Iot intelligent irrigation control system based on particle swarm optimization

Abstract

The present application relates to the field of intelligent control, especially to an Internet of Things intelligent irrigation control system based on particle swarm optimization, which comprises a data acquisition module, a particle swarm optimization module and an irrigation strategy generation module; after the data acquisition module collects data, the particle swarm optimization module is used to obtain local optimal particles through particle iteration, and finally the irrigation strategy generation module is used to complete the formulation of optimal irrigation strategy; by using the particle swarm optimization algorithm, the present application dynamically adjusts the irrigation strategy by fusing historical and real-time irrigation data, accurately locates the optimal irrigation amount and interval, overcomes the limitations of traditional particle position adjustment only based on irrigation amount and frequency, significantly improves the accuracy and effect of the irrigation strategy, efficiently utilizes irrigation resources, saves water and increases yield and quality; meanwhile, the present application determines the iteration termination according to the repeated degree of real-time particle position variation, strengthens the adaptability and robustness of the system, and makes it flexible to adapt to the needs of variable irrigation areas and crop growth stages.

Description

Technical Field

[0001] This invention relates to the field of intelligent control, and more particularly to an IoT-based intelligent irrigation control system based on particle swarm optimization. Background Technology

[0002] In the process of agricultural modernization, traditional irrigation methods rely on manual experience, often leading to water waste and crop yield losses. With the development of Internet of Things (IoT) technology, real-time collection of multi-source data such as soil moisture and weather conditions provides a solid data foundation for precision irrigation. This enables irrigation systems to dynamically adjust irrigation strategies based on real-time data, avoiding the shortcomings of traditional irrigation methods and achieving efficient water resource utilization and increased crop yields.

[0003] The Internet of Things (IoT) technology, through sensor networks deployed in farmland, collects real-time data from multiple sources such as soil moisture and weather conditions, providing a solid data foundation for precision irrigation. After preprocessing and analysis, this data helps the system accurately determine the water requirements of crops, thereby enabling on-demand irrigation and improving irrigation efficiency and water resource utilization. The Particle Swarm Optimization (PSO) algorithm, with its global search capability, can dynamically optimize irrigation parameters, overcoming the bottleneck of local optima in complex environments encountered by traditional models.

[0004] However, in the process of IoT smart irrigation using conventional particle swarm optimization (PSO) algorithms, although irrigation volume and frequency can be determined based on historical PSO particle position data, in actual irrigation, the selection of the optimal particle point can only be achieved by updating the particle position by changing the irrigation volume and frequency to bring it closer to the global optimum. This method may result in the actual collected particles failing to reach the optimal objective function value, thus affecting the accuracy of the irrigation strategy and the optimization of irrigation effects. Summary of the Invention

[0005] To address the problem that when conventional particle swarm optimization (PSO) algorithms are applied to IoT smart irrigation, simply adjusting irrigation volume and frequency to update particle positions to approximate the global optimum may result in actual particles failing to reach the optimal objective function value, thus affecting the accuracy and effectiveness of the irrigation strategy, this invention provides an IoT smart irrigation control system based on particle swarm optimization, employing the following technical solution:

[0006] In the first aspect, the IoT-based intelligent irrigation control system based on particle swarm optimization includes: a data acquisition module for deploying multiple IoT sensor nodes within the irrigation area to acquire historical and real-time irrigation data; a particle swarm optimization module for iterating the particle swarm based on the acquired historical irrigation data to obtain the globally optimal particle positions of historical particles, recording position changes and objective function values; iterating the particle swarm based on the real-time irrigation data, determining the estimated objective function value of the real-time particles based on the objective function value of the historical particles closest to the real-time particles, evaluating the repetition degree of particles based on the position changes of particles over the most recent preset number of times, determining particle termination based on the repetition degree of particles, and obtaining locally optimal particles; and an irrigation strategy generation module for determining the optimal irrigation strategy based on the locally optimal particles, including irrigation amount and irrigation interval, to achieve precision irrigation; after the data acquisition module acquires data, the particle swarm optimization module iterates the particles to obtain locally optimal particles, and finally the irrigation strategy generation module completes the formulation of the optimal irrigation strategy.

