Multi-mode soil moisture content sensor layout method

Abstract

The invention discloses a multi-mode soil moisture content sensor layout method, and relates to the technical field of sensor network layout and optimization. The problem that soil moisture content monitoring quality and low operation cost cannot be considered at the same time in a traditional laying method is solved. According to the method, a soil moisture status state estimation vector and an error covariance estimation vector are obtained and corrected by using a constructed measurement and calculation model and a constructed measurement model, and an objective function is constructed by using the corrected error covariance estimation vector and the soil moisture status state estimation vector to minimize estimation error, state estimation deviation and sensor operation cost. And according to a set constraint condition, enabling the target function to be minimum for constraint, and carrying out optimization solution on the target function by utilizing an optimization algorithm to obtain an optimal layout position of each sensor. The method is mainly used for carrying out optimized layout design on the sensors in the farmland.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor network layout and optimization, and specifically provides a method for deploying multi-modal soil moisture sensors that can improve the monitoring accuracy of soil moisture, reduce operating costs while ensuring the monitoring coverage, and achieve efficient and economical deployment. Background Art

[0002] In modern agricultural production, the application of precision agriculture technology has gradually become an important means to improve crop yields and resource utilization efficiency. Precision agriculture relies on real-time monitoring of environmental variables such as soil, water, climate, and crop growth to support decision-making processes and optimize resource allocation. Soil moisture monitoring is one of the core technologies in precision agriculture management. With the increasing requirements for water resource management in agricultural production, how to achieve rational utilization of water resources through an efficient and accurate monitoring system has become an important issue in the development of modern agriculture. Traditional monitoring methods usually rely on deploying a large number of sensors in farmland to monitor various environmental factors by collecting soil moisture data in real time. However, in the application of large-scale agricultural production in complex environments, there are problems such as difficulty in ensuring the reliability of sensor monitoring data and high operation and maintenance costs.

[0003] Currently, most sensor deployment methods rely on empirical deployment or grid deployment. Empirical deployment methods rely on human experience and usually ignore the impact of environmental factors such as terrain, soil type, and water flow on monitoring, making it difficult to adapt to the dynamic changes of different farmland environments. Although grid deployment can provide uniform spatial coverage, this method often incurs high costs, and due to the measurement errors of the sensors themselves and external interference factors, there may be certain uncertainties in the sensor data, which can affect the accuracy and reliability of soil moisture monitoring. Therefore, when deploying sensors, it is necessary to consider various uncertainties in agricultural production, optimize the sensor deployment method while ensuring monitoring accuracy, reduce system operation costs, and improve the reliability of monitoring data, which is the key to optimizing the sensor deployment location.

[0004] Therefore, in view of the deficiencies of the current technology, the present invention proposes a method for deploying multi-modal soil moisture sensors. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that traditional deployment methods cannot simultaneously take into account the quality of soil moisture monitoring and low operating costs, and provides a method for deploying multi-modal soil moisture sensors.

[0006] A method for deploying multi-modal soil moisture sensors, the method includes:

[0007] S1. Initialization stage:

[0008] Determine the number of grid cells N representing the dynamic changes of the plots through grid-based plots;

[0009] Determine the number of sensors, where the number of sensors is less than N, and all sensors are distributed in different grid cells;

[0010] Initialize the positions of the sensors, the initial values of the soil moisture state estimation vector, and the initial value P of the error covariance estimation vector; the soil moisture state is the soil temperature and humidity state within the grid cell where the sensor is located;

[0011] S2. Model construction stage:

[0012] Construct the measurement model and the measurement model for the grid cells;

[0013] The measurement model is used to obtain the soil moisture state estimation vector x caused by measurement at time k by combining the initialized sensor positions and the initial values of the soil moisture state estimation vector k ;

[0014] According to x k , Q k , and P, obtain the error covariance estimation vector at time k Among them, Q k is the process noise covariance matrix of the soil moisture state estimation vector at time k, k is the time index, k = 1, 2,..., T′, and T′ is the total number of sampling times;

[0015] The measurement model is used to obtain the soil moisture state measurement vector z caused by measurement at time k by combining the soil moisture state estimation vector x k ; k ;

[0016] S3. Correction stage:

[0017] Denoise x k to obtain the denoised soil moisture state estimation vector

[0018] According to obtain H k ; among them, H k is the Jacobian matrix of the nonlinear measurement function at time k;

[0019] Use z k to correct to obtain the corrected soil moisture state estimation vector at time k

[0020] Use H k to correct to obtain the corrected error covariance estimation vector P at time k k ;

[0021] S4. Optimization stage:

[0022] Based on the corrected soil moisture state estimation vector at each moment and the error covariance estimation vector P k , an objective function is constructed with the aim of minimizing the estimation error, state estimation deviation, and sensor operation cost. The objective function is minimized according to the set constraints, and an optimization algorithm is used to optimize and solve the objective function to obtain the optimal deployment positions of each sensor.

