Intelligent irrigation automatic execution system and method based on centralized control architecture

Abstract

The invention belongs to the technical field of agricultural irrigation, and discloses an intelligent irrigation automatic execution system and method based on a centralized control architecture. The method comprises the following steps: collecting sensing data of an irrigation area, and inputting the sensing data into a pre-constructed irrigation strategy mapping table to obtain an irrigation strategy of the irrigation area; real-time sensing data is obtained and analyzed, a total state coefficient is generated, whether the irrigation amount is reasonable or not is judged based on the total state coefficient, and an adjustment strategy is generated when the irrigation amount is unreasonable; adjusting the irrigation amount based on the generated adjustment strategy; distributing the adjusted irrigation volume to each sub-region, and executing irrigation operation; the abnormal coefficient of each sub-region is calculated according to the abnormal values of the real-time dimension and the historical dimension of each sub-region in the irrigation region, accurate monitoring and abnormal evaluation of the growth and soil cooperation state of each sub-region can be realized, and the sub-regions of the irrigation region can be sorted according to the abnormal coefficients, so that the irrigation strategies are dynamically increased and decreased, and the irrigation efficiency is improved. And resource waste caused by traditional uniform irrigation is avoided.

Description

Technical Field

[0001] This invention relates to the field of agricultural irrigation technology, and more specifically, to a smart irrigation automatic execution system and method based on a centralized control architecture. Background Technology

[0002] With the global water shortage becoming increasingly severe, improving irrigation efficiency in agriculture, a core water-consuming sector, has become a crucial issue for ensuring food security and sustainable agricultural development. Traditional manual and timed irrigation methods rely on experience and judgment, resulting in significant water waste, insufficient irrigation precision, and weak crop growth protection, making them unsuitable for the demands of large-scale, precision-based modern agricultural production. Against this backdrop, smart irrigation technology, leveraging the Internet of Things, sensors, and artificial intelligence, is gradually becoming a core solution to address the pain points of agricultural irrigation.

[0003] Currently, most smart irrigation systems on the market adopt a distributed control architecture, with sensors and execution units deployed independently in each irrigation sub-area. Irrigation actions are triggered by single-point monitoring data. Relying solely on single-point monitoring data can easily lead to imbalances in irrigation accuracy. Moreover, existing irrigation strategies are mostly based on static thresholds or historical experience, and can only respond to single environmental parameters (such as soil moisture). They cannot be dynamically adjusted in multiple dimensions by combining crop growth status and weather forecast data, resulting in low efficiency. In the irrigation allocation stage, most adopt a uniform allocation method without considering the collaborative differences between different sub-areas, leading to a mismatch between resource allocation and actual needs.

[0004] In view of this, the present invention proposes a smart irrigation automatic execution system and method based on a centralized control architecture to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: A smart irrigation automatic execution method based on a centralized control architecture, the method includes: Collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table to obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. Real-time sensing data is acquired during the irrigation cycle and recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and adjustment strategies are generated for irrigation cycle time points where the irrigation amount is unreasonable. Based on the generated adjustment strategies, the irrigation amount for irrigation cycles with unreasonable irrigation amounts is adjusted. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The adjusted irrigation amount is recorded as the actual irrigation amount. Based on the real-time sensing data of each sub-region within the irrigation cycle, the coordination index of each sub-region is calculated. Based on the coordination index of all sub-regions within the irrigation area, the anomaly coefficient of each sub-region within the irrigation area is calculated. Based on the anomaly coefficient, the actual irrigation amount is allocated to each sub-region, and irrigation operations are performed.

[0006] Furthermore, the method for constructing the irrigation strategy mapping table is as follows: Collect multiple sets of sensing data corresponding to the irrigation area under historical time series and label them as historical sensing data; the sensing data includes soil sensing data, meteorological sensing data and growth status sensing data; simultaneously collect irrigation strategies and irrigation effect data under each set of historical sensing data, and combine each set of historical sensing data with the corresponding irrigation strategies and irrigation effect data to form multiple sets of sensing-strategy-effect data; build a dataset based on multiple sets of sensing-strategy-effect data; Each set of historical sensing data in the dataset is standardized, and the standardized sets of historical sensing data are then fused to form corresponding fused feature vectors. K-means clustering is used to cluster all fused feature vectors to generate multiple sets of sensing state clusters, and a unique discrete label is assigned to each set of sensing state clusters. Based on discrete labels, each group of perception-policy-effect data in the dataset is integrated into perception state cluster discrete label-policy-effect data. Based on all perception state cluster discrete label-policy-effect data, a full dataset is constructed. The K-means clustering method is used to perform secondary clustering on the full dataset to obtain multiple policy-effect clusters. At the same time, each policy-effect cluster is bound to a corresponding irrigation policy label and irrigation effect score. The frequency of occurrence of discrete labels for each perception state cluster in each strategy effect cluster is counted. The frequency of occurrence of discrete labels for each perception state cluster is compared with a preset frequency threshold. Discrete labels for perception state clusters with a frequency greater than the frequency threshold are selected and recorded as high-frequency discrete labels. Each high-frequency discrete label is bound to the corresponding irrigation strategy label and irrigation effect score to form multiple sets of association pairs, and the contribution of each set of association pairs is calculated. Based on the multiple sets of association pairs and the corresponding contribution, an irrigation strategy mapping table is constructed.

[0007] Furthermore, the method for obtaining the irrigation strategy for the irrigated area is as follows: Acquire the sensing data of the irrigation area and record it as the latest sensing data. Input the latest sensing data into the irrigation strategy mapping table, match all corresponding association pairs, and select the irrigation strategy label corresponding to the association pair with the highest contribution as the irrigation strategy of the irrigation area.

[0008] Furthermore, the method for calculating contribution is as follows: The frequency of occurrence of the high-frequency discrete label corresponding to the association pair is divided by the total frequency of occurrence of the discrete label of the perception state cluster in the corresponding strategy effect cluster to obtain the occurrence frequency of the corresponding association pair, which is recorded as the first contribution. The mutual information between the high-frequency discrete label corresponding to the association pair and the corresponding irrigation effect score is obtained and recorded as the second contribution. The irrigation effect score corresponding to the association pair is standardized to obtain the third contribution. The first contribution, the second contribution, and the third contribution are multiplied together and then normalized to obtain the contribution.

[0009] Furthermore, the method for determining whether the irrigation amount at a given time point in the irrigation cycle is reasonable is as follows: The irrigation time point with unreasonable irrigation amount is denoted as irrigation time point Q, the irrigation time point preceding irrigation time point Q is denoted as irrigation time point E, and the time between irrigation time point E and irrigation time point Q is denoted as one irrigation cycle. Real-time sensing data includes real-time soil sensing data, real-time meteorological sensing data, and real-time growth status sensing data. The real-time sensing data within the irrigation cycle is analyzed to obtain the growth status coefficient and soil status coefficient of irrigation time point Q. The average of the growth status coefficient and the soil status coefficient is used to calculate the total status coefficient of irrigation time point Q. Based on the total status coefficient, a reasonable threshold range for irrigation amount at irrigation time point Q is obtained. If the set irrigation amount at irrigation time point Q is not within the corresponding reasonable threshold range, the irrigation amount at irrigation time point Q is judged to be unreasonable.

