A method for predicting a harvesting time window and scheduling batch harvesting for stable quality of cannabichromene

Abstract

This invention provides a method for predicting the harvest window and scheduling staggered harvesting for stable quality of oxalic acid, relating to the field of modern oxalic acid harvesting technology. The method includes: acquiring historical meteorological data, plot spatial data, and harvesting, transportation, and pretreatment capacity data to generate a baseline staggered planting and production plan; collecting physiological development data, stress data, and non-destructive testing data, and combining this with laboratory testing to establish a quality index prediction model to determine the actual optimal harvesting time window; constructing a staggered harvesting scheduling model using an asymmetric loss function; implementing microclimate intervention when allowable deviations are exceeded; and controlling the maximum safe post-harvest retention time based on a post-harvest quality decay model. This invention is beneficial for improving the stable yield and harvest quality of oxalic acid raw materials.

Description

Technical Field

[0001] This invention belongs to the field of modern harvesting technology for sarsaparilla acid, specifically relating to a method for predicting harvesting time windows and scheduling batch harvesting to ensure stable sarsaparilla acid quality. Background Technology

[0002] Rosemary and other plants contain caryopsisic acid, an important natural active ingredient whose content and stability directly affect the quality of raw materials and their subsequent utilization value. Unlike common crops that rely mainly on biomass accumulation for harvesting, caryopsisic acid is a secondary metabolite affected by both plant development and environmental stress, and its content changes are highly dynamic and phased.

[0003] Under large-scale planting conditions, existing harvesting methods mainly have the following problems: Firstly, current production methods often rely on planting days, plant height, leaf color, or experience to determine the harvest time, which makes it difficult to accurately reflect the true changing trend of the target component, oxalic acid, leading to harvesting too early or too late. Secondly, adjacent plots are likely to reach a higher quality state within a similar time frame under similar climatic conditions. However, the limited manpower, machinery, transportation and pre-processing capacity for harvesting means that some plots, although reaching a better state, cannot be harvested in time. Third, existing scheduling methods are unable to reflect the asymmetry of quality loss. For target components such as oxalic acid, early harvesting usually results in insufficient yield or content, while delayed harvesting may lead to faster oxidative degradation or quality decline. The loss mechanisms of the two are different, and if a symmetrical scheduling logic is still used, it will be difficult to achieve stable high-quality harvesting. Fourth, existing technologies typically lack a coordinated mechanism that integrates harvest window identification, batch scheduling, on-site correction, and post-harvest time limit control, making it difficult to balance field harvesting organization efficiency with post-harvest quality maintenance.

[0004] Therefore, there is a need for a harvesting method that can combine growth status, environmental stress, batch scheduling, on-site intervention and post-harvest time limit control in a closed-loop coordinated manner to improve the quality stability of sarsaparilla acid and the efficiency of large-scale production organization. Summary of the Invention

[0005] To overcome the problems in the background technology, this invention develops a method for predicting harvest window and scheduling batch harvesting for stable quality of oxalic acid, which solves the problems of difficulty in predicting the harvest window and high incidence of capacity squeeze in the large-scale harvesting of oxalic acid.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: A method for predicting harvest window and scheduling staggered harvesting to ensure stable quality of oxalic acid includes the following steps: S1. Before planting, obtain historical meteorological data, plot spatial data, and harvesting, transportation and pre-processing capacity data of the target planting area, determine the target harvesting season window, and generate a benchmark batch planting and production schedule based on the daily effective processing capacity. S2. During the crop growth cycle, continuously collect physiological development data representing the physiological development status of the plant, stress data representing the degree of environmental stress, and non-destructive testing data representing changes in canopy components. Obtain samples according to preset sampling rules for laboratory testing. Pair and calibrate the laboratory test results with the synchronously collected multi-source data to establish or update the quality index prediction model. Use the quality index prediction model to predict the oxalic acid-related quality score of each target plot in the future preset time period, and determine the actual optimal harvest time window of each target plot and the time window drift amount relative to the benchmark batch planting and production plan. S3. Under the constraints of daily effective processing capacity, transportation capacity and preprocessing capacity, construct a batch harvesting scheduling model with different loss functions for early harvesting and delayed harvesting, and solve for the harvesting date, harvesting order and resource allocation scheme of each target plot. S4. When the total scheduling loss exceeds the preset threshold and / or the planned harvest date of a target plot deviates from the center of its actual optimal harvest time window by more than the allowable harvest deviation, microclimate intervention is implemented for the corresponding target plot; after intervention, data is re-collected, the actual optimal harvest time window is updated, and scheduling solution is executed again. S5. Generate harvesting operation instructions based on the solution results, and calculate the maximum safe retention time threshold after harvesting based on ambient temperature, light conditions and post-harvest quality decay model. Control the total time from cutting to entering the cool storage area or pre-treatment station within the maximum safe retention time threshold after harvesting.