[0007] The data acquisition module acquires historical and real-time irrigation data within the irrigation area. The particle swarm optimization module uses this data for iterative optimization, dynamically adjusting the irrigation strategy. During the iteration process, the degree of repetition of particles is evaluated to determine whether to terminate the iteration, thereby obtaining locally optimal particles. Finally, the irrigation strategy generation module determines the optimal irrigation strategy based on the locally optimal particles, including irrigation amount and irrigation interval, achieving precision irrigation. This not only improves the accuracy and effectiveness of the irrigation strategy but also enhances the system's adaptability and robustness, while reducing system energy consumption and operating costs, ensuring efficient use of irrigation resources and healthy crop growth.

[0008] Optionally, the steps of obtaining the position change and the objective function value include:

[0009] Using each data point in the historical irrigation data as a particle, the objective function value of each particle is calculated based on the difference between the average soil moisture content after irrigation and the preset optimal moisture content.

[0010] Particle swarm optimization is performed on all data points in the historical irrigation data to obtain the position of each particle after each iteration.

[0011] By quantitatively analyzing historical irrigation data, the iteration position of each particle can be accurately located, effectively improving the scientific nature and accuracy of irrigation strategies, gradually approaching the global optimal solution, thereby achieving efficient and water-saving intelligent irrigation, ensuring the efficient use of irrigation resources and the healthy growth of crops.

[0012] Optionally, the method for calculating the estimated objective function value includes:

[0013] A predetermined number of historical particles are used as neighboring particles. The Euclidean distance between the real-time particle and the neighboring particles is calculated. The product of the reciprocal of the Euclidean distance and the objective function value of each neighboring particle is summed to obtain the estimated objective function value of the real-time particle.

[0014] By considering the influence of neighboring particles on real-time particles, the predicted objective function value of real-time particles is dynamically adjusted by integrating neighborhood information. This allows for a more accurate assessment of the optimization level of real-time particles, improving the global search capability and accuracy of the particle swarm optimization algorithm. Consequently, the adjustment of irrigation strategies becomes more scientific and reasonable, effectively enhancing the intelligence level and irrigation effect of the irrigation system.

[0015] Optionally, the method for calculating the degree of repetition includes:

[0016] Calculate the ratio between the range of each feature for the last preset number of real-time particles and the range of each feature for all real-time particles, and use a negative exponential function to exponentially decay the ratio to obtain the relative range of each feature. Average the relative ranges of all features to obtain the average relative range of each feature. Sum the average relative ranges of each feature for the last preset number of particles and take the average value to obtain the average relative range of the last preset number of particles.

[0017] Calculate the ratio between the standard deviation of the predicted objective function of the final preset number of particles and the standard deviation of the predicted objective function values ​​of all particles, and use a negative exponential function for exponential mapping to obtain the relative change of the standard deviation; multiply the average relative range by the relative change of the standard deviation to obtain the real-time particle repetition rate.

[0018] By quantitatively analyzing the variation amplitude of particles in various characteristics and the fluctuation of objective function values, the degree of particle repetition can be accurately assessed, thus providing a scientific basis for iteration termination, effectively balancing iteration efficiency and accuracy, avoiding over-computation, ensuring that the algorithm quickly converges to a stable solution in complex environments, improving the accuracy and reliability of irrigation strategy optimization, and enhancing the overall performance and adaptability of the system.

[0019] Optionally, the step of obtaining the locally optimal particle includes:

[0020] The system determines when to terminate a particle iteration by setting a stopping threshold based on the repetition rate of the particles. If the repetition rate of the real-time particles is less than or equal to the stopping threshold, the iteration continues; otherwise, if it is greater than the stopping threshold, the iteration of the real-time particles is terminated, and the final preset number of real-time particles are used to select the optimal particle.