[0010] Furthermore, the calculation methods for the growth state coefficient and the soil state coefficient are as follows: Acquire real-time growth status sensing data at irrigation time point Q, and generate growth status value at irrigation time point Q based on the real-time growth status sensing data at irrigation time point Q; acquire real-time meteorological sensing data at irrigation time point Q, generate growth correction coefficient based on the real-time meteorological sensing data at irrigation time point Q, add 1 to the growth correction coefficient, multiply by the growth status value, and then perform normalization to obtain the growth status coefficient at irrigation time point Q. Real-time soil sensing data for irrigation time points E and Q are acquired. Soil state values ​​for irrigation time points E and Q are calculated based on the real-time soil sensing data and denoted as soil state value E and soil state value Q, respectively. The difference between soil state value Q and soil state value E is denoted as the deviation value. Based on real-time meteorological sensing data acquired during the irrigation cycle, the corresponding deviation standard value for the irrigation cycle is obtained. The deviation value is divided by the deviation standard value to obtain the corresponding deviation condition value. Soil state value Q is multiplied by the deviation condition value and then normalized to obtain the soil state coefficient for irrigation time point Q.

[0011] Furthermore, the adjustment strategies include strategies to increase irrigation volume and strategies to decrease irrigation volume. The two endpoints of the reasonable irrigation volume threshold range are denoted as the reasonable threshold Y and the reasonable threshold Z, and the set irrigation volume is denoted as the expected irrigation volume. If the expected irrigation volume is less than the reasonable threshold Y, an strategy to increase irrigation volume is generated; if the expected irrigation volume is greater than the reasonable threshold Z, a strategy to decrease irrigation volume is generated. The next irrigation time point corresponding to the irrigation cycle time point is denoted as irrigation time point S. Forecast meteorological sensing data between irrigation time point Q and irrigation time point S is obtained, and combined with real-time meteorological sensing data of irrigation time point Q for comprehensive analysis, an irrigation volume correction factor is generated. When an strategy to increase irrigation volume is generated, the reasonable threshold Y and reasonable threshold Z are averaged to obtain the standard irrigation volume. The expected irrigation volume is subtracted from the standard irrigation volume to obtain the initial increase, and the initial increase is multiplied by the irrigation volume correction factor to obtain the final increase, denoted as the irrigation increase. When a strategy to decrease irrigation volume is generated, the expected irrigation volume is subtracted from the standard irrigation volume to obtain the initial decrease, and the initial decrease is divided by the irrigation volume correction factor to obtain the final decrease, denoted as the irrigation decrease.

[0012] Furthermore, the method for obtaining the irrigation amount correction factor is as follows: Based on the acquired forecast meteorological sensing data, the data is input into the Penman-Monteith equation to obtain the predicted total evapotranspiration between irrigation time point Q and irrigation time point S. Simultaneously, all historical evapotranspiration between irrigation time point Q and irrigation time point S are acquired. The mean evapotranspiration QS is calculated by averaging all historical evapotranspiration. The predicted total evapotranspiration is then divided by the mean evapotranspiration QS and normalized to obtain the future meteorological-driven stress index. Based on the real-time meteorological sensing data of irrigation time point Q, the actual evapotranspiration at irrigation time point Q is obtained. Simultaneously, all historical evapotranspiration at irrigation time point Q are acquired. The mean evapotranspiration Q is calculated by averaging all historical evapotranspiration. The actual evapotranspiration is then divided by the mean evapotranspiration Q and normalized to obtain the real-time water deficit index. The real-time water deficit index and the future meteorological-driven stress index are weighted and summed to obtain the irrigation amount correction factor.

[0013] Furthermore, the method for allocating the actual irrigation amount to each sub-region based on the anomaly coefficient is as follows: The irrigation area is divided into multiple sub-regions. Growth status and soil status values ​​are obtained for each sub-region. The mean of these values ​​is calculated to obtain the synergy index for each sub-region. Real-time synergy indices for all sub-regions at irrigation time point Q are obtained, and their mean values ​​are calculated to obtain the real-time synergy index mean. The difference between the real-time synergy index of each sub-region and its mean value is used as the first outlier. Real-time synergy indices for each sub-region between irrigation time points E and Q are obtained, and a real-time synergy index function is constructed for each sub-region. The integral of the real-time synergy index function for each sub-region over the period from irrigation time point E to irrigation time point Q is calculated to obtain the historical cumulative value for each sub-region. The mean of the historical cumulative values ​​for all sub-regions between irrigation time points E and Q is calculated to obtain the historical cumulative mean. The difference between the historical cumulative value of each sub-region and its mean value is used as the second outlier. The first and second outliers are standardized and then summed to obtain the anomaly coefficient for each sub-region. Based on the anomaly coefficients, the actual irrigation amount is allocated to each sub-region for irrigation operations.

[0014] A smart irrigation automatic execution system based on a centralized control architecture, implementing the aforementioned smart irrigation automatic execution method based on a centralized control architecture, the system includes: The irrigation strategy generation module is used to collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table, and obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. The adjustment strategy generation module is used to acquire sensing data within the irrigation cycle in real time, which is recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and an adjustment strategy is generated for irrigation cycle time points where the irrigation amount is unreasonable. The adjustment strategy execution module is used to adjust the irrigation amount for irrigation cycles with unreasonable irrigation amounts based on the generated adjustment strategies. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The irrigation volume allocation module is used to record the adjusted irrigation volume as the actual irrigation volume, calculate the coordination index of each sub-region based on the real-time sensing data of each sub-region within the irrigation cycle, calculate the anomaly coefficient of each sub-region within the irrigation area based on the coordination index of all sub-regions within the irrigation area, and allocate the actual irrigation volume to each sub-region based on the anomaly coefficient to execute the irrigation operation.

[0015] The technical effects and advantages of the intelligent irrigation automatic execution system and method based on a centralized control architecture of this invention are as follows: This invention integrates multi-dimensional synchronous sensing data, historical irrigation strategies, and effect data. Through secondary clustering, it generates sensing state clusters and strategy effect clusters. Based on frequency proportion, mutual information, and effect score, it constructs a multi-dimensional contribution degree, thereby selecting an appropriate irrigation strategy. This not only improves irrigation accuracy and water resource utilization efficiency but also makes the decision-making interpretable and traceable through quantitative evaluation of contribution. It effectively solves the problems of imbalance between strategy universality and accuracy, insufficient dynamic response, and low decision-making transparency in existing technologies. While ensuring crop yield, it significantly reduces irrigation costs and has stronger scenario adaptability and continuous evolution capabilities.

[0016] This invention clarifies the quantitative conditions for triggering increase / decrease strategies by setting reasonable threshold ranges. It uses the average value of the range as the standard irrigation amount to ensure benchmark stability. Then, it combines the Penman-Monteith equation with historical data to calculate the future weather-driven stress index and the real-time water deficit index, generating a correction factor to dynamically adjust the irrigation amount. This invention achieves unambiguous and executable strategy triggering through clear threshold judgments, and adjusts by integrating future trends and current conditions through correction factors, ultimately keeping the irrigation amount within a reasonable range. This not only accurately matches the actual water needs of crops and reduces water waste, but also proactively responds to weather changes and anchors the current water situation, effectively improving irrigation efficiency and crop growth protection capabilities.