[0007] Furthermore, in S1, seasonal periods characterized by high sunshine, strong ultraviolet radiation, low precipitation, and large diurnal temperature range are identified based on multi-year historical meteorological data as the target harvesting season window; the effective daily processing capacity is the minimum value among the area processing capacity corresponding to the harvesting capacity, transportation capacity, and pre-treatment capacity, wherein the harvesting capacity is calculated according to the following formula: Sh = N × v × Tsafe × eta; In the formula, Sh is the daily harvesting capacity, N is the number of harvesting operation units, v is the harvesting area per unit time of a single harvesting operation unit, Tsafe is the length of the safe operation period, and eta is the equipment utilization coefficient.

[0008] Furthermore, S1 also includes spatial allocation of different batches of crops based on the digital elevation model of the target planting area, slope aspect, slope position, soil water retention and heat accumulation differences; wherein, batches expected to develop slowly or allowed to be appropriately delayed are allocated to plots with low heat accumulation and high water retention, and batches expected to mature faster are allocated to plots with high heat accumulation, strong sunlight or fast evaporation.

[0009] Furthermore, the physiological development data in S2 includes at least one or more of the following: effective accumulated temperature, number of growing days, plant height, number of branches, canopy coverage, or leaf area index; the stress data includes at least one or more of the following: light intensity, ultraviolet radiation intensity, soil moisture content, soil tension, air temperature and humidity, leaf temperature, wind speed, or vapor pressure difference; the non-destructive testing data includes at least one or more of the following: canopy multispectral data, canopy hyperspectral data, canopy thermal infrared data, or vegetation index data; the laboratory testing data includes at least one or more of the following: oxalic acid content (CA), oxalic acid to degradation product ratio (R), and effective component yield per unit area (Y); the laboratory testing employs high-performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS); the canopy hyperspectral data in the non-destructive testing data specifically extracts reflectance data containing characteristic wavelengths of 230 nm and 280 nm.

[0010] Furthermore, a comprehensive quality score Q(t) is constructed, and the set of dates that satisfy Q(t) ≥ theta × Qmax is determined as the actual optimal harvesting time window, where Qmax is the maximum quality score within the preset prediction period, and theta is the time window threshold coefficient; Q(t) is calculated by the following formula: Q(t) = w1 × Norm[CA(t)] + w2 × Norm[R(t)] + w3 × Norm[Y(t)]; In the formula, w1, w2, and w3 are weights and w1+w2+w3=1; the difference between the center date tcenter of the time window and the planned harvest date tplanned in the baseline batch planting and production schedule is defined as the time window drift: Deltat=tcenter-tplanned.

[0011] Further, in S3, let x(i,t) represent the decision variable for whether plot i is harvested on date t. The scheduling model satisfies the constraints that each plot is harvested only once, the daily harvested area does not exceed the daily effective processing capacity, the daily transportation volume does not exceed Mmax, and the daily preprocessing volume does not exceed Pmax. The loss Li(t) when plot i is harvested on date t... When t≤topt,i, Li(t)=alpha_i×(topt,it); When t>topt,i, Li(t)=beta_i×[exp(lambda_i×(t-topt,i))-1], Where topt,i is the optimal harvesting date for plot i, and beta_i is greater than alpha_i; the goal of solving the batch harvesting scheduling model is to minimize the total loss.

[0012] Furthermore, in S4, measures such as water replenishment, shading, cooling, or reduction of stress intensity are implemented for target plots that mature too early, while measures such as water control, increased light exposure, or increased stress intensity are implemented for target plots that mature too late. The microclimate intervention lasts for 2-5 days. During the intervention, soil moisture content, leaf temperature, light conditions, and non-destructive testing data of the canopy are re-collected according to a retesting cycle of 24-48 hours, and the actual optimal harvest time window is updated. To avoid irreversible damage caused by excessive intervention, safety boundaries for soil moisture content, leaf temperature, and environmental stress intensity are pre-set and monitored in real time during the intervention process.

[0013] Furthermore, in S5, the maximum safe retention time threshold Tlimit after harvest is calculated based on the post-harvest quality decay model; the oxalic acid retention rate of the sample at different time points is determined in advance through post-harvest retention experiments under different temperature, light, and stacking conditions, and the decay rate constant k is obtained by fitting; when the post-harvest quality change is represented by a first-order decay model, we have: C(t) = C0 × exp(-k × t); Tlimit = -ln(theta_p) / k; In the formula, C0 is the initial oxalic acid content after harvest, C(t) is the oxalic acid content after time t, k is the decay rate constant, and theta_p is the allowable retention rate. When the on-site queuing information or transportation congestion prediction indicates that the expected retention time will exceed Tlimit, the system will automatically adjust the harvesting order, reduce the priority of subsequent plot cutting, or suspend the cutting of new plots.