[0021] By dynamically evaluating the degree of particle iteration and setting a stopping threshold accordingly, it is possible to accurately determine whether a particle is trapped in a local optimum, thereby effectively avoiding invalid iterations, improving the convergence speed and efficiency of the algorithm, ensuring that a high-quality irrigation strategy can be obtained quickly under limited computing resources, and enhancing the real-time performance and adaptability of the system.

[0022] Optionally, the step of determining the optimal irrigation strategy includes:

[0023] The particle with the largest objective function value among the last preset number of real-time particles is taken as the local optimal particle, and the operation data of the local optimal particle is taken as the current optimal solution to complete the local optimal irrigation control.

[0024] Optionally, the irrigation data includes:

[0025] Environmental data: Soil moisture content, soil moisture content information at different locations within the irrigation area; Soil temperature, soil temperature information at different locations within the irrigation area; Light intensity, light intensity information at different locations within the irrigation area; Rainfall, rainfall information within the irrigation interval.

[0026] Operational data: Irrigation volume, water volume information for each irrigation; time interval, time interval information between two irrigations; and preprocessing.

[0027] Optionally, the preprocessing steps include:

[0028] The irrigation data is cleaned to remove outliers and noise to ensure its accuracy and reliability. Data normalization is performed to unify irrigation data of different dimensions. Missing data is interpolated to fill in gaps in the irrigation data. Time synchronization is also performed to align the timestamps of the irrigation data.

[0029] The present invention has the following effects:

[0030] 1. This invention utilizes a particle swarm optimization algorithm, combined with historical and real-time irrigation data, to dynamically adjust irrigation strategies, enabling more precise determination of optimal irrigation amounts and intervals. This avoids the limitations of traditional methods that merely adjust irrigation amounts and frequencies to update particle positions, thereby improving the accuracy and effectiveness of irrigation strategies, ensuring efficient use of irrigation resources, reducing water waste, and enhancing crop yield and quality.

[0031] 2. This invention assesses the degree of repetition by analyzing the real-time position changes of particles and determines whether to terminate the iteration accordingly. The iteration termination strategy based on the degree of repetition effectively balances the efficiency and accuracy of the iteration, ensuring that it approaches the global optimum as closely as possible without overcomputation. Furthermore, preprocessing steps such as data cleaning, normalization, interpolation, and time synchronization improve the accuracy and reliability of the data, enhancing the system's adaptability and robustness, enabling it to better meet the needs of different irrigation areas and crop growth stages. Attached Figure Description

[0032] Figure 1 This is a module connection diagram of the IoT smart irrigation control system based on particle swarm optimization according to an embodiment of the present invention. Detailed Implementation

[0033] This invention discloses an IoT-based intelligent irrigation control system based on particle swarm optimization, with reference to... Figure 1 It includes: a data acquisition module, a particle swarm optimization module, and an irrigation strategy generation module. After the data acquisition module collects data, the particle swarm optimization module performs particle iteration to obtain locally optimal particles, and finally, the irrigation strategy generation module completes the formulation of the optimal irrigation strategy.

[0034] A data acquisition module is used to deploy multiple IoT sensor nodes within the irrigation area to acquire historical and real-time irrigation data. In one embodiment, the irrigation data may include:

[0035] Environmental data includes: soil moisture content (information on soil moisture content at different locations within the irrigation area); soil temperature (information on soil temperature at different locations within the irrigation area); light intensity (information on light intensity at different locations within the irrigation area); and rainfall (information on rainfall during irrigation intervals). Operational data includes: irrigation volume (information on the amount of water used in each irrigation); and time interval (information on the time interval between two irrigations). Preprocessing is performed on these data. The sampling frequency is once a day, primarily based on the cyclical nature of crop growth and the stability of environmental parameters. Crop growth status and water requirements typically change on a daily cycle, while environmental parameters such as soil moisture content and soil temperature remain relatively stable within a day. Therefore, a daily sampling frequency effectively captures key data while avoiding misjudgments caused by short-term fluctuations. Furthermore, a lower sampling frequency helps reduce system energy consumption and data processing costs, improves operational efficiency, reduces maintenance workload, and ensures long-term stable system operation. This frequency setting can also be flexibly adjusted according to specific needs to adapt to the requirements of different irrigation areas and crop growth stages.