[0017] This invention divides the irrigation area into multiple sub-regions and generates a synergy index that synergistically reflects crop growth and soil condition. Based on the outliers in both real-time and historical dimensions of each sub-region within the irrigation area, anomaly coefficients are calculated for each sub-region. This enables precise monitoring and anomaly assessment of the growth and soil synergy status of each sub-region. Furthermore, it allows for the ranking of the sub-regions within the irrigation area based on the anomaly coefficients, thereby dynamically adjusting irrigation strategies. This avoids the resource waste of traditional uniform irrigation, making irrigation allocation more aligned with the actual needs of each sub-region and ensuring irrigation quality and the stability of crop growth. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module block diagram of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] In one embodiment, a method for automatically executing smart irrigation based on a centralized control architecture is disclosed, such as... Figure 1 As shown, the execution method mainly includes: Collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table to obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. Real-time sensing data is acquired during the irrigation cycle and recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and adjustment strategies are generated for irrigation cycle time points where the irrigation amount is unreasonable. Based on the generated adjustment strategies, the irrigation amount for irrigation cycles with unreasonable irrigation amounts is adjusted. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The adjusted irrigation amount is recorded as the actual irrigation amount. The anomaly coefficient of each sub-region within the irrigation area is calculated. Based on the anomaly coefficient and the adjustment strategy, the actual irrigation amount is allocated to each sub-region, and irrigation operations are performed.

[0021] Through the above scheme, this application first constructs an irrigation strategy mapping table based on historical sensing data, irrigation strategies, and irrigation effect data. Then, it collects sensing data from the irrigation area and inputs the corresponding sensing data into the corresponding irrigation strategy mapping table. The irrigation strategy for the irrigation area is obtained from the mapping table, and the irrigation cycle and irrigation amount for each time point in the irrigation cycle are set according to the obtained irrigation strategy. During the irrigation cycle, the system acquires real-time sensing data (soil sensing data, meteorological sensing data, and growth status sensing data) and records it as real-time sensing data. Based on the analysis of this real-time sensing data, a total state coefficient for the corresponding irrigation time point is generated, and the corresponding state coefficient is used to determine the irrigation strategy mapping table. If the irrigation amount is deemed unreasonable, the system generates an adjustment strategy to adjust the irrigation amount. This adjustment includes either increasing or decreasing the irrigation amount. Specifically, the system dynamically corrects the planned irrigation amount by combining pre-set reasonable threshold ranges, meteorological correction factors, and other parameters. If an increase in irrigation is needed, the final increment is calculated based on the difference between the standard and estimated irrigation amounts, combined with the correction factor. If a decrease in irrigation is needed, the final reduction is calculated based on the difference between the estimated and standard irrigation amounts, combined with the correction factor, thus generating the actual irrigation amount adapted to real-time needs. Finally, the system finely allocates the actual irrigation amount based on the anomaly coefficients of each sub-region within the irrigation area, ultimately achieving automated irrigation operations for each sub-region through centralized control.

[0022] The method for obtaining the irrigation strategy of the irrigation area is as follows: acquire the sensing data of the irrigation area, record it as the latest sensing data, input the latest sensing data into the irrigation strategy mapping table, match all corresponding association pairs, select the irrigation strategy label corresponding to the association pair with the highest contribution among the association pairs, and use it as the irrigation strategy of the irrigation area. The method for constructing the irrigation strategy mapping table is as follows: collect multiple sets of sensing data corresponding to the irrigation area under historical time series and label them as historical sensing data; the sensing data includes soil sensing data, meteorological sensing data, and growth status sensing data; simultaneously collect irrigation strategies and irrigation effect data under each set of historical sensing data, and combine each set of historical sensing data with the corresponding irrigation strategies and irrigation effect data to form multiple sets of sensing-strategy-effect data; based on the multiple sets of sensing-strategy-effect data, construct a dataset; Each set of historical sensing data in the dataset is standardized, and then the standardized sets are fused to form corresponding fused feature vectors. K-means clustering is used to cluster all fused feature vectors, generating multiple sensing state clusters, and a unique discrete label is assigned to each cluster. Based on these discrete labels, each sensing-policy-effect data set is integrated into sensing state cluster discrete label-policy-effect data. A full dataset is constructed based on all sensing state cluster discrete label-policy-effect data. K-means clustering is then used to perform binary search on the full dataset. Secondary clustering yields multiple policy effect clusters, and each policy effect cluster is associated with a corresponding irrigation policy label and irrigation effect score. The frequency of occurrence of discrete labels for each perception state cluster within each policy effect cluster is statistically analyzed. The frequency of each perception state cluster's discrete label is compared with a preset frequency threshold, and discrete labels for perception state clusters with a frequency greater than the threshold are selected and designated as high-frequency discrete labels. Each high-frequency discrete label is then associated with its corresponding irrigation policy label and irrigation effect score, forming multiple association pairs, and the contribution of each association pair is calculated. Based on these association pairs and their corresponding contributions, an irrigation policy mapping table is constructed. The contribution is calculated as follows: the frequency of occurrence of the high-frequency discrete label corresponding to the association pair is divided by the total frequency of occurrence of the discrete label of the perception state cluster in the corresponding strategy effect cluster to obtain the occurrence frequency of the corresponding association pair, which is recorded as the first contribution; the mutual information between the high-frequency discrete label corresponding to the association pair and the corresponding irrigation effect score is obtained, which is recorded as the second contribution; the irrigation effect score corresponding to the association pair is standardized to obtain the third contribution; the first contribution, the second contribution, and the third contribution are multiplied together and then normalized to obtain the contribution.

