CN111757266B - UAV data acquisition trajectory algorithm based on solar power supply type agricultural Internet of things - Google Patents

Abstract

The invention discloses a UAV data acquisition track algorithm based on a solar power supply type agricultural Internet of things, which comprises the steps of selecting a solar node as a cluster head aiming at a farmland UAV-WSN, then determining a clustering center by adopting an FCM clustering algorithm based on the optimal cluster number, and mapping the cluster head node by combining the clustering center to deploy the solar node; providing 3 different UAV track design schemes of WSN non-clustering, WSN clustering but no cluster head fusion data and WSN clustering and cluster head fusion data, and then calculating the number of traversal nodes and the clustering number required by the area under the 3 schemes according to the given CRLB; carrying out UAV path planning according to the node distribution uniformity, the node residual energy and the solar energy acquisition value to obtain 3 UAV path planning routes; and calculating the energy consumption of the nodes corresponding to the 3 routes, and selecting an optimal scheme through the energy consumption and the UAV flight distance. The UAV data acquisition path with low energy consumption, short path length and wide data acquisition range can be designed by the invention.

Description

UAV data acquisition trajectory algorithm based on solar power supply type agricultural Internet of things

Technical Field

The invention relates to the technical field of UAV trajectory design, in particular to a UAV data acquisition trajectory algorithm based on a solar power supply type agricultural Internet of things.

Background

A Wireless Sensor Network (WSN) is an important carrier for transforming an agricultural production mode and promoting the high-efficiency development of agriculture as an important component of an agricultural Internet of things system. The traditional WSN is generally composed of a large number of sensor nodes with small volume, less energy storage and low price and a plurality of data fusion centers, data are transmitted among the sensor nodes to the Fusion Center (FC) in a clustering mode, a non-clustering mode, a single-hop mode and a multi-hop mode, and for the static FC, early death of the sensor nodes and the energy hot area problem of the WSN are easily caused. In order to optimize the problem, UAV (unmanned aerial vehicle) is introduced into the WSN, network energy consumption of the WSN can be balanced by adopting the UAV crop mobile FC, and the problems that the mobile path and data acquisition stability are influenced by complex environmental factors of the farmland can be effectively avoided.

In recent years, distributed estimation in WSNs has attracted considerable research interest. A key feature of distributed estimation is that the sensors SN estimate scalar parameters, signal vectors, or other parameters in coordination with the limitations inherent in WSNs (e.g., energy-limited, bandwidth-limited, and measurement accuracy-limited), with the main design goal being to conserve resources while achieving acceptable estimation performance. One basic problem of UAV-WSN distributed estimation is UAV trajectory design, which needs to consider three factors: first, the UAV flight path is upper bounded for UAV self-limits; secondly, according to the requirement of distributed estimation precision, sufficiently large number of quantized data of nodes, such as minimum CRLB estimation parameters, need to be acquired; and finally, the WSN is prevented from dying nodes early, and the energy consumption of all the nodes is balanced. Therefore, the key to the problem of how to improve the performance of distributed estimation is how to find a path with low energy consumption, short path length and wide data acquisition range by the UAV.

Disclosure of Invention

The invention aims to overcome the defects and shortcomings of the prior art and provides a UAV data acquisition track algorithm based on a solar power supply type agricultural Internet of things.

The purpose of the invention is realized by the following technical scheme: a UAV data acquisition track algorithm based on a solar power supply type agricultural Internet of things comprises the following steps:

s1, aiming at a farmland UAV-WSN, selecting a solar node as a cluster head, then determining a clustering center by adopting an FCM clustering algorithm based on the optimal cluster number, and mapping the cluster head node by combining the clustering center to deploy the solar node;

s2, providing 3 different UAV trajectory design schemes of WSN non-clustering, WSN clustering but no cluster head fusion data and WSN clustering and cluster head fusion data, and then calculating the number of traversal nodes and the number of clusters required by the area under the 3 schemes according to the given CRLB;

s3, planning UAV paths according to the distribution uniformity of the nodes, the residual energy of the nodes and the solar energy collection value to obtain UAV path planning routes corresponding to the 3 schemes;

and S4, calculating node energy consumption corresponding to the 3 routes, and selecting an optimal scheme through the energy consumption and the UAV flight distance, wherein the optimal scheme is the final data acquisition track of the UAV.