[0023] Preferably, the expression of the measurement model is:

[0024] x k = f(x k-1 , u k ) + w k ;

[0025] f(·) is a non-linear measurement function;

[0026] x k is the soil moisture state estimation vector at time k, and the humidity value of the nth grid cell, the temperature value of the nth grid cell, n = 1, 2... N;

[0027] u k is the control vector at time k;

[0028] w k is the process noise at time k, w k satisfies the normal distribution, and the process noise covariance matrix of the soil moisture state estimation vector at time k where represents the variance of the humidity process noise, represents the standard deviation of the humidity process noise, represents the variance of the temperature process noise, represents the standard deviation of the temperature process noise;

[0029] The expression of the measurement model is:

[0030] z k = h(x k ) + v k ;

[0031] h(·) is a non-linear measurement function;

[0032] z k is the soil moisture state measurement vector at time k;

[0033] v k is the observation noise at time k, vk obeys a normal distribution, and the observation noise covariance matrix of the soil moisture state measurement vector at time k wherein, represents the variance of the humidity observation noise at time k, represents the standard deviation of the humidity observation noise at time k, represents the variance of the temperature observation noise at time k, represents the standard deviation of the temperature observation noise at time k.

[0034] Preferably, in step S2, according to x k , Q k and P, the error covariance estimation vector at time k is obtained by the following implementation method:

[0035] S21. Let the process noise w k in x k be 0, and the denoised soil moisture state estimation vector

[0036] at time k is obtained; S22. According to k-1 , the Jacobian matrix F

[0037] of the nonlinear measurement function at time k - 1 is obtained, where k-1 S23. According to F k and P, the error covariance estimation vector

[0038] at time k is obtained. Preferably,

[0039] where T is the transpose. k Preferably, in step S3, x is denoised to obtain the denoised soil moisture state estimation vector

[0040] by the following implementation method: k Let the process noise w k in x

[0041] be 0, and the denoised soil moisture state estimation vector k is corrected by using z to obtain the corrected soil moisture state estimation vector at time k by the following implementation method:

[0042] S311. According to and R k , the Kalman gain K k at time k is calculated;

[0043] S312. According to K k and z k , correct to obtain the corrected soil moisture state estimation vector

[0044] Preferably, where T is the transpose.

[0045] Preferably, in step S3, use H k to correct to obtain the corrected error covariance estimation vector P k at time k, and the implementation method is:

[0046]

[0047] where P k is the corrected error covariance estimation vector at time k.

[0048] Preferably, the constructed objective function is:

[0049]

[0050] Constraint conditions:

[0051] q ∈ M;

[0052] n' < N;

[0053] q is the sensor position vector combination, q = {m1, m2,..., m n′}; m i is the position of sensor i, i = 1, 2... n';

[0054] Tr(P k ) is the trace of P k ;

[0055] x 目标 is the target state vector;

[0056] γ is the weight factor of the cost function;

[0057] n' is the total number of sensors;

[0058] O i is the operating cost of the i-th sensor;

[0059] M is the combination vector of all combination forms of sensor placement positions.

[0060] Preferably,

[0061] Advantages of the present invention:

[0062] The present invention proposes a method for arranging multi-modal soil moisture sensors. Through the initialization stage, model construction stage, correction stage and optimization stage, the soil moisture state is estimated, the measured soil moisture value is corrected through state estimation, and the optimized sensor arrangement scheme is designed and determined considering the monitoring quality and monitoring cost, thereby improving the reliability of monitoring data and reducing the operation cost of the system.

[0063] By optimizing the sensor arrangement scheme, while real-time monitoring the soil moisture, the present invention uses state estimation to improve the reliability of monitoring data, reduces the sensor operation cost at the same time, thereby improving the monitoring quality and reducing the monitoring cost, providing a more accurate basis for environmental regulation in agricultural production, and enhancing the intelligent level of crop growth management.