[0036] The irrigation data is cleaned to remove outliers and noise to ensure its accuracy and reliability. Data normalization is performed to unify irrigation data of different dimensions. Missing data is interpolated to fill in gaps in the irrigation data. Time synchronization is also performed to align the timestamps of the irrigation data.

[0037] The particle swarm optimization module iterates the particle swarm based on collected historical irrigation data to obtain the global optimal particle position of historical particles, record position changes and objective function values; it iterates the particles on real-time irrigation data, determines the estimated objective function value of real-time particles based on the objective function value of the historical particles closest to the real-time particles, evaluates the repetition degree of particles based on the position changes of particles in the most recent preset number of times, determines particle termination based on the repetition degree of particles, and obtains the local optimal particles.

[0038] In this embodiment, real-time data points are first obtained based on real-time irrigation data and integrated into the historical particle swarm to drive iterative movement towards the globally optimal particle position. Since environmental data at the particle position (e.g., soil moisture content, soil temperature, light intensity, and rainfall within the irrigation interval) remain constant in the real-time data, only operational data (e.g., current irrigation amount, and the irrigation interval from the previous one) are adjusted to encourage the particles to move closer to the globally optimal solution. During the iteration process, the estimated objective function value for the real-time particle's iteration position is calculated based on the objective function values ​​of historical particles with Euclidean distances close to the real-time particles. Furthermore, by analyzing the changes in the particle's recent position and objective function value, the degree of repetition of the particle's position is assessed to determine whether the particle has fallen into a repetitive state, thereby deciding whether to terminate the particle's iteration process. The specific steps are as follows:

[0039] In other words, select the particles closest to the real-time particles from the historical particles. The number of particles, which can be adjusted according to specific circumstances, is to ensure that only the data points most relevant to the current particle are considered. This allows for further calculation of the weights of neighboring particles, determined by the Euclidean distance between the real-time particle and its neighbors. The closer the distance, the closer the two particles are in the solution space, and therefore the greater the influence of the neighboring particles on the real-time particle. A larger distance results in a larger weight. The contributions of all neighboring particles are summed to obtain the sum of the real-time particle's contribution and the estimated objective function value, reflecting the optimization level of the real-time particle at its current position.

[0040] Specifically, the objective function is estimated to satisfy the following relationship:

[0041] ;

[0042] In the formula, Indicates the first The predicted objective function value for each real-time particle. This represents the number of neighboring particles. Indicates the first The first real-time particle The Euclidean distance between neighboring particles. Indicates the first The objective function value of a neighboring particle. A preset number of historical particles are used as neighboring particles. The Euclidean distance between the real-time particle and the neighboring particles is calculated. The inverse of the Euclidean distance is multiplied by the objective function value of each neighboring particle to obtain the estimated objective function value of the real-time particle.

[0043] Furthermore, since real-time particles rarely reach the global optimum directly during iteration, they typically fluctuate around local optima. Therefore, it's necessary to analyze the recent position changes of real-time particles to determine if they are approaching a local optimum. Real-time particles continuously update their positions during iteration, attempting to find the optimal solution. However, due to the influence of complex environments and multidimensional data, particles often struggle to reach the global optimum directly, instead lingering around local optima. By analyzing the changes in particle positions in recent iterations, we can assess whether the particle is repeatedly fluctuating around a local optimum. If the changes in particle position are small, it indicates that the particle may have approached a local optimum. In optimization problems, a local optimum is the point where the objective function value is optimal within a certain local region. Particles may get trapped in local optima during iteration, preventing them from continuing to approach the global optimum.