[0014] A batch harvesting collaborative system for implementing a harvesting window prediction and batch harvesting scheduling method for ensuring stable quality of oxalic acid includes: The planning module is used to acquire historical meteorological data, plot spatial data, and harvesting, transportation, and preprocessing capacity data, and generate a baseline batch planting and production schedule. The perception and prediction module is used to collect physiological development data, stress data and non-destructive testing data, and establish or update the quality index prediction model based on the pairing and calibration of laboratory test results and multi-source data, and output the actual optimal harvesting time window and time window drift for each target plot. The scheduling solution module is used to generate a batch harvesting scheduling scheme based on an asymmetric penalty mechanism under the constraints of daily effective processing capacity, transportation capacity and preprocessing capacity. The intervention control module is used to implement microclimate intervention and trigger closed-loop update for the target plot when the total scheduling loss exceeds the preset threshold or the target plot exceeds the allowable harvesting deviation. The collaborative execution module is used to generate harvesting operation instructions and control the maximum safe dwell time after harvesting.

[0015] Furthermore, the sensing and prediction module is communicatively connected to distributed IoT monitoring nodes, drones or ground non-destructive testing terminals and laboratory testing terminals; the intervention control module is communicatively connected to irrigation, shading or environmental regulation actuators; the collaborative execution module is communicatively connected to harvesting operation terminals, transportation terminals and pre-processing terminals, and writes the maximum safe retention time threshold after harvesting into the harvesting operation instructions.

[0016] The beneficial effects of this invention are: 1. This invention combines the asymmetric losses caused by early and late harvesting during the formation of oxalic acid quality by constructing a batch harvesting scheduling model that uses different loss functions for early and late harvesting. This helps to alleviate harvesting conflicts caused by multiple plots entering the harvestable state at similar times and improves the stable yield of oxalic acid raw materials.

[0017] 2. When the total loss of harvest scheduling exceeds a preset threshold or the target plot exceeds the allowable harvest deviation, the present invention implements microclimate intervention on the corresponding plot, and re-identifies the time window and solves the scheduling after the intervention, thereby dynamically correcting the maturity rhythm of local plots, alleviating the squeeze on harvesting resources under large-scale planting conditions, and ensuring harvesting quality.

[0018] 3. This invention combines the determination of the target harvesting season window, the allocation of land space, the monitoring of multi-source data, the prediction of quality indicators, the scheduling of batch harvesting, the intervention of microclimate, and the control of the maximum safe retention time after harvesting, which is conducive to improving the systematicness and synergy of the harvesting organization of oxalic acid and realizing the stable and high-quality production of raw materials. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall process of the method of the present invention; Figure 2 This is a schematic diagram of the prediction, scheduling, intervention, and closed-loop update process of the present invention; Figure 3 This is a schematic diagram of the batch harvesting coordination system of the present invention; Figure 4 This is a schematic diagram of the asymmetric penalty function of the present invention; Figure 5This is a data fusion architecture diagram of the quality index prediction model of this invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and beneficial effects of the present invention clearer, the preferred embodiments of the present invention will be described in detail below to facilitate understanding by those skilled in the art.

[0021] Example 1 See Figure 1 This embodiment provides a method for predicting harvest window and scheduling staggered harvesting to ensure stable quality of ursolic acid: Before planting, obtain historical meteorological data of the target planting area for the past 5-20 years, and statistically analyze indicators such as total radiation, ultraviolet radiation, precipitation, daily maximum temperature, daily minimum temperature and diurnal temperature range on a daily or weekly scale; through threshold screening, empirical statistics or cluster analysis, identify continuous periods of high light, strong ultraviolet radiation, low precipitation and large diurnal temperature range that are more related to the accumulation of oxalic acid, and use these as the target harvesting season window; Simultaneously, data on harvesting, transportation, and pre-processing capabilities are acquired; the harvesting capability data includes at least the number of harvesting units N, the harvested area v per unit time per harvesting unit, the safe operating period length Tsafe, and the equipment utilization coefficient eta. Preferably, the harvesting capability is calculated using the following formula: Sh = N × v × Tsafe × eta; In the formula, Sh is the daily harvesting capacity, N is the number of harvesting operation units, v is the harvesting area per unit time of a single harvesting operation unit, Tsafe is the length of the safe operation period, and eta is the equipment utilization coefficient. Furthermore, the harvesting capacity, transportation capacity, and pretreatment capacity are converted into corresponding area treatment capacities, and the minimum value among them is taken as the daily effective treatment capacity Sday. Based on the daily effective treatment capacity Sday and the target harvest season window length, the recommended planting date range for each batch is deduced, and a benchmark batch planting and production schedule is generated so that different batches of crops can enter the suitable harvesting state in a dispersed manner within the target harvest season window. In a preferred embodiment, spatial allocation is also carried out in combination with the digital elevation model of the target planting area, slope aspect, slope position, soil water retention and heat accumulation differences; batches that are expected to develop slowly or can be appropriately delayed are allocated to plots with low heat accumulation and high water retention, while batches that are expected to mature faster are allocated to plots with high heat accumulation, strong sunlight or fast evaporation, thereby reserving buffer space for subsequent scheduling by utilizing the differences in plot microenvironment.