[0064] Under the condition that the number of sensors remains unchanged, the method of the present invention improves the accuracy and stability of data collection through scientific arrangement, reduces unnecessary energy consumption and maintenance costs, thereby improving the scientific nature and intelligence of agricultural management and the resource utilization efficiency. Description of the drawings

[0065] Figure 1 is a schematic diagram of the principle of a method for arranging multi-modal soil moisture sensors according to the present invention;

[0066] Figure 2 is a flowchart of the particle swarm optimization algorithm in the prior art. Specific embodiments

[0067] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work belong to the scope of protection of the present invention.

[0068] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0069] Specific embodiment 1. Refer to Figure 1 This embodiment is described. A method for arranging multi-modal soil moisture sensors described in this embodiment includes:

[0070] S1. Initialization stage:

[0071] By gridifying the plot, determine the number N of grid units representing the dynamic changes of the plot;

[0072] Determine the number of sensors, where the number of sensors is less than N, and all sensors are distributed in different grid cells;

[0073] Initialize the positions of the sensors, the initial values of the soil moisture state estimation vector, and the initial value P of the error covariance estimation vector; the soil moisture state is the soil temperature and humidity state within the grid cell where the sensor is located;

[0074] S2. Model construction stage:

[0075] Construct the measurement model and the measurement model for the grid cell;

[0076] The measurement model is used to combine the initialized sensor positions and the initial values of the soil moisture state estimation vector to obtain the soil moisture state estimation vector x caused by the measurement at time k k ;

[0077] According to x k , Q k , and P, obtain the error covariance estimation vector at time k where Q k is the process noise covariance matrix of the soil moisture state estimation vector at time k, k is the time index, k = 1, 2,..., T′, and T′ is the total number of sampling times;

[0078] The measurement model is used to combine the soil moisture state estimation vector x k to obtain the soil moisture state measurement vector z caused by the measurement at time k k ;

[0079] S3. Correction stage:

[0080] Denoise x k to obtain the denoised soil moisture state estimation vector

[0081] According to obtain H k ; where H k is the Jacobian matrix of the nonlinear measurement function at time k;

[0082] Use z k to correct to obtain the corrected soil moisture state estimation vector at time k

[0083] Use H k to correct to obtain the corrected error covariance estimation vector P at time k k ;

[0084] S4. Optimization stage:

[0085] Based on the estimated vector of soil moisture state at each corrected moment and the error covariance estimation vector P k , an objective function is constructed with the aim of minimizing the estimation error, state estimation deviation, and sensor operation cost. The objective function is minimized according to the set constraints, and an optimization algorithm is used to optimize and solve the objective function to obtain the optimal layout positions of each sensor.

[0086] In this embodiment, through the initialization stage, model construction stage, correction stage, and optimization stage, the soil moisture state is estimated. The soil moisture measurement value is corrected through state estimation, and an optimized sensor layout scheme is designed considering the monitoring quality and monitoring cost, thereby improving the reliability of monitoring data and reducing the operation cost of the system. By optimizing the sensor layout scheme, while real-time monitoring the soil moisture, state estimation is used to improve the reliability of monitoring data, and at the same time reduce the sensor operation cost, thereby improving the monitoring quality and reducing the monitoring cost, providing a more accurate basis for environmental control in agricultural production, and enhancing the intelligent level of crop growth management.

[0087] The method of the present invention improves the accuracy and stability of data acquisition through scientific layout without changing the number of sensors, reduces unnecessary energy consumption and maintenance costs, thereby improving the scientific nature and intelligence of agricultural management and improving the resource utilization efficiency.

[0088] The optimization algorithm used can be implemented by existing technologies. Specifically, it can be implemented by the particle swarm optimization algorithm. Refer to Figure 2 , specifically, it can be:

[0089] a. Initialize the particle swarm: Randomly generate an initialized particle swarm. Each particle is represented by a binary string of a fixed length. A random initial velocity is assigned to each particle, and its individual optimal solution and global optimal solution are initialized.

[0090] b. Evaluate the fitness: Define a fitness function with the goal of minimizing the estimation error, state estimation deviation, and sensor operation cost. The function measures the quality of the sensor position. Calculate the fitness of the current position of each particle, and update the individual optimal solution and global optimal solution of the particle.

[0091] c. Update the velocity and position: Update the velocity and position of the particle according to the current velocity, individual optimal position, and global optimal position of the particle.

[0092] d. Termination condition: Repeat the steps of "evaluate - update velocity - update position" to gradually optimize the particle swarm. When the maximum number of iterations is reached or the fitness function value of the global optimal solution no longer improves significantly, the algorithm terminates.