[0023] The above scheme provides a specific method for determining irrigation strategies for irrigation areas. First, an irrigation strategy mapping table is constructed. Then, multiple sets of sensing data corresponding to the irrigation area are collected from historical time series and labeled as historical sensing data. This historical sensing data includes soil sensing data, meteorological sensing data, and growth status sensing data. Soil sensing data includes soil electrical conductivity, soil moisture content, and soil porosity; meteorological sensing data includes wind speed, light intensity, temperature, and humidity; and growth status sensing data includes leaf curl, canopy temperature, stomatal conductance, and leaf water potential. Simultaneously, irrigation strategies and irrigation effect data are collected for each set of historical sensing data. Irrigation strategies include the irrigation cycle and the irrigation amount at each time point in the irrigation cycle. Irrigation effect data includes yield increase rate, water saving rate, etc. Each set of historical sensing data is combined with the corresponding irrigation strategy and irrigation effect data to form multiple sets of sensing-strategy-effect data. Each set of sensing-strategy-effect data corresponds to specific sensing data, historical irrigation strategies, and irrigation effect data. A dataset is constructed based on these multiple sets of sensing-strategy-effect data. Each set of historical sensing data is standardized, and the standardized sets are then fused to form corresponding fusion feature vectors. For example, data such as soil electrical conductivity, soil moisture content, evaporation, light intensity, temperature, leaf water potential, canopy temperature, and plant height are standardized and normalized to the 0-1 range. The data are then fused into corresponding fusion feature vectors. K-means clustering is used to cluster all fusion feature vectors, generating multiple perception state clusters, and each perception state cluster is assigned a unique discrete label. Based on these discrete labels, each perception-policy-effect data set in the dataset is integrated into perception state cluster discrete label-policy-effect data. Based on all perception state cluster discrete label-policy-effect data, a full dataset is constructed. K-means clustering is then used to perform secondary clustering on the full dataset, resulting in multiple policy-effect clusters. Each policy-effect cluster represents a complete association of "perception state → irrigation policy → irrigation effect," and each policy-effect cluster is bound to a corresponding irrigation policy label and irrigation... The effectiveness score is calculated by manually setting irrigation strategy labels based on the corresponding irrigation strategies, and then standardizing the irrigation effectiveness score by weighting and summing factors such as yield increase rate and water saving rate in the irrigation effectiveness data. The specific weights can be set based on human experience. The frequency of occurrence of discrete labels for each perception state cluster in each strategy effectiveness cluster is counted. The frequency of occurrence of discrete labels for each perception state cluster is compared with a preset frequency threshold, which can be obtained based on the professional knowledge and experience of personnel in this field. Discrete labels for perception state clusters with a frequency greater than the frequency threshold are selected and recorded as high-frequency discrete labels. Each high-frequency discrete label is then bound to the corresponding irrigation strategy label and irrigation effectiveness score to form multiple sets of association pairs.For example, a strategy effect cluster P001 (with its corresponding irrigation strategy label BQ1 and irrigation effect score PF1) contains multiple discrete labels for perception state clusters (such as perception state cluster discrete labels S001, S002, S003, S004, S005, etc.). After screening, it is found that the frequency of S003 and S004 is greater than a preset frequency threshold, so S003 and S004 are recorded as high-frequency discrete labels. The irrigation strategy label and irrigation effect score corresponding to the strategy effect cluster P001 are bound to each other, thus obtaining two association pairs (S003-BQ1-PF1 and S004-BQ1-PF1). Similarly, a strategy effect cluster P002 (with its corresponding irrigation strategy label BQ2 and irrigation effect score PF1) contains multiple discrete labels for perception state clusters (such as perception state cluster discrete labels S001, S002, S003, S004, S005, etc.). The irrigation effect score (PF2) contains multiple discrete labels for the perception state cluster (such as discrete labels S001, S002, S003, S006, S007, etc.). After screening, it was found that the frequency of S003, S005, and S001 is greater than the preset frequency threshold. Therefore, S003, S005, and S001 are recorded as high-frequency discrete labels. They are then bound to the irrigation strategy label and irrigation effect score corresponding to the strategy effect cluster P002, respectively, to obtain three association pairs (S003-BQ2-PF2, S005-BQ2-PF2, and S001-BQ2-PF2). The contribution of each association pair is calculated. The specific calculation method for the contribution is as follows: the high-frequency discrete labels corresponding to the association pair are... The frequency of a label occurrence is divided by the total frequency of discrete labels in the perceived state cluster within the corresponding strategy effect cluster to obtain the occurrence frequency of the corresponding association pair, which is denoted as the first contribution. For example, if the strategy effect cluster P001 contains a total of 490 discrete labels, and the high-frequency discrete label corresponding to the association pair S003-BQ1-PF1 occurs 280 times, then its first contribution is 280 / 490. The first contribution reflects the historical association frequency between the perceived state and the irrigation strategy in the association pair; a higher value indicates that the perception state has been historically more commonly bound to the strategy, and the strategy has stronger universality. The mutual information between the high-frequency discrete label corresponding to the association pair and the corresponding irrigation effect score is obtained and denoted as the second contribution. Mutual information can be obtained by calculating the high-frequency discrete label and the irrigation effect score. The first contribution is obtained by the relative entropy (such as KL divergence) between the joint probability distribution and the respective marginal probability distribution of irrigation effect scores. The specific calculation method is a current technique in the field of information theory, which will not be described in detail here. The second contribution reflects the closeness of the correlation between the perceived state and the irrigation effect score. The higher the value, the more stable the effect of the corresponding strategy and the stronger the strategy adaptability under the perceived state. After standardizing the irrigation effect scores corresponding to the correlation pairs, the third contribution is obtained. The third contribution reflects the actual effect of the irrigation strategy. The higher the value, the better the strategy has performed in history and the greater the value of the strategy. Therefore, the first, second, and third contribution are multiplied together and then normalized to obtain the contribution score. The larger the contribution score, the better the corresponding irrigation strategy.Based on multiple sets of association pairs and their corresponding contributions, an irrigation strategy mapping table is constructed. Finally, the perceived data of the irrigation area is obtained and recorded as the latest perceived data. This latest perceived data is input into the irrigation strategy mapping table to match the corresponding association pairs. The matching method can be based on cosine similarity. For example, the latest perceived data is standardized and set as vector B. The standard fusion feature vector within the discrete label of the perceived state cluster corresponding to the association pair is denoted as vector A. The standard fusion feature vector can be obtained based on the mean of all fusion feature vectors of the discrete label of the perceived state cluster. The cosine similarity between vector B and all vectors A is calculated, and the association pair corresponding to vector A with the highest similarity is selected for matching. Since there may be multiple irrigation strategy labels corresponding to the matched association pairs, such as association pair S003-BQ1-PF1 and association pair S003-BQ2-PF... 2. The two sets of perceptual state clusters share the same discrete label, but their bound irrigation strategy labels differ. Therefore, the irrigation strategy label corresponding to the association pair with the highest contribution among all association pairs is selected as the irrigation strategy for the irrigation area. By integrating multi-dimensional synchronous perceptual data, historical irrigation strategies, and effect data, perceptual state clusters and strategy effect clusters are generated through secondary clustering. Multi-dimensional contribution is constructed based on frequency proportion, mutual information, and effect score, thereby selecting a suitable irrigation strategy. This not only improves irrigation accuracy and water resource utilization efficiency but also makes the decision-making interpretable and traceable through quantitative evaluation of contribution. It effectively solves the problems of imbalance between strategy universality and accuracy, insufficient dynamic response, and low decision-making transparency in existing technologies. While ensuring crop yield, it significantly reduces irrigation costs and has stronger scenario adaptability and continuous evolution capabilities.

[0024] The method for determining whether the irrigation amount at a given time point in an irrigation cycle is reasonable is as follows: Irrigation time point Q is designated as the time point in an irrigation cycle where the irrigation amount is unreasonable; the preceding irrigation cycle time point E is designated as the time point in an irrigation cycle; the time between irrigation time point E and irrigation time point Q is defined as one irrigation cycle. Real-time sensing data is acquired, including real-time soil sensing data, real-time meteorological sensing data, and real-time growth status sensing data. The real-time sensing data within the irrigation cycle is analyzed to obtain the growth status coefficient and soil status coefficient at irrigation time point Q. The average of the growth status coefficient and the soil status coefficient is used to calculate the total status coefficient of irrigation time point Q. Based on the total status coefficient, a reasonable threshold range for irrigation amount at irrigation time point Q is obtained. If the set irrigation amount at irrigation time point Q is not within the corresponding reasonable threshold range, then the irrigation amount at irrigation time point Q is determined to be unreasonable. The calculation methods for the growth state coefficient and soil state coefficient are as follows: Real-time growth state sensing data for irrigation time point Q is acquired, and a growth state value for irrigation time point Q is generated based on this data. Real-time meteorological sensing data for irrigation time point Q is acquired, and a growth correction coefficient is generated based on this data. The growth correction coefficient is then incremented by 1, multiplied by the growth state value, and normalized to obtain the growth state coefficient for irrigation time point Q. Real-time soil sensing data for irrigation time points E and Q is acquired, and soil state values ​​for irrigation time points E and Q are calculated based on this data, denoted as soil state value E and soil state value Q, respectively. The difference between soil state value Q and soil state value E is recorded as the deviation value. Based on the real-time meteorological sensing data acquired during the irrigation cycle, the corresponding deviation standard value for the irrigation cycle is obtained. The deviation value is divided by the deviation standard value to obtain the corresponding deviation condition value. The soil state value Q is multiplied by the deviation condition value and then normalized to obtain the soil state coefficient for irrigation time point Q.