Preferably, in step S1, the FCM clustering algorithm based on the optimal cluster number is used to determine the clustering center, and the process specifically includes:

s11, calculating a clustering number K according to the optimal clustering number formula:

wherein K is equal to the number m of cluster heads, and

n is the number of common nodes; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; c is the region side length d₂Is the distance between the cluster head and the UAV; e_mpEnergy consumption parameters of the multi-path attenuation model; e_elecEnergy consumption is transmitted for unit length;

s12, initializing a membership degree matrix u_ij：

Giving a clustering number K and a total number N of nodes, considering the minimum objective function in the cluster of each clustering group, and adopting a fuzzy set to set a membership set of each node to each cluster as U, namely the membership of the jth node belonging to the ith cluster is U_ijThe center of the group is c_iAnd is and

the jth node, i.e., node x_j；

In order to make the distance between each node and the center of the group to which the node belongs effective when determining the node to which the node belongs and the distance between the node and the center of the group to which the node does not belong ineffective when determining the node to which the node belongs, the importance of controlling the membership degree by considering a fuzzy number q is considered, so that the determination x_jMeter belonging to group iThe calculation value is:

when node x_jThe smaller the degree of membership to the i-th group, even if the distance is large, the fuzzy number q makes it possible to reduce the number of the groups

The distance is therefore equivalent to invalid when determining the position to which it belongs;

taking into account the initial degree of membership u_ijIntroducing a logistic chaotic model to carry out membership u on the influence of the final clustering effect of the FCM algorithm_ijThe mathematical model of the logistic chaotic model is as follows:

a_n+1＝μ*a_n*(1-a_n)

wherein, a_nValue generated for the nth iteration, a₁Is a random value in the range of (0-1); a is_n+1A value generated for the (n +1) th iteration; mu is an element of [0 to 4 ∈]；a_nThe value generated by iteration is in a chaotic state and has ergodicity;

s13, adopting Lagrange multiplier and respectively pairing u_ijAnd c_iAnd (3) calculating a partial derivative according to the following calculation formula:

wherein, c_lIs the cluster center of the first cluster;

s14, for all nodes and all clusters, the objective function is:

solving the condition of satisfying the constraint

Objective function of time J^(t)Minimum value of (1), J^(t)Representing the objective function value at the t iteration;

s15, if J^(t)-J^(t-1)I < epsilon, where epsilon is the minimum deviation value, J^(t-1)Representing the objective function value at the t-1 th iteration, the algorithm ends, cluster center c_iNamely the clustering center; otherwise, go back to step S13.

Further, in step S1, the solar energy node is deployed by mapping the cluster head node with the cluster center, specifically:

selecting a node as a clustering center by adopting nearest neighbor mapping, and taking c_iDenotes the cluster center of the ith cluster, x_ijRepresenting the jth intra-cluster node of the ith cluster, and selecting the node as follows:

min||c_i-x_ij||

and after the position mapping is completed, selecting a mapping node to deploy for the solar node.

Preferably, in step S2, the design scheme of UAV trajectories without clustering by WSNs specifically includes: all the traversal nodes are in direct communication with the UAV, the UAV sends information to the traversal nodes, and the traversal nodes send data to the UAV;

the design scheme of the UAV trajectory with WSN clustering and no cluster head fused with data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, and the cluster head nodes forward the data to the UAV;

the design scheme of the UAV trajectory of the WSN clustered and cluster head fused data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, the cluster head fuses the data, and then the data are forwarded to the UAV.

Preferably, in step S2, the number of traversal nodes and the number of clusters required for the area under 3 schemes are calculated according to the given CRLB, as follows:

(1) for WSNs not clustered:

a cluster-free adaptive quantization scheme is proposed:

calculating CRLB under the condition that WSN is not clustered through a cluster-free distributed estimation formula:

wherein N is the number of nodes;

indicating the normalized frequency, P (tau), of the sensor used as a threshold_n-1K Δ) is the quantization threshold τ of the n-1 th node_n-1Probability at k Δ; p is a radical of_w(x) Is w_kA Probability Density Function (PDF); f_w(x) Is w_kComplementary Cumulative Density Function (CCDF); θ is an estimated environmental physical quantity of the physical world, i.e. a parameter estimated from quantized data received from the UAV;

is an estimate of the parameter θ; w is a_kIs the observed noise of the kth node in the region; Δ represents a parametric quantization step; k represents the kth node in the area; tau is_kAccumulating early transmission data for a node from other sensor nodes and using the accumulated value as a threshold for its 1-bit quantifier;

then calculating the number of traversal nodes required by the region which is closest to and smaller than the given region CRLB by adopting a dichotomy;

(2) for WSN clustering but no fusion data: the calculation of locally estimated CRLB and the required traversal node number of the area is the same as the calculation of the WSN non-clustering;

(3) for WSN clustering and cluster head fusion data:

a clustering adaptive quantization scheme is proposed: the globally estimated CRLB may be represented by a weighted sum of the locally estimated CRLB;

calculating CRLB under the condition that WSN is clustered and cluster heads fuse data through a clustering distributed estimation formula:

wherein,

a weighting coefficient representing a local estimate;

and

distributed estimation for the ith cluster and the jth cluster respectively;

then calculating the number of members in the cluster when the CRLB of a given area is met under the condition of different cluster numbers, wherein the maximum cluster number is the number of solar nodes in the area, and the minimum cluster number is 2;

and comparing the total energy consumption of the WSN under different cluster numbers, and selecting the cluster number and the member number in the cluster under the condition of minimum energy consumption.

Preferably, for node distribution uniformity:

calculating the distribution uniformity f of the nodes through a distribution uniformity potential function_rWherein, the distribution uniformity potential function calculation formula of the r-th node is as follows:

in the formula, n is the total number of nodes in the region; l represents the extension number of other nodes in the area; k represents the kth node in the area; s represents the s-th continuation node of the node in the region; x is the number of_rI.e. represents the r-th node;

to a node x_rThe s-th continuation node of the k-th node in the area;

the uniformity potential function calculation formula of all nodes in the region is as follows:

when the distribution of the nodes in the region is more uniform, the total uniformity potential function value is smaller;

for node residual energy:

calculating the weight of the residual energy of the common node by the following calculation formula:

in the formula, E_startResidual energy is the node; e_RestIs the initial energy of the node;

calculating the weight of the residual energy of the solar node by the following calculation formula:

in the formula, E_ForeTo predict the energy harvesting value, the value E can be harvested from solar energy as follows_havEstimating;

when the residual energy of the node is more, the weight of the residual energy is smaller;

for solar energy collection values:

the solar energy collection can be quantized into a solar energy collection energy model represented by a trigonometric function of different peak values according to factors such as time, weather and position, a solar energy collection value is calculated through the model, and according to the rising and falling conditions of the sun, the solar energy collection is started at 6 am and finished at 6 pm;

the solar energy collection energy model is as follows:

E_hav＝z*sin(x)-rand

in the formula, E_havCollecting a value for solar energy; z is the solar energy collection peak value under different conditions of sunny days and cloudy days; x is solar energy collection time point conversionThe value of the one or more of the one,

time is the number of seconds from the current time to 6 am; rand is a random value that simulates factors that affect energy harvesting.

Further, the UAV path planning process according to the node distribution uniformity, the node residual energy, and the solar energy collection value is as follows:

s31, selecting a traversal node of the UAV based on a minimized cost equation, wherein the minimized cost equation is as follows:

in the formula, N_iThe number of the nodes selected in the ith cluster is obtained; w is a_ijThe residual energy weight of the jth node in the ith cluster is obtained; f. of_ijThe uniformity potential function value of the jth node in the ith cluster is obtained;

s32, considering the UAV path planning problem of the traversal node obtained by considering the minimum value of the cost function obtained by the simulated annealing algorithm, which may be regarded as a traveling quotient problem, that is, a shortest path planning problem that starts from an initial city, traverses all other cities, and finally returns to the initial city, and solving the traveling quotient problem means traversing all nodes except the initial node once in a connected graph, and planning a Hamilton loop except the shortest one, where a mathematical model defining the traveling quotient problem is as follows:

given a connectivity graph P ═ (C)_P,L_P) Wherein, C_PFor a set of node numbers, L_PIs a collection of edges between nodes; set node set C_P1,2_P＝{(r,j,w_rj)|r,j∈C_P,w_rj∈R⁺}，w_rjIs the distance between the r-th node and the j-th node, the Hamilton loop is a set of edges that contains all the nodes

