants is uniformly lowered after triggering collective coordination, taking a value between zero and one. This coefficient is determined during the system debugging phase based on a trade-off between the strength of collective protection and avoiding excessive intervention: if the coefficient is too small, the lowering will be too drastic, causing excessive intervention in all plants, resulting in resource waste and excessive intervention; if it is too large, the strength of collective protection will be insufficient.

[0142] Please see Figure 2 As shown, the intelligent maintenance and management system for garden plant growth based on multi-source sensing specifically includes:

[0143] The daytime feature extraction module is used to acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle. It analyzes and calculates the light radiation data and leaf-temperature temperature difference data within a preset first time window to generate photothermal hysteresis feature data, and accumulates it during the daytime operation cycle to obtain daytime cumulative data.

[0144] The daytime intervention triggering module is used to generate and output the first physical intervention command when the photothermal hysteresis characteristic data is greater than the preset first threshold.

[0145] The daytime feedback module is used to acquire the temporal change trajectory of the leaf-temperature difference data in the first time window after the execution of the first physical intervention command and to calculate the first recovery slope data. The first recovery slope data is used to adjust the length of the first time window of the next daytime operation cycle.

[0146] The attitude correction module is used to acquire the red-to-red spectral ratio data and leaf night-sensing tilt angle data of the target plant during the nighttime operating cycle. Based on the daytime cumulative data, the spatial deviation correction of the leaf night-sensing tilt angle data is performed to obtain the daytime corrected tilt angle data.

[0147] The nighttime feature synthesis module is used to analyze the red-far-red spectral ratio data within a preset second time window to obtain spectral integral data, and calculate the photosensitivity imbalance feature data by combining the spectral integral data with the daytime correction tilt angle data.

[0148] The nighttime intervention trigger module is used to generate and output a second physical intervention command when the photosensitivity imbalance characteristic data is greater than a preset second threshold.

[0149] The nighttime feedback module is used to acquire the temporal change trajectory of the blade nighttime tilt angle data in the second time window after the execution of the second physical intervention command, calculate the second recovery slope data, and use the second recovery slope data to update the value of the first threshold of the next daytime operating cycle according to the preset mapping function.

[0150] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the above functions can be divided into different functional modules to complete all or part of the functions described above.

[0151] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units through some interfaces, and may be electrical, mechanical, or other forms.

[0152] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent maintenance and management of garden plant growth based on multi-source perception, characterized in that, include: Acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle, analyze and calculate the light radiation data and leaf-temperature temperature difference data within the preset first time window, generate light and heat hysteresis characteristic data, and accumulate them during the daytime operation cycle to obtain daytime cumulative data. When the photothermal hysteresis feature data is greater than a preset first threshold, a first physical intervention command is generated and output. The temporal change trajectory of the leaf temperature difference data in the first time window after the execution of the first physical intervention command is obtained and the first recovery slope data is calculated. The length of the first time window of the next daytime operation cycle is adjusted using the first recovery slope data. Acquire the red-to-red spectral ratio data and leaf night-sensing tilt angle data of the target plant during the nighttime operating cycle, and obtain the daytime corrected tilt angle data by spatial bias correction of the leaf night-sensing tilt angle data based on the daytime cumulative data. The red-to-red spectral ratio data is analyzed within a preset second time window to obtain spectral integral data, and the spectral integral data is calculated with the daytime correction tilt angle data to obtain photosensitivity imbalance characteristic data. When the photosensitive imbalance feature data is greater than a preset second threshold, a second physical intervention command is generated and output. The temporal change trajectory of the blade night-sensing tilt angle data in the second time window after the execution of the second physical intervention command is obtained and the second recovery slope data is calculated. The value of the first threshold of the next daytime operation cycle is updated using the second recovery slope data according to a preset mapping function.