[0022] Example 2 See Figure 1 and Figure 5This embodiment describes the process of collecting physiological development data, stress data, and non-destructive testing data; pairing and calibrating laboratory test results with multi-source data; establishing or updating quality indicator prediction models; and identifying the actual optimal harvesting time window and the amount of time window drift. During the crop growth cycle, multi-source data are continuously collected from each target plot. This multi-source data includes physiological development data, stress data, and non-destructive testing data. The physiological development data includes at least one or more of the following: effective accumulated temperature, number of growing days, plant height, number of branches, canopy coverage, and leaf area index. The stress data includes at least one or more of the following: light intensity, ultraviolet radiation intensity, soil moisture content, soil tension, air temperature and humidity, leaf temperature, wind speed, and vapor pressure difference. The non-destructive testing data includes at least one or more of the following: canopy multispectral data, canopy hyperspectral data, canopy thermal infrared data, and vegetation index data. Preferably, one IoT monitoring node is set up for every 20-50 mu (approximately 13-2 hectares), and the sampling cycle of the IoT monitoring node is 10-30 minutes. Every 2-4 days, a drone platform is used to perform flight data collection during a relatively stable lighting period, and whiteboard calibration, dark current calibration, or radiometric calibration are performed before the flight. In addition to the drone platform, the non-destructive testing terminal can also use ground-based fixed or mobile testing equipment. To establish the correspondence between non-destructive testing results and actual chemical indicators, samples are drawn from each target plot according to a preset sampling rule for laboratory testing. Preferably, 10-20 plants are randomly selected from each target plot every 5-7 days, and representative leaves or tender branches are taken for testing. In another preferred embodiment, 12 plants are randomly selected from each micro-plot every 7 days for testing. The laboratory testing employs high-performance liquid chromatography or liquid chromatography-mass spectrometry, and the detection indicators include at least one or more of the following: sarsaparilla oxalate content (CA), sarsaparilla oxalate to degradation product ratio (R), and effective component yield per unit area (Y). Pair and calibrate laboratory test results with non-destructive testing data, physiological development data, and stress data collected from the same plot of land at the same or similar times to establish or update quality indicator prediction models; In a preferred embodiment, the canopy hyperspectral data in the nondestructive testing data undergoes preprocessing before modeling. This preprocessing includes at least one or more of the following: spectral smoothing, standard normal variable transformation, derivative enhancement, outlier removal, and feature selection, to reduce the impact of noise, background scattering, and environmental fluctuations on the inversion results. After preprocessing, reflectance features including the 230nm and 280nm characteristic bands are extracted from the canopy hyperspectral data, and / or ratio features, difference features, derivative spectrum features, or narrowband exponential features composed of the 230nm and 280nm characteristic bands are extracted. The aforementioned features were paired and calibrated with the oxalic acid content (CA) and the ratio of oxalic acid to degradation products (R) obtained from laboratory testing to establish a quality index prediction model for retrieving the quality status of oxalic acid. Specifically, the band near 230 nm was used to characterize the spectral response related to oxalic acid, and the band near 280 nm was used to characterize the spectral response related to degradation products. In practical applications, adjacent narrow bands can also be selected near the aforementioned characteristic bands, or a feature selection algorithm can be used to determine band combinations highly correlated with oxalic acid content and the ratio of oxalic acid to degradation products to improve the robustness and adaptability of the model. The quality indicator prediction model may employ partial least squares regression, random forest, gradient boosting model, temporal neural network or a combination thereof, with the aim of outputting the quality change curve related to the quality of oxalic acid within a future preset time period. In a preferred embodiment, a comprehensive quality score Q(t) is constructed to identify the actual optimal harvesting time window, wherein Q(t) is calculated using the following formula: Q(t)=w1×Norm[CA(t)]+w2×Norm[R(t)]+w3×Norm[Y(t)]; In the formula, w1, w2, and w3 are weights, and w1 + w2 + w3 = 1. The maximum quality score within the preset prediction period is denoted as Qmax. The set of dates that satisfy Q(t) ≥ theta × Qmax is selected as the actual optimal harvesting time window, where theta is the time window threshold coefficient. w1, w2, and w3 can be preset according to the target product's focus on the content of effective ingredients, degradation risk, and yield per unit area, or they can be determined by training based on historical batch quality evaluation results; theta can be preset according to the width of the harvest time window to be controlled. The higher the value of theta, the narrower the optimal harvest time window will be identified. Let tcenter be the center date of the actual optimal harvest time window, and tplanned be the planned harvest date in the baseline batch planting and production schedule. Then, the time window drift Delta is defined as: Deltat = tcenter - tplanned; When Deltat is positive, it indicates that the corresponding plot matures relatively late; when Deltat is negative, it indicates that the corresponding plot matures relatively early.