[0044] Specifically, after the real-time particle completes at least 5 iterations, the real-time particle positions and corresponding estimated objective function values ​​from the last 5 iterations are collected. Based on this data, the repetition rate of the particle positions is calculated to determine whether the particle has fallen into a local optimum, and thus decide whether to terminate the iteration process. The specific steps include:

[0045] Specifically, the degree of repetition satisfies the following relationship:

[0046] ;

[0047] In the formula, Indicates the degree of repetition of particles in real time. This indicates the final number of local particles. This indicates the number of items contained in a data point. Indicates the last The first real-time particle The range of the data items Represents the first of all real-time particles The range of the data items Indicates the last The standard deviation of the predicted objective function value for each real-time particle. The standard deviation of the predicted objective function values ​​for all real-time particles is represented. Represented by natural numbers An exponential function with base 0.

[0048] It should be noted that, Used to measure the final The first particle The range of the data item and the first real-time particle The ratio of the ranges of the data items. The smaller the ratio, the better. The local particle in the first The smaller the change in the data, the more likely the particle has entered the convergence phase. Reflecting the final The ratio of the standard deviation of the predicted objective function value for each particle to the standard deviation of the predicted objective function values ​​for all particles. The smaller this ratio, the better the final result. The smaller the fluctuation in the predicted objective function value of a particle, the more likely the particle is to be in a convergent state.

[0049] Specifically, in order to determine whether to terminate the real-time particle iteration process, the degree of particle repetition is compared with a preset stopping threshold.

[0050] Specifically, if the repetition rate of a real-time particle is less than or equal to the stopping threshold, it indicates that the particle has not yet fully converged and needs to continue iterating for further optimization. Conversely, if the repetition rate is greater than the stopping threshold, it means that the particle has reached a relatively stable convergence state. In this case, the iteration process of the real-time particle is terminated, and the optimal particle is selected from the last preset number of real-time particles as the optimal solution for the current iteration. This iteration termination strategy based on repetition rate effectively balances the efficiency and accuracy of the iteration, ensuring that it approaches the global optimum as closely as possible without overcomputation. The stopping threshold is... It can be adjusted according to the specific circumstances.

[0051] The irrigation strategy generation module determines the optimal irrigation strategy based on local optimal particles, including irrigation amount and irrigation interval, to achieve precision irrigation.

[0052] To determine the locally optimal irrigation strategy, the particle with the largest objective function value is first selected from the last five real-time particles and identified as the locally optimal particle. Then, the operational data corresponding to this locally optimal particle, such as irrigation amount and irrigation interval, is taken as the current optimal solution. Based on this optimal operational data, the system will execute locally optimal irrigation control, thereby achieving precision irrigation, ensuring efficient use of irrigation resources, and promoting healthy crop growth.

[0053] By evaluating the objective function value of each particle, the particle that performs best in the current iteration is selected, and the operation parameters (such as irrigation amount and irrigation interval) corresponding to its position are considered the local optimum closest to the global optimum. Subsequently, the operation data corresponding to this local optimum particle is taken as the current optimum, and the system will perform locally optimal irrigation control based on this optimal operation data. This dynamic adjustment ensures that the irrigation system achieves good irrigation results under different environmental conditions, avoiding water waste and over-irrigation or water shortage of crops, thereby achieving precision irrigation, ensuring efficient use of irrigation resources, and promoting healthy crop growth.

[0054] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An IoT based smart irrigation control system based on particle swarm optimization characterized by, The application relates to an irrigation strategy optimization method and device. The data acquisition module is used for deploying a plurality of Internet of Things sensor nodes in an irrigation area to acquire historical and real-time irrigation data. The particle swarm optimization module is used for performing particle swarm iteration according to the acquired historical irrigation data, acquiring a global optimal particle position of historical particles, recording position change conditions and target function values, performing particle iteration on real-time irrigation data, determining an estimated target function value of the real-time particle according to a target function value of a historical particle closest to the real-time particle, evaluating a repetition degree of the real-time particle based on position changes of the real-time particle in the last preset number of times, judging particle termination according to the repetition degree of the real-time particle, and acquiring a locally optimal particle. In the iteration process, the estimated target function value of the iteration position of the real-time particle data is calculated according to the target function value of the historical particle data closest to the real-time particle in terms of the Euclidean distance, the repetition degree of the particle position is evaluated by analyzing the position and target function value changes of the particle in the last several times, and the locally optimal particle is acquired. The irrigation strategy generation module is used for determining an optimal irrigation strategy according to the locally optimal particle, the optimal irrigation strategy including an irrigation amount and an irrigation interval, so as to realize precise irrigation. After the data acquisition module acquires data, the particle swarm optimization module is used for performing particle iteration to acquire the locally optimal particle, and finally the irrigation strategy generation module is used for determining the optimal irrigation strategy.