Including the initial nodeTwice for each node, once for each other node, m_r,m_j∈C_PAnd r ≠ j;

for edge set L_PThe selection of the element(s) is represented as follows:

the objective function of the Hamilton loop is then expressed as follows:

assuming that the previous state of the system is x (n), the state of the system is changed into x (n)) +1 according to the set index, the energy of the system is correspondingly changed from E (n) to E (n +1), and the receiving probability P of the system from x (n) to x (n)) +1 is defined as:

the system is different according to different scene requirements, and global area search and local cluster search are carried out in the system;

s33, solving the minimum value of the minimized cost equation in the step S31 by adopting a simulated annealing algorithm, wherein the smaller the cost function f of the node is, the greater the probability of selecting the node is, and thus the selected traversal node in the region is solved;

and S34, calculating the minimum value of the objective function of the Hamilton loop according to the selected traversal nodes in the area to obtain the shortest path of the UAV.

Further, the process of step S33 is as follows:

s331, in order to make the cooling process faster, firstly determining a better initial solution w through a Monte Carlo algorithm, and calculating an appropriate value f (w);

s332, considering that the selected nodes are discrete, selecting any one element in the random updating solution to generate a new solution w 'to calculate a new proper value f (w');

s333, then calculating an adaptation increment Δ f ═ f (w ') -f (w), and determining whether Δ f is greater than 0, if Δ f is less than or equal to 0, accepting a new solution according to Metropolis criterion, and if Δ f is greater than 0, accepting a new solution w ═ w ', f (w) ═ f (w ');

s334, slowly cooling, judging whether a termination condition is met, if not, returning to the step S332 again;

if so, the operation is ended, and the final solution is the minimum value of the minimization cost equation.

Preferably, in step S4, the node energy consumption models of the UAV trajectory planning route corresponding to the 3 schemes are as follows:

(1) based on non-cluster distributed estimation, the WSN is not clustered:

energy consumption of each node:

wherein,

energy consumption to receive UAV information for a node;

sending data to the UAV energy consumption for the node; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; r_cFor communication radius, to ensure full connectivity of the network, R_c＝2*R_S，R_SIs the sensing radius;

(2) based on non-cluster distributed estimation, the WSN forms clusters but the cluster heads do not fuse the energy consumption model of data:

let the number of cluster members of the ith cluster be m_iEnergy consumption of cluster head node

The method comprises the following steps:

(a) receiving UAV clustering information energy consumption

(b) Energy consumption for sending cluster information to cluster members

(c) Energy consumption for receiving cluster member data

(d) Energy consumption for sending data to UAV

Energy consumption of cluster members

The method comprises the following steps:

(e) energy consumption for receiving cluster head clustering information

(f) Energy consumption for receiving other node broadcast quantized data

(g) Energy consumption for sending data to cluster head

Finally, the energy consumption of the cluster head nodes and the energy consumption of the cluster members are respectively as follows:

(3) based on clustered distributed estimation, the WSN is clustered and the cluster heads fuse the energy consumption model of the data:

energy consumption of cluster head node

Energy consumption of member of Hezhou cluster

Respectively as follows:

wherein m is_i*E_daFusing data energy consumption for cluster heads;

the optimal scheme is selected through energy consumption and UAV flight distance, and specifically comprises the following steps:

calculating the total energy consumption E of the network based on the node energy consumption model_Total；

Then compare the total energy consumption E of the 3 schemes_TotalAnd selecting an optimal scheme according to the UAV flight distance Len corresponding to the UAV trajectory planning route, wherein the scheme selection function is as follows:

min f＝a₁E_Total+a₂Len

in the formula, a₁Is the energy consumption weight; a is₂Is a path length weight; wherein a is₁And a₂According to the actual situation, if the network life cycle proportion is larger, the setting a is₁The larger, and similarly, the more the UAV path length is constrained, the setting a₂The larger.