[0023] Example 3 See Figure 2 and Figure 4 This embodiment describes the batch harvesting scheduling model in step S3: After determining the actual optimal harvesting time window for each plot, a batch harvesting scheduling model is constructed: the target planting area is divided into m plots, the area of ​​plot i is Ai, and the decision variable x(i,t) indicates whether plot i is harvested on date t; the scheduling model satisfies the constraints that each plot is harvested only once, the daily harvesting area does not exceed the daily effective processing capacity, the daily transportation volume does not exceed Mmax, and the daily preprocessing volume does not exceed Pmax. To reflect the difference in losses between early and late harvesting, different loss functions are set for them. Let the optimal harvesting date for plot i be topt,i. Then, when plot i is harvested on date t, its loss Li(t) can be expressed as: When t≤topt,i Li(t) = alpha_i × (topt, it); When t > topt,i Li(t)=beta_i×[exp(lambda_i×(t-topt,i))-1]; In the formula, alpha_i is the penalty coefficient for early harvesting, beta_i and lambda_i are the penalty coefficients for delayed harvesting, and beta_i is greater than alpha_i. alpha_i, beta_i and lambda_i can be obtained by fitting the degree of loss to the target quality index caused by early and delayed harvesting in historical batches, where lambda_i is used to characterize the acceleration of the increase in losses from delayed harvesting over time. The scheduling scheme is solved by minimizing the sum of losses of each plot. Preferably, mixed integer programming, linear programming after piecewise linear approximation, or heuristic optimization methods can be used to solve the scheme, thereby obtaining the harvesting date, harvesting order, and resource allocation scheme for each target plot.

[0024] Example 4 See Figure 2 This embodiment mainly describes step S4, which is that when the total scheduling loss exceeds a preset threshold and / or the target plot exceeds the allowable harvesting deviation, microclimate intervention is implemented on the corresponding target plot, and time window identification and scheduling solution are re-performed after the intervention: When the scheduling solution results show that the total loss exceeds the preset threshold, and / or the planned harvest date of the target plot deviates from the center of its actual optimal harvest time window by more than the allowable harvest deviation, microclimate intervention is implemented on the corresponding target plot to change its subsequent growth and quality index evolution process.

[0025] For plots that mature early and may crowd out harvestable resources in the short term, mitigation measures are preferred, including water replenishment, shading, cooling, or reducing stress intensity, to delay the arrival of their peak quality. For plots that mature late and may slip out of the target harvest season window or miss resource gaps, promotion measures are preferred, including water control, increased light exposure, or increased stress intensity, to appropriately advance their peak quality. In a preferred embodiment, for plots that mature early, water is supplemented by micro-sprinkler irrigation or drip irrigation, and a shade net with a shading rate of 30%-50% is set up, preferably maintaining the soil moisture content at 65%-80% of field capacity; for plots that mature late, it is preferable to control the irrigation amount or stop irrigation for a short period of time, cancel the shading and increase the light exposure, preferably controlling the soil moisture content at 40%-55% of field capacity, and making it close to but not exceed the preset wilting risk threshold; Preferably, the microclimate intervention lasts for 2-5 days. During the intervention, soil moisture content, leaf temperature, light conditions, and canopy non-destructive testing data are re-collected at 24-48 hour retesting intervals, and the actual optimal harvesting time window for each target plot is updated. If the center of the updated time window returns to the allowable deviation range, the intervention is stopped and the scheduling solution is executed again; if it still does not return to the allowable deviation range, one or more interventions can be implemented until the preset stopping condition is met. To avoid irreversible damage from excessive intervention, safety boundaries for soil moisture content, leaf temperature, and environmental stress intensity can be preset and monitored in real time during the intervention. When the monitored values ​​approach or reach the safety boundaries, the intervention intensity should be reduced or the intervention should be stopped.