[0025] The above technical solution provides a specific method for determining whether the irrigation amount at a given time point in an irrigation cycle is reasonable. First, the irrigation time point where the irrigation amount is unreasonable is recorded as irrigation time point Q. The irrigation time point preceding irrigation time point Q is recorded as irrigation time point E. The time between irrigation time point E and irrigation time point Q is recorded as one irrigation cycle. Real-time sensing data is acquired, including real-time soil sensing data, real-time meteorological sensing data, and real-time growth status sensing data. Real-time growth status sensing data for irrigation time point Q is acquired. Based on this data, a growth status value for irrigation time point Q is generated. The calculation method for the growth status value is as follows: the real-time growth status sensing data is divided into positive and negative values. The data affecting growth status are categorized into positive and negative impacts. Positive impacts are indicated by higher values, such as leaf curling and canopy temperature; negative impacts are indicated by lower values, such as stomatal conductance and leaf water potential. The data is normalized and then weighted to obtain a comprehensive value for positive and negative impacts. Finally, the comprehensive value for positive impacts is obtained by dividing the comprehensive value for positive impacts by the comprehensive value for negative impacts and then normalizing again. The weights of the sensing data can be determined based on the sensitivity of historical growth status sensing data to the physiological effects on the corresponding organisms. Sensing data with higher sensitivity has a higher weight. It is known that a larger growth status value indicates a more severe water shortage. Different meteorological sensing data can affect growth status parameters; for example, excessively high temperatures can lead to decreased enzyme activity and inhibited growth, while excessively high wind speeds can accelerate canopy water loss. Therefore, real-time meteorological sensing data at irrigation time point Q is obtained, and each meteorological sensing data point is quantified to obtain the corresponding quantified impact value. The quantified impact values ​​corresponding to each meteorological sensing data point can be normalized, while positive impacts (such as higher values ​​for temperature and wind speed) are normalized. Normalization is performed, and for negative impacts (such as lower humidity values ​​indicating higher impact), reverse normalization is used to map each meteorological sensing data into a quantified impact value in the 0-1 range. It can be seen that the larger the quantified impact value, the greater the impact on the growth state parameters. The growth correction coefficient is obtained by weighting and summing all quantified impact values. The weight of each quantified impact value can be determined according to the sensitivity of the corresponding meteorological sensing data to the physiological impact on organisms. The sensing data with higher sensitivity has a higher weight. Finally, the growth correction coefficient is added by 1, multiplied by the growth state value, and then normalized to obtain the growth state coefficient at irrigation time point Q. It can be seen that the larger the value of the growth state coefficient, the more severe the water shortage.In this way, growth status sensing data is accurately divided into two categories of indicators: positive and negative impacts. Weights are assigned based on their sensitivity to crop physiology. Then, real-time meteorological sensing data is integrated to generate correction coefficients. Finally, a quantifiable growth status coefficient is obtained to determine the water shortage situation. This not only reflects the true water stress state of crops more accurately and comprehensively, avoiding the one-sided misjudgment of a single indicator, but also achieves dynamic coupling between crop physiology and the real-time environment through meteorological correction. This improves the scientificity and accuracy of water shortage judgment and provides a reliable quantitative basis for the subsequent fine adjustment of irrigation amount. Real-time soil sensing data at irrigation time points E and Q are acquired, and soil state values ​​at these times are calculated. The calculation method is as follows: real-time soil sensing data is divided into positively and negatively impacted data. Positively impacted data, such as soil porosity, indicates a higher value and more severe water shortage. Negatively impacted data, such as soil conductivity and soil moisture content, indicates a lower value and more severe water shortage. The positively impacted data is normalized and then weighted to obtain a comprehensive positive soil state value. The negatively impacted data is then weighted to obtain a comprehensive positive soil state value. The negative soil impact comprehensive value is obtained by weighted summation after normalization. The positive soil impact comprehensive value is then divided by the negative soil impact comprehensive value and normalized again to obtain the soil state value. The weights of each positive and negative soil perception data point can be determined based on the sensitivity of historical soil perception data to the physiological effects on corresponding organisms; the higher the sensitivity of the perception data, the higher its weight. It can be seen that the larger the soil state value, the more severe the water shortage. The soil state value at irrigation time point E is denoted as soil state value E, and the soil state value at irrigation time point Q is denoted as soil state value Q. The difference between soil state value Q and soil state value E is... The value is denoted as the deviation value, which represents the actual fluctuation range of soil condition within an irrigation cycle. Based on real-time meteorological sensing data acquired during the irrigation cycle, the corresponding standard deviation value for that cycle is obtained. The method for obtaining the standard deviation value is as follows: the acquired real-time meteorological sensing data is input into the Penman-Monteith equation, which is the international standard method recommended by the FAO for calculating reference crop evapotranspiration. This is existing technology in the field of agricultural hydrology and will not be elaborated upon here. Real-time evapotranspiration is obtained, and then all real-time evapotranspiration values ​​within the irrigation cycle are summed to obtain the total evapotranspiration value, based on historical data. Cluster analysis was performed on the historical total evapotranspiration and corresponding historical deviation values ​​during the Shitong irrigation cycle to obtain the standard deviation value for the corresponding total evapotranspiration. The clustering method is an existing technology and will not be described in detail here. Then, the deviation value was divided by the standard deviation value to obtain the corresponding deviation status value. The deviation status value reflects the degree to which the soil state change exceeds the normal fluctuations driven by meteorology. The larger the value, the more abnormal the change and the higher the possibility of water shortage. Finally, the soil state value Q at irrigation time point Q was multiplied by the deviation status value and then normalized to obtain the soil state coefficient at irrigation time point Q. The soil state coefficient indicates the final comprehensive quantitative value of soil water shortage. The larger the value, the more severe the soil water shortage.In this way, soil sensing data is divided into positive and negative impact indicators and weighted and normalized by physiological sensitivity to accurately quantify soil state values. Then, real-time evapotranspiration is calculated based on the Penman-Monteith equation, and the deviation standard value driven by meteorology is determined by cluster analysis to determine whether the fluctuation of soil state exceeds the normal range driven by meteorology. Finally, a soil state coefficient that can be directly quantified is generated. This not only reflects the true water shortage status of the soil more accurately and dynamically, avoiding the one-sided misjudgment of a single indicator or fixed threshold, but also improves the scientificity and adaptability of the judgment through meteorological coupling correction, providing a reliable quantitative basis for the subsequent fine adjustment of irrigation amount. After obtaining the growth state coefficient and soil state coefficient, the average of the growth state coefficient and soil state coefficient is calculated to obtain the total state coefficient at irrigation time point Q. Based on the total state coefficient, a reasonable threshold range for irrigation amount at irrigation time point Q is obtained. This reasonable threshold range can be obtained by collecting a matching dataset of the total state coefficient and the corresponding optimal irrigation amount within historical irrigation cycles, and using cluster analysis to establish a corresponding mapping model. This maps the total state coefficient to the corresponding reasonable threshold range for irrigation amount in the mapping model. This is existing technology and will not be described in detail. If it is found that the irrigation amount set at irrigation time point Q is not within the corresponding reasonable threshold range, it indicates that the set irrigation amount is unreasonable. Therefore, an adjustment strategy is generated to adjust the irrigation amount.