Compared with the prior art, the invention has the following advantages and effects:

(1) according to the UAV data acquisition track algorithm based on the solar power supply type agricultural Internet of things, the optimal cluster number-based FCM clustering algorithm is combined to select the deployment position of the solar node, then 2 modules are designed in the UAV track design part, the UAV data acquisition track algorithm comprises a node selection module and a UAV path planning module, finally the optimal scheme is selected by comparing the shortest path length and the node energy consumption obtained by different schemes, and finally the UAV data acquisition path with low energy consumption, short path length and wide data acquisition range is designed, so that the UAV data acquisition can be balanced in the UAV flight path length, the energy consumption problem of the sensor node and the distribution uniformity of the data acquisition node, the distributed estimation performance is improved, and the network life is prolonged.

(2) In the UAV data acquisition trajectory algorithm, in order to ensure the accuracy of the estimated data acquired by farmland information, the UAV needs to acquire a sufficient number of nodes to ensure that the required data can be restored within an allowable error, so that the number of the nodes of the required acquired data is calculated based on a given CRLB (cross-correlation between the nodes and the data), and the accuracy of restoring the data is ensured. In addition, whether clustering fusion data cause different calculation formulas for solving traversal nodes by the CRLB, and whether clustering fusion data directly affect the UAV flight path length and the WSN total energy consumption, the invention also provides 3 UAV track design schemes for calculating the number of nodes respectively, comparing resources consumed by different schemes, and selecting the optimal scheme from the calculated nodes, so that the designed UAV data acquisition track can better meet the requirements of data acquisition stability and resource saving.

Drawings

Fig. 1 is a flowchart of the UAV data acquisition trajectory algorithm based on the internet of things for solar powered agriculture of the present invention.

FIG. 2 is a simulation diagram of a farm environment with nodes deployed.

Figure 3 is a UAV trajectory planning circuit diagram for scenario 1.

Fig. 4 is a schematic diagram of the energy consumption of the node in the case of the 100 th round of operation scheme 1 for 4 farmland areas.

Figure 5 is a UAV trajectory planning circuit diagram for scenario 2.

Fig. 6 is a schematic diagram of the energy consumption of the node in case of the 60 th round of operation scheme 2 for 4 farmland areas.

Figure 7 is a UAV trajectory planning circuit diagram for scenario 3.

Fig. 8 is a schematic diagram of the energy consumption of the node in case of the 10 th run of scheme 3 for 4 farmland areas.

Detailed Description

The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.

Examples

The embodiment discloses a UAV data acquisition track algorithm based on a solar power supply type agricultural Internet of things, which comprises the steps of firstly determining and selecting a solar node as a cluster head, adopting an FCM clustering algorithm based on the optimal cluster number and combining with clustering center mapping to select a solar node deployment position, then dividing a UAV track design part into 2 modules for design, wherein the UAV track design part comprises a node selection module and a UAV path planning module, and finally selecting an optimal scheme by comparing the shortest path length and the node energy consumption obtained by different schemes. As shown in fig. 1, the steps are as follows:

s1, because the energy consumption of the cluster head is far larger than that of a common node, in consideration of the energy consumption balance problem, a solar node in a farmland UAV-WSN is selected as the cluster head, then a clustering center is determined by adopting an FCM (Fuzzy C-Means) clustering algorithm based on the optimal cluster number, and the cluster head node is mapped by combining the clustering center to deploy the solar node.

The farmland environment is divided into a plurality of regions, the areas of the regions are the same, and the distance between the regions is ignored. The basic idea of the FCM (Fuzzy C-Means) clustering algorithm is to minimize an objective function of nodes divided into the same class, apply the FCM clustering algorithm to the WSN clusters, and divide the nodes in an area into K clusters according to similarity, so that the energy consumption of cluster distribution inequality is effectively reduced.