[0026] Example 5 See Figure 1 To reduce the risk of continued quality decline after harvest, this embodiment incorporates post-harvest waiting time into unified control and establishes a post-harvest quality decay model in advance through post-harvest retention test; specifically, under different temperature, light and stacking conditions, the oxalic acid retention rate of freshly harvested samples is measured at different time points, and the decay rate constant k is obtained by fitting. In a preferred embodiment, a first-order decay model is used to represent the post-harvest quality change: C(t) = C0 × exp(-k × t) In the formula, C0 is the initial oxalic acid content after harvest, C(t) is the oxalic acid content after time t, and k is the decay rate constant. If the allowable retention rate theta_p is used as the minimum acceptance criterion, then the maximum safe retention time threshold Tlimit after sampling is: Tlimit = -ln(theta_p) / k; In the formula, theta_p is the allowable retention rate, preferably 0.90-0.98. The k can be obtained by fitting post-harvest retention tests under different temperature, light and stacking conditions, and dynamically corrected in practical applications by combining real-time temperature, light conditions and queuing status of transportation or pretreatment. When generating harvesting operation instructions based on scheduling results, in addition to including the plot number, planned harvesting start time, planned harvesting end time, harvesting sequence, corresponding harvesting team, transportation handover node, and pre-processing station information, the maximum safe post-harvest dwell time threshold Tlimit is also written into the harvesting operation instructions. The system controls the total time from cutting to entering the cool storage area or pre-processing station to not exceed Tlimit; When on-site queuing information or transportation congestion predictions indicate that the expected dwell time will exceed the Tlimit, the system automatically adjusts the harvesting sequence, lowers the priority of subsequent plot cutting, or suspends new plot cutting to reduce quality loss caused by post-harvest oxidation.

[0027] Example 6 This embodiment illustrates the comprehensive application process of the method of the present invention: Large-scale rosemary plantations divide the target area into multiple micro-plots and generate a baseline batch planting and production plan before the start of the harvest season based on historical meteorological data, plot spatial data, and harvesting, transportation, and pre-processing capacity data. During the growth period, the system continuously collects physiological development data, stress data, and non-destructive testing data from each plot, and randomly selects 12 samples from each micro-plot every 7 days for high-performance liquid chromatography analysis to establish and update the quality indicator prediction model. Within a specific scheduling cycle, the scheduling solution module solved multiple plots and found that some plots would have total scheduling losses exceeding the preset threshold due to time window drift. For micro-plot No. 23, the intervention control module issued a promoting intervention instruction to reduce irrigation for three consecutive days, control soil moisture content at approximately 50% field capacity, and close the shading components in the area to enhance mild environmental stress and advance its optimal harvest time window. For micro-plot No. 41, the intervention control module issued a mitigating intervention instruction to supplement water through micro-sprinkler irrigation for two consecutive days and deploy shading components with a shading rate of approximately 40% to reduce the intensity of local environmental stress and delay the arrival of its optimal harvest time window. The perception and prediction module reacquired field data and non-destructive testing data 48 hours after the intervention and updated the quality prediction results. The updated results showed that the time window center of micro-plot No. 23 was brought forward from October 11 to October 9, and the time window center of micro-plot No. 41 was postponed from October 8 to October 9, both of which were within the allowable deviation range. The scheduling solution module re-executed the solution based on this and obtained the updated batch harvesting scheduling plan. In terms of post-harvest collaborative control, the collaborative execution module generates harvesting operation instructions based on the updated scheduling results and sends them to the harvesting operation terminal, transportation terminal, and pre-processing terminal respectively. Assuming the average temperature on the day of harvesting is 28℃, the post-harvest attenuation rate constant k is 0.017h^-1 based on historical calibration. If the minimum allowable retention rate theta_p is taken as 0.95, then Tlimit is calculated to be approximately 3.0 hours = -ln(0.95) / 0.017. Based on this, the system controls the total time from cutting to entering the cool storage area or pre-processing station to not exceed 3.0 hours. If the expected dwell time exceeds the limit due to changes in the transportation or pre-processing queue status, the harvesting sequence is automatically adjusted or the cutting of new plots is suspended.