2. The method of claim 1, wherein the method further comprises: determining a plant health index of the plant based on the plant health data; and determining a plant health index of the plant based on the plant health data. The generated photothermal hysteresis feature data specifically includes: The rate of change of light radiation and the rate of change of temperature difference are obtained by performing derivative calculations on the light radiation data and the leaf temperature difference data, respectively. Monitor the optical radiation data, and record the moment when the absolute value of the change in the optical radiation data relative to the previous moment exceeds the preset jump variable as the jump start moment; Starting from the jump start time, the duration for which the optical radiation data remains deviated from the optical radiation baseline value before the jump start time is counted, and the jump end time is recorded when the optical radiation data falls back to the preset neighborhood of the optical radiation baseline value. When the duration between the start time of the transition and the end time of the transition is less than a preset duration, the data segment between the start time of the transition and the end time of the transition is determined to be a pseudo-transition, and the rate of change of light radiation and the rate of change of temperature difference at the corresponding time in the data segment are removed. The product of the removed rate of change of light radiation and the rate of change of temperature difference is integrated within the first time window to obtain the photothermal hysteresis characteristic data. The percentage of eliminated moments within the first time window is calculated relative to the total number of moments. When the percentage of eliminated moments exceeds a preset upper limit, the length of the first time window is extended proportionally to the percentage of eliminated moments and then re-integrated. 3.The method of claim 2, wherein the method further comprises: determining a growth state of the garden plant according to the growth state of the garden plant in the growth state database; and determining a growth state of the garden plant according to the growth state of the garden plant in the growth state database. Accumulated daytime data is obtained, specifically including: The daytime operating cycle is divided into a preset number of sub-periods, and the cumulative value of the photothermal hysteresis characteristic data is calculated for each sub-period. The cumulative value of each time period is compared with a preset feature upper limit, and the cumulative value of the time period that exceeds the feature upper limit is limited to the feature upper limit; The cumulative values ​​for each time period after the amplitude limiting are weighted and summed according to their corresponding weights to obtain the daytime cumulative data; When the number of cumulative values ​​in each time period that exceed the upper limit of the feature exceeds the preset number of anomalies, the current day period is determined to be an abnormal day, and the cumulative data of the day period is multiplied by a preset attenuation coefficient and then output.

4. The method of claim 1, wherein the method further comprises: determining a plant health index of the plant based on the plant health data; and determining a plant health index of the plant based on the plant health data. The step of adjusting the length of the first time window of the next daytime running cycle using the first recovery slope data specifically includes: The leaf-temperature temperature difference data within the first time window after the execution of the first physical intervention command is obtained as the first recovery response trajectory. Linear fitting is performed on the first recovery response trajectory to obtain the first recovery slope data and the first goodness of fit characterizing the goodness of fit, and the difference between the first and last values ​​of the first recovery response trajectory is extracted as the first recovery amplitude. When the first goodness of fit is lower than the preset first goodness of fit lower limit, or the first recovery amplitude is less than the preset first recovery amplitude lower limit, the intervention is determined to be an invalid response, and the length of the first time window is not updated based on the first recovery slope data. When the first goodness of fit is not lower than the first goodness of fit lower limit and the first recovery amplitude is not less than the first recovery amplitude lower limit, the intervention is determined to be an effective response, and the length of the first time window of the next day period operation cycle is adjusted according to the absolute value of the first recovery slope data.

5. The intelligent maintenance and management method for garden plant growth based on multi-source sensing as described in claim 3, characterized in that: Based on the cumulative daytime data, spatial bias correction is performed on the leaf night-sensing tilt angle data to obtain the daytime corrected tilt angle data. The specific process includes: Extract the spatial deviation amplitude of the blade night-sensing tilt angle data relative to the preset reference tilt angle; Extract the high-frequency jitter component of the blade night-sensing tilt angle data. When the amplitude of the high-frequency jitter component exceeds the preset jitter amplitude, determine that the corresponding data segment is an external force disturbance and remove it from the calculation of the spatial deviation amplitude. The attitude correction offset is obtained by converting the accumulated daytime data according to a preset ratio. The attitude correction offset increases as the accumulated daytime data increases. The daytime correction tilt angle data is obtained by subtracting the attitude correction offset from the spatial deviation amplitude. When the attitude correction offset is greater than the spatial deviation amplitude, the daytime correction tilt data is set to zero and the preset proportion of the daytime cumulative data participating in the correction of the next nighttime operating cycle is reduced.

6. The intelligent maintenance and management method for garden plant growth based on multi-source sensing as described in claim 1, characterized in that: The process of analyzing the far-infrared spectral ratio data within a preset second time window to obtain spectral integral data specifically includes: Perform a discrete Fourier transform on the time series of the far-infrared spectral ratio data within the second time window to obtain the amplitude spectrum with frequency as the independent variable; In the amplitude spectrum, frequency components within a preset frequency range are selected as candidate interference components; For each candidate interference component, extract the amplitude of a preset number of adjacent frequency components that are outside the preset frequency range, and use the statistical value of the amplitude of the preset number of adjacent frequency components as the background amplitude of the candidate interference component. The difference between the amplitude of the candidate interference component and the background amplitude is calculated as the interference excess. When the interference excess is greater than zero, the interference excess is subtracted from the amplitude of the candidate interference component to bring it back to the background amplitude. When the interference excess is not greater than zero, the amplitude of the candidate interference component is kept unchanged to obtain the processed amplitude spectrum. Perform a discrete inverse Fourier transform on the processed amplitude spectrum to obtain the suppressed red-far-red spectral ratio data; The suppressed red-to-red spectral ratio data is integrated within the second time window to obtain the spectral integral data.