[0026] The adjustment strategies include strategies to increase irrigation volume and strategies to decrease irrigation volume. The two endpoints of the reasonable irrigation volume threshold range are denoted as reasonable threshold Y and reasonable threshold Z, and the set irrigation volume is denoted as the expected irrigation volume. If the expected irrigation volume is less than reasonable threshold Y, an strategy to increase irrigation volume is generated; if the expected irrigation volume is greater than reasonable threshold Z, a strategy to decrease irrigation volume is generated. The next irrigation time point corresponding to the irrigation cycle time point is denoted as irrigation time point S. Forecast meteorological sensing data between irrigation time point Q and irrigation time point S is obtained, and combined with real-time meteorological sensing data of irrigation time point Q for comprehensive analysis, an irrigation volume correction factor is generated. When an strategy to increase irrigation volume is generated, the average of reasonable threshold Y and reasonable threshold Z is calculated to obtain the standard irrigation volume. The expected irrigation volume is subtracted from the standard irrigation volume to obtain the initial increase, and the initial increase is multiplied by the irrigation volume correction factor to obtain the final increase, denoted as the irrigation increase. When a strategy to decrease irrigation volume is generated, the expected irrigation volume is subtracted from the standard irrigation volume to obtain the initial decrease, and the initial decrease is divided by the irrigation volume correction factor to obtain the final decrease, denoted as the irrigation decrease. The method for obtaining the irrigation amount correction factor is as follows: Based on the acquired forecast meteorological sensing data, the data is input into the Penman-Monteith equation to obtain the predicted total evapotranspiration between irrigation time point Q and irrigation time point S. Simultaneously, all historical evapotranspiration between irrigation time point Q and irrigation time point S are acquired. The mean evapotranspiration QS is calculated by averaging all historical evapotranspiration. The predicted total evapotranspiration is divided by the mean evapotranspiration QS and then normalized to obtain the future meteorological-driven stress index. Based on the real-time meteorological sensing data of irrigation time point Q, the actual evapotranspiration at irrigation time point Q is obtained. Simultaneously, all historical evapotranspiration at irrigation time point Q is acquired. The mean evapotranspiration Q is calculated by averaging all historical evapotranspiration. The actual evapotranspiration is divided by the mean evapotranspiration Q and then normalized to obtain the real-time water deficit index. The real-time water deficit index and the future meteorological-driven stress index are weighted and summed to obtain the irrigation amount correction factor.

[0027] The above scheme provides a specific method for adjusting irrigation volume based on adjustment strategies. Firstly, the adjustment strategies include increasing irrigation volume and decreasing irrigation volume. The two endpoints of the reasonable irrigation volume threshold range are denoted as the reasonable threshold Y and the reasonable threshold Z. The set irrigation volume is denoted as the expected irrigation volume. If the expected irrigation volume is less than the reasonable threshold Y, it indicates that the set irrigation volume is insufficient to meet the current irrigation needs of the irrigation area, and an increase in irrigation volume is required, thus generating an increase in irrigation volume strategy. Conversely, if the expected irrigation volume is greater than the reasonable threshold Z, it indicates that the set irrigation volume exceeds the current irrigation needs of the irrigation area, and a decrease in irrigation volume is required to conserve water, thus generating a decrease in irrigation volume strategy. Specific adjustments need to be combined with meteorological sensing. The data undergoes comprehensive analysis and adjustment. For example, the next irrigation time point after irrigation time point Q is denoted as irrigation time point S. Forecast meteorological sensing data between irrigation time point Q and irrigation time point S is acquired. Combined with real-time meteorological sensing data of irrigation time point Q, a comprehensive analysis is performed to generate an irrigation amount correction factor. Based on the acquired forecast meteorological sensing data, it is input into the Penman-Monteith equation to obtain the expected total evapotranspiration between irrigation time point Q and irrigation time point S. At the same time, all historical evapotranspiration between irrigation time point Q and irrigation time point S is acquired. The mean evapotranspiration QS is obtained by averaging all historical evapotranspiration. The expected total evapotranspiration is divided by the mean evapotranspiration QS and then normalized to obtain the future meteorological driver. The stress index, specifically the future weather-driven stress index, provides a forward-looking assessment of the impact of meteorological conditions on crop water demand. A higher future weather-driven stress index indicates a greater degree of water deficit. Based on real-time meteorological data at irrigation time point Q, the actual evapotranspiration at Q is obtained, along with all historical evapotranspiration data for Q. The mean evapotranspiration Q is calculated by averaging all historical evapotranspiration data. The actual evapotranspiration is then divided by the mean evapotranspiration Q and normalized to obtain the real-time water deficit index. The real-time water deficit index provides an immediate assessment of the actual water deficit faced by crops. A higher value indicates that the current water consumption evapotranspiration of the crop is higher than the historical average for the same period, suggesting a more severe current water deficit. Therefore, the real-time water deficit index and the future weather-driven stress index are weighted and summed to obtain the irrigation amount correction factor. The weight of each index can be determined by field experiments combined with the experience of experts in this field. The larger the irrigation amount correction factor, the higher the degree of water shortage, and the smaller the irrigation amount correction factor, the lower the urgency of water shortage. Therefore, when generating an increased irrigation amount strategy, the average of the reasonable threshold Y and the reasonable threshold Z is used to calculate a standard irrigation amount. The expected irrigation amount is subtracted from the standard irrigation amount to obtain the initial increase. Then, the initial increase is multiplied by the irrigation amount correction factor to obtain the final increase, which is recorded as the irrigation increase. It can be seen that if the irrigation amount correction factor is larger, more irrigation amount needs to be increased to ensure irrigation demand.Similarly, when generating a strategy to reduce irrigation, the initial reduction is obtained by subtracting the standard irrigation from the expected irrigation. The final reduction is obtained by dividing the initial reduction by the irrigation correction factor, and is denoted as the irrigation reduction. It can be seen that the larger the irrigation correction factor, the higher the water demand of the crop. Therefore, when reducing the irrigation, the corresponding reduction in irrigation should also be less to ensure irrigation demand. Thus, the final reduction is obtained by dividing the initial reduction by the irrigation correction factor. This solution clarifies the quantitative conditions for triggering irrigation increase / decrease strategies by setting reasonable threshold ranges. The average value of these ranges is used as the standard irrigation amount to ensure baseline stability. Furthermore, by combining the Penman-Monteith equation with historical data, future weather-driven stress indices and real-time water deficit indices are calculated, generating correction factors to dynamically adjust irrigation amounts. This approach achieves both unambiguous and executable strategy triggering through clear threshold judgments and adjustments based on future trends and current conditions through correction factors. Ultimately, it ensures that irrigation amounts remain within a reasonable range, accurately matching actual crop water needs, reducing water waste, proactively responding to weather changes, and anchoring current water conditions, effectively improving irrigation efficiency and crop growth protection.

[0028] The method for allocating actual irrigation volume to various sub-regions based on anomaly coefficients is as follows: The irrigation area is divided into multiple sub-regions; the growth status value and soil status value of each sub-region are obtained; the average of the growth status value and soil status value is calculated to obtain the synergy index of each sub-region; the real-time synergy index of all sub-regions at irrigation time point Q is obtained; the average of the real-time synergy indices of all sub-regions is calculated to obtain the real-time synergy index mean; the difference between the real-time synergy index of each sub-region and the real-time synergy index mean is used as the first anomaly; the real-time synergy index of each sub-region between irrigation time point E and irrigation time point Q is obtained. The system constructs a real-time collaborative exponential function for each sub-region; calculates the integral of the real-time collaborative exponential function for each sub-region between irrigation time point E and irrigation time point Q to obtain the historical cumulative value for each sub-region; calculates the average of the historical cumulative values ​​for all sub-regions between irrigation time point E and irrigation time point Q to obtain the historical cumulative mean; the difference between the historical cumulative value and the historical cumulative mean for each sub-region is used as the second outlier; the first outlier and the second outlier are standardized and then added together to obtain the outlier coefficient for each sub-region; and the actual irrigation amount is allocated to each sub-region for irrigation operations based on the outlier coefficient for each sub-region.