[0028] Example 7 See Figure 3 and Figure 5 This embodiment provides a batch harvesting collaborative system for implementing the aforementioned method, including a planning module, a sensing and prediction module, a scheduling and solving module, an intervention and control module, and a collaborative execution module; The planning module is used to acquire historical meteorological data, plot spatial data, and harvesting, transportation, and preprocessing capacity data, and to generate a baseline batch planting and production schedule. Preferably, the planning module is deployed on a farm management server, industrial control host, or central computing platform. The perception and prediction module is connected to distributed IoT monitoring nodes, drones or ground non-destructive testing terminals and laboratory testing terminals to establish or update quality index prediction models and output the actual optimal harvesting time window and time window drift for each target plot. The scheduling solution module receives the baseline batch planting and production plan, time window identification results and resource capacity data, and generates a batch harvesting scheduling scheme. The intervention control module is connected to the irrigation, shading or environmental regulation actuators and is used to implement microclimate intervention on the target plot and trigger closed-loop update when the total scheduling loss exceeds the preset threshold or the target plot exceeds the allowable harvesting deviation. The collaborative execution module is communicatively connected to the harvesting terminal, the transportation terminal, and the preprocessing terminal. It is used to generate harvesting operation instructions and control the maximum safe residence time after harvesting, and write the maximum safe residence time threshold after harvesting into the operation instructions.

[0029] The above description is merely a preferred embodiment of the present invention. Any equivalent substitutions, modifications, or improvements made by those skilled in the art to the data sources, parameter ranges, model forms, execution mechanism forms, and scheduling algorithm implementation methods without departing from the spirit and essence of the present invention should fall within the protection scope of the present invention.

Claims

1. A method for predicting harvest window and scheduling staggered harvesting to ensure stable quality of oxalic acid, characterized in that, Includes the following steps: S1. Before planting, obtain historical meteorological data, plot spatial data, and harvesting, transportation and pre-processing capacity data of the target planting area, determine the target harvesting season window, and generate a benchmark batch planting and production schedule based on the daily effective processing capacity. S2. During the crop growth cycle, continuously collect physiological development data representing the physiological development status of the plant, stress data representing the degree of environmental stress, and non-destructive testing data representing changes in canopy components. Obtain samples according to preset sampling rules for laboratory testing. Pair and calibrate the laboratory test results with the synchronously collected multi-source data to establish or update the quality index prediction model. Use the quality index prediction model to predict the oxalic acid-related quality score of each target plot in the future preset time period, and determine the actual optimal harvest time window of each target plot and the time window drift amount relative to the benchmark batch planting and production plan. S3. Under the constraints of daily effective processing capacity, transportation capacity and preprocessing capacity, construct a batch harvesting scheduling model with different loss functions for early harvesting and delayed harvesting, and solve for the harvesting date, harvesting order and resource allocation scheme of each target plot. S4. When the total scheduling loss exceeds the preset threshold and / or the planned harvest date of a target plot deviates from the center of its actual optimal harvest time window by more than the allowable harvest deviation, microclimate intervention is implemented for the corresponding target plot; after intervention, data is re-collected, the actual optimal harvest time window is updated, and scheduling solution is executed again. S5. Generate harvesting operation instructions based on the solution results, and calculate the maximum safe retention time threshold after harvesting based on ambient temperature, light conditions and post-harvest quality decay model. Control the total time from cutting to entering the cool storage area or pre-treatment station within the maximum safe retention time threshold after harvesting.

2. The method according to claim 1, characterized in that, In S1, seasonal periods characterized by high sunshine, strong ultraviolet radiation, low precipitation, and large diurnal temperature range are identified based on multi-year historical meteorological data as the target harvesting season window; the effective daily processing capacity is the minimum value among the area processing capacity corresponding to the harvesting capacity, transportation capacity, and pre-treatment capacity, wherein the harvesting capacity is calculated according to the following formula: Sh = N × v × Tsafe × eta; In the formula, Sh is the daily harvesting capacity, N is the number of harvesting operation units, v is the harvesting area per unit time of a single harvesting operation unit, Tsafe is the length of the safe operation period, and eta is the equipment utilization coefficient.

3. The method according to claim 1, characterized in that, S1 also includes spatial allocation of different batches of crops based on the digital elevation model of the target planting area, slope aspect, slope position, soil water retention and heat accumulation differences; wherein, batches expected to develop slowly or allowed to be appropriately delayed are allocated to plots with low heat accumulation and high water retention, and batches expected to mature faster are allocated to plots with high heat accumulation, strong sunlight or fast evaporation.

4. The method according to claim 1, characterized in that, The physiological development data in S2 includes at least one or more of the following: effective accumulated temperature, number of growing days, plant height, number of branches, canopy coverage, or leaf area index; the stress data includes at least one or more of the following: light intensity, ultraviolet radiation intensity, soil moisture content, soil tension, air temperature and humidity, leaf temperature, wind speed, or vapor pressure difference; the non-destructive testing data includes at least one or more of the following: canopy multispectral data, canopy hyperspectral data, canopy thermal infrared data, or vegetation index data; the laboratory testing data includes at least one or more of the following: oxalic acid content (CA), oxalic acid to degradation product ratio (R), and effective component yield per unit area (Y); the laboratory testing employs high-performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS).