[0093] Furthermore, the expression of the measurement model is:

[0094] x k = f(x k-1 , u k ) + w k ;

[0095] f(·) is a non - linear measurement function;

[0096] x k is the soil moisture state estimation vector at time k, and is the humidity value of the nth grid cell, is the temperature value of the nth grid cell, n = 1, 2... N;

[0097] u k is the control vector at time k;

[0098] w k is the process noise at time k, w k follows a normal distribution, and the process noise covariance matrix of the soil moisture state estimation vector at time k where represents the variance of the humidity process noise, represents the standard deviation of the humidity process noise, represents the variance of the temperature process noise, represents the standard deviation of the temperature process noise;

[0099] The expression of the measurement model is:

[0100] z k = h(x k ) + v k ;

[0101] h(·) is a non - linear measurement function;

[0102] z k is the soil moisture state measurement vector at time k;

[0103] v k is the observation noise at time k, v k follows a normal distribution, and the observation noise covariance matrix of the soil moisture state measurement vector at time k where, represents the variance of the humidity observation noise at time k, represents the standard deviation of the humidity observation noise at time k, represents the variance of the temperature observation noise at time k, represents the standard deviation of the temperature observation noise at time k.

[0104] Furthermore, in step S2, according to x k , Q kand P to obtain the error covariance estimation vector at time k The implementation method includes:

[0105] S21. Let the process noise w k in k be 0 to obtain the denoised soil moisture state estimation vector at time k

[0106] S22. According to obtain the Jacobian matrix F k-1 of the nonlinear measurement function at time k - 1, where

[0107] S23. According to F k-1 , Q k and P to obtain the error covariance estimation vector at time k where T is the transpose.

[0108] In this preferred embodiment, the specific process of obtaining is given, which can determine the estimation error of the denoised soil moisture state estimation vector of the system, reduce the uncertainty in the soil moisture state estimation of the system through subsequent adjustment, thereby improving the accuracy of the state estimation and ensuring that the state estimation can effectively verify the quality of the sensor monitoring data.

[0109] Furthermore, in step S3, the denoising of x k to obtain the denoised soil moisture state estimation vector is implemented as follows:

[0110] Let the process noise w k in x k be 0 to obtain the denoised soil moisture state estimation vector at time k

[0111] Use z k to correct to obtain the corrected soil moisture state estimation vector at time k The implementation method includes:

[0112] S311. According to and R k , calculate the Kalman gain K k at time k;

[0113] S312. According to K k and z k , correct to obtain the corrected soil moisture state estimation vector

[0114] In this preferred embodiment, the corrected soil moisture state estimation vector is given. The specific process can reduce the influence of environmental noise and sensor measurement errors on soil moisture monitoring, making the estimation result more accurate and reliable, thereby improving the monitoring accuracy.

[0115] Specifically, in step S3, H k is used to perform correction to obtain the corrected error covariance estimation vector P k at time k, and the implementation method is:

[0116]

[0117] Among them,

[0118] Furthermore, the constructed objective function is:

[0119]

[0120] Constraint conditions:

[0121] q ∈ M;

[0122] n' < N;

[0123] q is the combination vector of sensor position vectors, q = {m1, m2,..., m n′}; m i is the position of sensor i, i = 1, 2... n';

[0124] Tr(P k ) is the trace of P k ;

[0125] x 目标 is the target state vector;

[0126] γ is the weight factor of the cost function;

[0127] n' is the total number of sensors;

[0128] O i is the operating cost of the i-th sensor;

[0129] M is the combination vector of all combination forms of sensor placement positions.

[0130] In this preferred embodiment, the specific form of the constructed objective function is given. This objective function considers the influence of estimation error, state estimation deviation, and sensor operating cost on the accuracy of sensor temperature and humidity measurement, the stability of state estimation, and the energy consumption control of sensors, and has the effects of improving the quality of soil moisture monitoring, reducing measurement deviation, and optimizing resource allocation.

[0131] Although the present invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. Accordingly, it should be understood that numerous modifications may be made to the exemplary embodiments, and other arrangements may be designed, without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in connection with a separate embodiment may be used in other described embodiments.