[0029] The above technical solution provides a specific method for allocating actual irrigation volume to various sub-regions. Since the irrigated area is generally large, to utilize the irrigation volume more rationally, this solution divides the irrigated area into multiple sub-regions and obtains the growth status and soil status values ​​for each sub-region. The average of the growth status and soil status values ​​is used to calculate the synergy index for each sub-region. The synergy index indicates the water shortage status of each sub-region. Then, the real-time synergy index of all sub-regions at irrigation time point Q is obtained, and the average of the real-time synergy indices of all sub-regions is used to calculate the real-time synergy index mean. The real-time synergy index of each sub-region is then compared with the average of the real-time synergy indices. The difference is used as the first outlier. The first outlier indicates the real-time overall difference between a single sub-region and all sub-regions of the irrigated area at the irrigation time point. Since all sub-regions are in the same irrigation area, the closer the first outlier is to 0, the higher the level of synergy between crop growth and soil condition in that sub-region. The closer it is to the overall average level, the lower the probability of an anomaly. Conversely, the further the first outlier deviates from 0, the higher the probability of an anomaly. Then, the real-time synergy index of each sub-region between irrigation time point E and irrigation time point Q is obtained, and a real-time synergy index function for each sub-region is constructed. The real-time synergy index function of each sub-region is calculated from irrigation time point E to irrigation time point Q. Integrating between time points Q yields the historical cumulative value for each sub-region. The average historical cumulative value of all sub-regions between irrigation time point E and irrigation time point Q is calculated to obtain the historical cumulative mean. The difference between the historical cumulative value of each sub-region and the historical cumulative mean is used as the second outlier. The second outlier indicates the overall historical difference between a single sub-region and all sub-regions of the irrigated area within an irrigation cycle prior to the irrigation time point. Similarly, the further the second outlier deviates from 0, the further the cumulative performance of crop growth and soil condition synergy in that sub-region deviates from the overall average over a past period. Therefore, the first and second outliers are separated... After standardization, the values ​​are summed to obtain the anomaly coefficients for each sub-region. This allows for the assessment of anomalies in each sub-region based on historical data and the current status of irrigation data points. Then, irrigation volume is allocated rationally according to the anomalies in each sub-region. For example, when generating an increased irrigation strategy, the sub-regions are sorted from largest to smallest based on their anomaly coefficients, and more irrigation volume is allocated to the top-ranked sub-regions. Conversely, when generating a decreased irrigation strategy, the sub-regions are sorted from smallest to largest based on their anomaly coefficients, and more irrigation volume is reduced for the top-ranked sub-regions. This achieves a rational allocation of irrigation volume, ensuring both irrigation efficiency and quality.This method divides the irrigation area into multiple sub-regions and generates a synergy index that reflects crop growth and soil condition. By combining outliers from both real-time and historical dimensions to calculate anomaly coefficients, it achieves precise monitoring and anomaly assessment of growth and soil synergy in each sub-region. Furthermore, it dynamically adjusts irrigation strategies based on anomaly coefficient ranking, avoiding the resource waste of traditional uniform irrigation. This ensures irrigation allocation better meets the actual needs of each sub-region, guaranteeing irrigation quality and crop growth stability.

[0030] In one embodiment, a smart irrigation automatic execution system based on a centralized control architecture is disclosed for implementing a smart irrigation automatic execution method based on a centralized control architecture, such as... Figure 2 As shown, the system includes: The irrigation strategy generation module is used to collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table, and obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. The adjustment strategy generation module is used to acquire sensing data within the irrigation cycle in real time, which is recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and an adjustment strategy is generated for irrigation cycle time points where the irrigation amount is unreasonable. The adjustment strategy execution module is used to adjust the irrigation amount for irrigation cycles with unreasonable irrigation amounts based on the generated adjustment strategies. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The irrigation volume allocation module is used to record the adjusted irrigation volume as the actual irrigation volume, calculate the coordination index of each sub-region based on the real-time sensing data of each sub-region within the irrigation cycle, calculate the anomaly coefficient of each sub-region within the irrigation area based on the coordination index of all sub-regions within the irrigation area, and allocate the actual irrigation volume to each sub-region based on the anomaly coefficient to execute the irrigation operation.

[0031] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0032] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0033] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A smart irrigation automatic execution method based on a centralized control architecture, characterized in that, The methods include: Collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table to obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. Real-time sensing data is acquired during the irrigation cycle and recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and adjustment strategies are generated for irrigation cycle time points where the irrigation amount is unreasonable. Based on the generated adjustment strategies, the irrigation amount for irrigation cycles with unreasonable irrigation amounts is adjusted. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The adjusted irrigation amount is recorded as the actual irrigation amount. Based on the real-time sensing data of each sub-region within the irrigation cycle, the coordination index of each sub-region is calculated. Based on the coordination index of all sub-regions within the irrigation area, the anomaly coefficient of each sub-region within the irrigation area is calculated. Based on the anomaly coefficient, the actual irrigation amount is allocated to each sub-region, and irrigation operations are performed.

2. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 1, characterized in that, The method for constructing the irrigation strategy mapping table is as follows: Collect multiple sets of sensing data corresponding to the irrigation area under historical time series and label them as historical sensing data; the sensing data includes soil sensing data, meteorological sensing data and growth status sensing data; simultaneously collect irrigation strategies and irrigation effect data under each set of historical sensing data, and combine each set of historical sensing data with the corresponding irrigation strategies and irrigation effect data to form multiple sets of sensing-strategy-effect data; build a dataset based on multiple sets of sensing-strategy-effect data; Each set of historical sensing data in the dataset is standardized, and the standardized sets of historical sensing data are then fused to form corresponding fused feature vectors. K-means clustering is used to cluster all fused feature vectors to generate multiple sets of sensing state clusters, and a unique discrete label is assigned to each set of sensing state clusters. Based on discrete labels, each group of perception-policy-effect data in the dataset is integrated into perception state cluster discrete label-policy-effect data. Based on all perception state cluster discrete label-policy-effect data, a full dataset is constructed. The K-means clustering method is used to perform secondary clustering on the full dataset to obtain multiple policy-effect clusters. At the same time, each policy-effect cluster is bound to a corresponding irrigation policy label and irrigation effect score. The frequency of occurrence of discrete labels for each perception state cluster in each strategy effect cluster is counted. The frequency of occurrence of discrete labels for each perception state cluster is compared with a preset frequency threshold. Discrete labels for perception state clusters with a frequency greater than the frequency threshold are selected and recorded as high-frequency discrete labels. Each high-frequency discrete label is bound to the corresponding irrigation strategy label and irrigation effect score to form multiple sets of association pairs, and the contribution of each set of association pairs is calculated. Based on the multiple sets of association pairs and the corresponding contribution, an irrigation strategy mapping table is constructed.

3. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 2, characterized in that, The method for obtaining the irrigation strategy for the irrigation area is as follows: Acquire the sensing data of the irrigation area and record it as the latest sensing data. Input the latest sensing data into the irrigation strategy mapping table, match all corresponding association pairs, and select the irrigation strategy label corresponding to the association pair with the highest contribution as the irrigation strategy of the irrigation area.

4. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 3, characterized in that, The method for calculating contribution is as follows: The frequency of occurrence of the high-frequency discrete label corresponding to the association pair is divided by the total frequency of occurrence of the discrete label of the perception state cluster in the corresponding strategy effect cluster to obtain the occurrence frequency of the corresponding association pair, which is recorded as the first contribution. The mutual information between the high-frequency discrete label corresponding to the association pair and the corresponding irrigation effect score is obtained and recorded as the second contribution. The irrigation effect score corresponding to the association pair is standardized to obtain the third contribution. The first contribution, the second contribution, and the third contribution are multiplied together and then normalized to obtain the contribution.

5. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 1, characterized in that, The method for determining whether the irrigation amount at a given time point in the irrigation cycle is reasonable is as follows: The irrigation time point with unreasonable irrigation amount is denoted as irrigation time point Q, the irrigation time point preceding irrigation time point Q is denoted as irrigation time point E, and the time between irrigation time point E and irrigation time point Q is denoted as one irrigation cycle. Real-time sensing data includes real-time soil sensing data, real-time meteorological sensing data, and real-time growth status sensing data. The real-time sensing data within the irrigation cycle is analyzed to obtain the growth status coefficient and soil status coefficient of irrigation time point Q. The average of the growth status coefficient and the soil status coefficient is used to calculate the total status coefficient of irrigation time point Q. Based on the total status coefficient, a reasonable threshold range for irrigation amount at irrigation time point Q is obtained. If the set irrigation amount at irrigation time point Q is not within the corresponding reasonable threshold range, the irrigation amount at irrigation time point Q is judged to be unreasonable.

6. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 5, characterized in that, The calculation methods for the growth state coefficient and the soil state coefficient are as follows: Acquire real-time growth status sensing data at irrigation time point Q, and generate growth status value at irrigation time point Q based on the real-time growth status sensing data at irrigation time point Q; acquire real-time meteorological sensing data at irrigation time point Q, generate growth correction coefficient based on the real-time meteorological sensing data at irrigation time point Q, add 1 to the growth correction coefficient, multiply by the growth status value, and then perform normalization to obtain the growth status coefficient at irrigation time point Q. Real-time soil sensing data for irrigation time points E and Q are acquired. Soil state values ​​for irrigation time points E and Q are calculated based on the real-time soil sensing data and denoted as soil state value E and soil state value Q, respectively. The difference between soil state value Q and soil state value E is denoted as the deviation value. Based on real-time meteorological sensing data acquired during the irrigation cycle, the corresponding deviation standard value for the irrigation cycle is obtained. The deviation value is divided by the deviation standard value to obtain the corresponding deviation condition value. Soil state value Q is multiplied by the deviation condition value and then normalized to obtain the soil state coefficient for irrigation time point Q.

7. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 6, characterized in that, The adjustment strategies include strategies to increase irrigation volume and strategies to decrease irrigation volume. The two endpoints of the reasonable irrigation volume threshold range are denoted as reasonable threshold Y and reasonable threshold Z, and the set irrigation volume is denoted as the expected irrigation volume. If the expected irrigation volume is less than reasonable threshold Y, an strategy to increase irrigation volume is generated; if the expected irrigation volume is greater than reasonable threshold Z, a strategy to decrease irrigation volume is generated. The next irrigation time point corresponding to the irrigation cycle time point is denoted as irrigation time point S. Forecast meteorological sensing data between irrigation time point Q and irrigation time point S is obtained, and combined with real-time meteorological sensing data of irrigation time point Q for comprehensive analysis, an irrigation volume correction factor is generated. When generating an increased irrigation strategy, the standard irrigation amount is calculated by averaging the reasonable thresholds Y and Z. The initial increase is obtained by subtracting the expected irrigation amount from the standard irrigation amount. The final increase is obtained by multiplying the initial increase by the irrigation amount correction factor, and is denoted as the irrigation increase. When generating a decreased irrigation strategy, the initial decrease is obtained by subtracting the standard irrigation amount from the expected irrigation amount. The final decrease is obtained by dividing the initial decrease by the irrigation amount correction factor, and is denoted as the irrigation decrease.

8. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 7, characterized in that, The method for obtaining the irrigation amount correction factor is as follows: Based on the acquired forecast meteorological sensing data, the data is input into the Penman-Monteith equation to obtain the expected total evapotranspiration between irrigation time point Q and irrigation time point S. At the same time, all historical evapotranspiration between irrigation time point Q and irrigation time point S are obtained. The mean evapotranspiration QS is obtained by averaging all historical evapotranspiration. The expected total evapotranspiration is divided by the mean evapotranspiration QS and then normalized to obtain the future meteorological driving stress index. Based on real-time meteorological sensing data at irrigation time point Q, the actual evapotranspiration at irrigation time point Q is obtained. At the same time, all historical evapotranspiration at irrigation time point Q is obtained. The mean evapotranspiration Q is obtained by averaging all historical evapotranspiration. The actual evapotranspiration is divided by the mean evapotranspiration Q and then normalized to obtain the real-time water deficit index. The irrigation amount correction factor is obtained by weighting and summing the real-time water deficit index and the future weather-driven stress index.

9. The intelligent irrigation automatic execution method based on a centralized control architecture according to claim 1, characterized in that, The method for allocating actual irrigation amounts to various sub-regions based on anomaly coefficients is as follows: The irrigation area is divided into multiple sub-regions. Growth status and soil status values ​​are obtained for each sub-region. The mean of these values ​​is calculated to obtain the synergy index for each sub-region. Real-time synergy indices for all sub-regions at irrigation time point Q are obtained, and their mean values ​​are calculated to obtain the real-time synergy index mean. The difference between the real-time synergy index of each sub-region and its mean value is used as the first outlier. Real-time synergy indices for each sub-region between irrigation time points E and Q are obtained, and a real-time synergy index function is constructed for each sub-region. The integral of the real-time synergy index function for each sub-region over the period from irrigation time point E to irrigation time point Q is calculated to obtain the historical cumulative value for each sub-region. The mean of the historical cumulative values ​​for all sub-regions between irrigation time points E and Q is calculated to obtain the historical cumulative mean. The difference between the historical cumulative value of each sub-region and its mean value is used as the second outlier. The first and second outliers are standardized and then summed to obtain the anomaly coefficient for each sub-region. Based on the anomaly coefficients, the actual irrigation amount is allocated to each sub-region for irrigation operations.

10. A smart irrigation automatic execution system based on a centralized control architecture, implementing the smart irrigation automatic execution method based on a centralized control architecture as described in any one of claims 1-9, characterized in that the system... include: The irrigation strategy generation module is used to collect sensing data of the irrigation area, input the sensing data into a pre-built irrigation strategy mapping table, and obtain the irrigation strategy of the irrigation area. The irrigation strategy includes the irrigation cycle and the irrigation amount at each time point of the irrigation cycle. The adjustment strategy generation module is used to acquire sensing data within the irrigation cycle in real time, which is recorded as real-time sensing data. The real-time sensing data is analyzed to generate a total state coefficient. Based on the total state coefficient, it is determined whether the irrigation amount at the corresponding irrigation cycle time point is reasonable, and an adjustment strategy is generated for irrigation cycle time points where the irrigation amount is unreasonable. The adjustment strategy execution module is used to adjust the irrigation amount for irrigation cycles with unreasonable irrigation amounts based on the generated adjustment strategies. The adjustment strategies include strategies to increase irrigation amount and strategies to decrease irrigation amount. The irrigation volume allocation module is used to record the adjusted irrigation volume as the actual irrigation volume, calculate the coordination index of each sub-region based on the real-time sensing data of each sub-region within the irrigation cycle, calculate the anomaly coefficient of each sub-region within the irrigation area based on the coordination index of all sub-regions within the irrigation area, and allocate the actual irrigation volume to each sub-region based on the anomaly coefficient to execute the irrigation operation.

Images