2. The IoT based smart irrigation control system using particle swarm optimization as claimed in claim 1, wherein, The step of acquiring the position change conditions and the target function values comprises the following steps: Each data point in the historical irrigation data is taken as a particle, and the target function value of each particle is calculated according to the gap between the average soil moisture content after particle irrigation and the preset optimal soil moisture content. The particle swarm iteration is performed on all data points in the historical irrigation data, and the particle position of each particle after each iteration is acquired.

3. The IoT based smart irrigation control system using particle swarm optimization as claimed in claim 1, wherein, The calculation method of the estimated target function value comprises the following steps: A preset number of historical particles are taken as neighborhood particles, the Euclidean distance between the real-time particle and the neighborhood particles is calculated, and the product sum of the reciprocal of the Euclidean distance and the target function values of the neighborhood particles is taken as the estimated target function value of the real-time particle.

4. The particle swarm optimization based Internet of Things smart irrigation control system of claim 1, wherein, The calculation method of the repetition degree comprises the following steps: The ratio between the range of the last preset number of real-time particles in each feature and the range of all real-time particles in each feature is calculated, the ratio is exponentially attenuated by using a negative exponential function, the relative range of each feature is obtained, the average relative range of all features is obtained, the average relative range of each feature of the last preset number of particles is summed, and the average value is taken, so that the average relative range of the last preset number of particles is obtained; The ratio between the standard deviation of the estimated target function of the last preset number of particles and the standard deviation of the estimated target function value of all particles is calculated, and the relative change of the standard deviation is obtained by using an exponential mapping of a negative exponential function; the average relative range is multiplied by the relative change of the standard deviation, so that the repetition degree of the real-time particle is obtained.

5. The particle swarm optimization based Internet of Things smart irrigation control system as claimed in claim 1, wherein, The step of acquiring the locally optimal particle comprises the following steps: The judgment of particle termination is performed according to the setting of a stop threshold value according to the iteration degree of the particle, and in response to the iteration degree of the real-time particle being less than or equal to the stop threshold value, the iteration is continued, otherwise, the iteration of the real-time particle is terminated, and the last preset number of real-time particles are obtained to select the optimal particle.

6. The particle swarm optimization based Internet of Things smart irrigation control system of claim 1, wherein, The step of determining the optimal irrigation strategy comprises: The particle with the maximum target function value in the last preset number of real-time particles is taken as a local optimal particle, and the operation data of the local optimal particle is taken as a current optimal solution, so as to complete the local optimal irrigation control.

7. The particle swarm optimization based Internet of Things smart irrigation control system of claim 1, wherein, The irrigation data comprises: environmental data: soil moisture, soil moisture information at different positions in the irrigation area; soil temperature, soil temperature information at different positions in the irrigation area; light intensity, light intensity information at different positions in the irrigation area; rainfall, rainfall information within the irrigation interval; operation data: irrigation amount, water amount information of each irrigation; time interval, time interval information between two irrigations; and preprocessing.

8. The IoT based smart irrigation control system using particle swarm optimization as claimed in claim 7, wherein, The preprocessing step comprises: data cleaning is performed on the irrigation data to remove outliers and noise, so as to ensure the accuracy and reliability of the irrigation data, data normalization is performed to unify irrigation data of different dimensions, interpolation is performed on missing data to fill in the gaps of the irrigation data, and time synchronization is performed to align the time stamps of the irrigation data.

Images