Claims

1. A multi-modal soil moisture sensor deployment method, characterized in that: The method includes: S1, initialization phase: By gridding the plot, the number of grid cells N representing the dynamic changes of the plot is determined; Determine the number of sensors, the number of sensors is less than N, and all sensors are distributed in different grid cells; Initialize the sensor position, the initial value of the soil moisture state estimation vector and the initial value P of the error covariance estimation vector; the soil moisture state is the soil temperature and humidity state in the grid unit where the sensor is located; S2, model construction stage: Construct the calculation model and measurement model of the grid cells; The measurement model is used to combine the initialized sensor position and the initial value of the soil moisture state estimation vector to obtain the soil moisture state estimation vector x caused by the measurement at time k k ; According to x k , Q k , and P, we get the error covariance estimation vector at time k Among them, Q k is the process noise covariance matrix of the soil moisture state estimation vector at time k, k is the time index, k = 1, 2, ..., T', T' is the total number of sampling moments; The measurement model is used to combine the soil moisture state estimation vector x k , we get the moisture state measurement vector z caused by the measurement at time k k ; S3, Correction Phase: x k Denoise to obtain the denoised soil moisture state estimation vector according to Get H k ; Among them, H k is the Jacobian matrix of the nonlinear measurement function at time k; Using z k right Correction is performed to obtain the corrected soil moisture state estimation vector at time k Using H k right Correction is performed to obtain the corrected error covariance estimation vector P at time k k ; S4, optimization stage: Based on the estimated vector of soil moisture state at each moment after correction and the error covariance estimate vector P k , the objective function is constructed with the purpose of minimizing the estimation error, state estimation deviation and sensor operation cost. The objective function is minimized according to the set constraints. The objective function is optimized and solved using the optimization algorithm to obtain the optimal layout position of each sensor.

2. A multi-modal soil moisture sensor deployment method according to claim 1, characterized in that: The expression of the estimation model is: x k =f(x k-1 ,u k )+w k ; f(·) is the nonlinear measurement function; x k is the estimated vector of soil moisture state at time k, and The humidity value of the nth grid cell, The temperature value of the nth grid cell, n = 1, 2...N; u k is the control vector at time k; w k is the process noise at time k, w k Satisfies normal distribution, and the process noise covariance matrix of the soil moisture state estimation vector at time k in represents the variance of humidity process noise, σ w1,k represents the standard deviation of the humidity process noise, represents the variance of the temperature process noise, σ w2,k represents the standard deviation of temperature process noise; The expression of the measurement model is: z k =h(x k )+v k ; h(·) is the nonlinear measurement function; z k is the soil moisture state measurement vector at time k; v k is the observation noise at time k, v k Satisfies normal distribution, and the observation noise covariance matrix of the soil moisture state measurement vector at time k is in, represents the variance of the humidity observation noise at time k, σ v1,k represents the standard deviation of the humidity observation noise at time k, represents the variance of the temperature observation noise at time k, represents the standard deviation of the temperature observation noise at time k.

3. A multi-modal soil moisture sensor deployment method according to claim 2, characterized in that: In step S2, according to x k , Q k And P, get the error covariance estimation vector at time k The implementation methods include: S21, let x k The process noise w k = 0, and the soil moisture state estimation vector after de-noising at time k is obtained S22, according to Get the Jacobian matrix F of the nonlinear measurement function at time k-1 k-1 ,in, S23, according to F k-1 , Q k and P, we get the error covariance estimation vector at time k 4. A multi-modal soil moisture sensor deployment method according to claim 3, characterized in that: Where T is the transpose.

5. The method for deploying a multi-modal soil moisture sensor according to claim 3, characterized in that: In step S3, x k Denoise to obtain the denoised soil moisture state estimation vector The implementation is: Let x k The process noise w k = 0, and the soil moisture state estimation vector after de-noising at time k is obtained Using z k right Correction is performed to obtain the corrected soil moisture state estimation vector at time k The implementation methods include: S311, according to and R k , calculate the Kalman gain K at time k k ; S312, according to K k and z k ,right Correction is performed to obtain the corrected soil moisture state estimation vector 6. A multi-modal soil moisture sensor deployment method according to claim 5, characterized in that: Where T is the transpose.

7. A multi-modal soil moisture sensor deployment method according to claim 6, characterized in that: In step S3, H k right Correction is performed to obtain the corrected error covariance estimation vector P at time k k The implementation is: Among them, P k is the corrected error covariance estimate vector at time k.

8. The method for deploying a multi-modal soil moisture sensor according to claim 1, characterized in that: The constructed objective function is: Constraints: q∈M; n′<N; q is the sensor position vector combination, q={m1,m2,…,m n′ };m i is the position of sensor i, i=1,2……n′; Tr(P k ) is P k traces; x 目标 is the target state vector; γ is the weight factor of the cost function; n′ is the total number of sensors; O i is the operating cost of the i-th sensor; M is the combination vector of all combinations of sensor placement positions.

9. The multi-modal soil moisture sensor deployment method according to claim 1, characterized in that:

Images