7. The intelligent maintenance and management method for garden plant growth based on multi-source sensing as described in claim 3, characterized in that: The value of the first threshold for the next day's operating cycle is updated using the second recovery slope data according to a preset mapping function, specifically including: The temporal change trajectory of the blade night-sensing tilt angle data within the second time window after the execution of the second physical intervention command is obtained as the second recovery response trajectory. Linear fitting is performed on the second recovery response trajectory to obtain the second recovery slope data and the second goodness of fit characterizing the goodness of fit, and the difference between the first and last values ​​of the second recovery response trajectory is extracted as the second recovery amplitude. When the second goodness of fit is lower than the preset second goodness of fit lower limit, or the second recovery amplitude is less than the preset second recovery amplitude lower limit, the intervention is determined to be an invalid response, and the first threshold is maintained unchanged; When the second goodness of fit is not lower than the second goodness of fit lower limit and the second recovery amplitude is not less than the second recovery amplitude lower limit, the intervention is determined to be an effective response, and the value of the first threshold of the next day period operation cycle is adjusted according to the absolute value of the second recovery slope data.

8. The intelligent maintenance and management method for garden plant growth based on multi-source sensing as described in claim 4, characterized in that: Adjusting the first time window also includes: Record the historical adjustment trajectory within a preset statistical number of daytime operation cycles for the length of the first time window; When the number of times the adjustment direction alternates between positive and negative during adjacent daytime operation cycles in the historical adjustment trajectory exceeds the preset number of oscillations, it is determined to be an oscillation, and the preset length range is reduced according to the preset tightening coefficient; When the historical adjustment trajectory accumulates monotonically in the same direction within several daytime operating cycles of the preset statistical period, and the accumulated amount exceeds the preset drift limit, it is determined to be a drift, and the length of the first time window is reset to the preset reference length.

9. The intelligent maintenance and management method for garden plant growth based on multi-source sensing as described in claim 1, characterized in that: The target plant also includes multiple target plants, and the specific process includes: Configure corresponding preset first and second thresholds for each target plant; For each target plant, the photothermal hysteresis feature data are compared with its corresponding first threshold, and the photosensitive imbalance feature data are compared with its corresponding second threshold. The first threshold is updated based on the second recovery slope data corresponding to each of the target plants; The percentage of target plants whose first threshold was lowered within the same daytime operating cycle is counted. When the percentage exceeds the upper limit of the preset population percentage, the first threshold of all target plants is lowered according to the preset population coefficient.

10. A smart maintenance and management system for garden plant growth based on multi-source sensing, characterized in that: Specifically, it includes: The daytime feature extraction module is used to acquire light radiation data and leaf-temperature temperature difference data of the target plant during the daytime operation cycle. It analyzes and calculates the light radiation data and leaf-temperature temperature difference data within a preset first time window to generate photothermal hysteresis feature data, and accumulates it during the daytime operation cycle to obtain daytime cumulative data. The daytime intervention triggering module is used to generate and output a first physical intervention command when the photothermal hysteresis feature data is greater than a preset first threshold. The daytime feedback module is used to acquire the temporal change trajectory of the leaf temperature difference data in the first time window after the execution of the first physical intervention command and to calculate the first recovery slope data, and to adjust the length of the first time window of the next daytime operation cycle using the first recovery slope data. The attitude correction module is used to acquire the red-to-red spectrum ratio data and leaf night-sensing tilt angle data of the target plant during the nighttime operation cycle. Based on the daytime cumulative data, the spatial deviation correction of the leaf night-sensing tilt angle data is performed to obtain the daytime corrected tilt angle data. The nighttime feature synthesis module is used to analyze the red-far-red spectral ratio data within a preset second time window to obtain spectral integral data, and to calculate the photosensitivity imbalance feature data by combining the spectral integral data with the daytime correction tilt angle data. The nighttime intervention trigger module is used to generate and output a second physical intervention command when the photosensitivity imbalance feature data is greater than a preset second threshold. The nighttime feedback module is used to acquire the temporal change trajectory of the blade nighttime tilt angle data in the second time window after the execution of the second physical intervention command, calculate the second recovery slope data, and use the second recovery slope data to update the value of the first threshold of the next daytime operating cycle according to a preset mapping function.

Images