5. The method according to claim 4, characterized in that, The canopy hyperspectral data from the non-destructive testing data specifically extracts reflectance data containing characteristic bands of 230nm and 280nm; A comprehensive quality score Q(t) is constructed, and the set of dates that satisfy Q(t) ≥ theta × Qmax is determined as the actual optimal harvesting time window, where Qmax is the maximum quality score within the preset prediction period, and theta is the time window threshold coefficient; Q(t) is calculated by the following formula: Q(t) = w1 × Norm[CA(t)] + w2 × Norm[R(t)] + w3 × Norm[Y(t)]; In the formula, w1, w2, and w3 are weights and w1+w2+w3=1; the difference between the center date tcenter of the time window and the planned harvest date tplanned in the baseline batch planting and production schedule is defined as the time window drift: Deltat=tcenter-tplanned.

6. The method according to claim 1, characterized in that, In S3, let x(i,t) represent the decision variable for whether plot i is harvested on date t. The scheduling model satisfies the constraints that each plot is harvested only once, the daily harvested area does not exceed the daily effective processing capacity, the daily transportation volume does not exceed Mmax, and the daily preprocessing volume does not exceed Pmax. The loss Li(t) when plot i is harvested on date t is... When t≤topt,i, Li(t)=alpha_i×(topt,it); When t>topt,i, Li(t)=beta_i×[exp(lambda_i×(t-topt,i))-1], Where topt,i is the optimal harvesting date for plot i, and beta_i is greater than alpha_i; the goal of solving the batch harvesting scheduling model is to minimize the total loss.

7. The method according to claim 1, characterized in that, In S4, measures such as water replenishment, shading, cooling, or reduction of stress intensity are implemented for target plots that mature too early, while measures such as water control, increased light exposure, or increased stress intensity are implemented for target plots that mature too late. The microclimate intervention lasts for 2-5 days. During the intervention, soil moisture content, leaf temperature, light conditions, and non-destructive testing data of the canopy are re-collected according to a retesting cycle of 24-48 hours, and the actual optimal harvest time window is updated. To avoid irreversible damage caused by excessive intervention, safety boundaries for soil moisture content, leaf temperature, and environmental stress intensity are pre-set and monitored in real time during the intervention process.

8. The method according to claim 1, characterized in that, In S5, the maximum safe retention time threshold Tlimit after harvest is calculated based on the post-harvest quality decay model; the oxalic acid retention rate of the sample at different time points is determined in advance through post-harvest retention experiments under different temperature, light, and stacking conditions, and the decay rate constant k is obtained by fitting; when the post-harvest quality change is represented by a first-order decay model, we have: C(t) = C0 × exp(-k × t); Tlimit = -ln(theta_p) / k; In the formula, C0 is the initial oxalic acid content after harvest, C(t) is the oxalic acid content after time t, k is the decay rate constant, and theta_p is the allowable retention rate. When the on-site queuing information or transportation congestion prediction indicates that the expected retention time will exceed Tlimit, the system will automatically adjust the harvesting order, reduce the priority of subsequent plot cutting, or suspend the cutting of new plots.

9. A batch harvesting coordination system for implementing the method according to any one of claims 1-8, characterized in that, include: The planning module is used to acquire historical meteorological data, plot spatial data, and harvesting, transportation, and preprocessing capacity data, and generate a baseline batch planting and production schedule. The perception and prediction module is used to collect physiological development data, stress data and non-destructive testing data, and establish or update the quality index prediction model based on the pairing and calibration of laboratory test results and multi-source data, and output the actual optimal harvesting time window and time window drift for each target plot. The scheduling solution module is used to generate a batch harvesting scheduling scheme based on an asymmetric penalty mechanism under the constraints of daily effective processing capacity, transportation capacity and preprocessing capacity. The intervention control module is used to implement microclimate intervention and trigger closed-loop update for the target plot when the total scheduling loss exceeds the preset threshold or the target plot exceeds the allowable harvesting deviation. The collaborative execution module is used to generate harvesting operation instructions and control the maximum safe dwell time after harvesting.

10. The batch harvesting coordination system according to claim 9, characterized in that, The sensing and prediction module is communicatively connected to distributed IoT monitoring nodes, drones or ground non-destructive testing terminals and laboratory testing terminals; the intervention control module is communicatively connected to irrigation, shading or environmental regulation actuators; the collaborative execution module is communicatively connected to harvesting operation terminals, transportation terminals and pre-processing terminals, and writes the maximum safe retention time threshold after harvesting into the harvesting operation instructions.

Images