The optimal cluster number-based FCM clustering algorithm is adopted to map cluster head nodes in combination with a clustering center to perform deployment of the solar nodes, and the process specifically comprises the following steps:

s11, calculating a clustering number K according to the optimal clustering number formula:

wherein K is equal to the number m of cluster heads, and

n is the number of common nodes; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; c is the region side length; d₂Is the distance between the cluster head and the UAV; e_mpEnergy consumption parameters of the multi-path attenuation model; e_elecEnergy consumption is transmitted for unit length;

s12, initializing a membership matrix:

given the cluster number K and the total number N of nodes, considering the minimum objective function in the cluster of each cluster group, adopting a fuzzy set to set the membership set of each node to each cluster group as U, namely the jth node (namely the node x)_j) Membership of u to the ith group_ijThe center of the group is c_iAnd is and

in order to make the distance between each node and the center of the group to which the node belongs effective when determining the node to which the node belongs and the distance between the node and the center of the group to which the node does not belong ineffective when determining the node to which the node belongs, the importance of controlling the membership degree by considering a fuzzy number q is considered, so that the determination x_jThe calculated values belonging to group i are:

when node x_jThe smaller the degree of membership to the i-th group, even if the distance is large, the fuzzy number q makes it possible to reduce the number of the groups

The distance is therefore equivalent to invalid when determining the position to which it belongs;

taking into account the initial degree of membership u_ijIntroducing a logistic chaotic model to carry out membership u on the influence of the final clustering effect of the FCM algorithm_ijThe mathematical model of the logistic chaotic model is as follows:

a_n+1＝μ*a_n*(1-a_n)

wherein, a_nGenerating for the nth iterationValue of a₁Is a random value in the range of (0-1); a is_n+1A value generated for the (n +1) th iteration; mu is an element of [0 to 4 ∈]Generally, the value is 4; a is_nThe value generated by iteration is in a chaotic state and has ergodicity; a is_nAs element u in a membership matrix_ijBy iterating continuously, u can be initialized_ijInitializing;

s13, adopting Lagrange multiplier and respectively pairing u_ijAnd c_iAnd (3) calculating a partial derivative according to the following calculation formula:

wherein, c_lIs the cluster center of the first cluster;

s14, for all nodes and all clusters, the objective function is:

solving the condition of satisfying the constraint

Objective function of time J^(t)Minimum value of (1), J^(t)Representing the objective function value at the t iteration;

s15, if J^(t)-J^(t-1)I < epsilon, where epsilon is the minimum deviation value, J^(t-1)Representing the objective function value at the t-1 th iteration, the algorithm ends, cluster center c_iNamely the clustering center; otherwise, returning to the step S13;

s16, adopting nearest neighbor mapping to select nodes as clustering centers, and taking c as_iDenotes the cluster center of the ith cluster, x_ijRepresenting the jth intra-cluster node of the ith cluster, and selecting the node as follows:

min||c_i-x_ij||

and after the position mapping is completed, selecting a mapping node to deploy for the solar node.

And S2, providing 3 different UAV trajectory design schemes of WSN non-clustering, WSN clustering but cluster head non-fusion data and WSN clustering and cluster head fusion data, and then calculating the number of traversal nodes and the number of clustering required by the area under the 3 schemes according to the given CRLB.

The design scheme of the UAV trajectory without clustering the WSN specifically comprises the following steps: all the traversal nodes are in direct communication with the UAV, the UAV sends information to the traversal nodes, and the traversal nodes send data to the UAV.

The design scheme of the UAV trajectory with WSN clustering and no cluster head fused with data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, and the cluster head nodes forward the data to the UAV.

The design scheme of the UAV trajectory of the WSN clustered and cluster head fused data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, the cluster head fuses the data, and then the data are forwarded to the UAV.

Based on the given CRLB, the number of traversal nodes and the number of clusters required for the region under 3 schemes are calculated as follows:

(1) for WSNs not clustered:

a cluster-free adaptive quantization scheme is proposed: consider a fusion center consisting of M regions, each region deploying N, and a UAV_iA wireless sensor network of (1, 2.. M) sensor nodes, each sensor node x_jAnd carrying out noise observation on the unknown parameter theta. Suppose observation noise w_j,j＝1,2,...,N_iMean 0 and variance σ²Each sensor node observes an unknown and determined scalar parameter: x is the number of_j＝θ+w_j,j＝1,2,...,N_i. Assume that all sensors are due to bandwidth and power constraintsIt needs to be observed by { x }_jQuantized into one-bit binary data b_jB, the quantization formula of_j＝sgn{x_j-τ_j-1Then quantized data received from UAV b_jThe unknown parameter theta is estimated. Each sensor node accumulates early transmission data from other sensor nodes and uses the accumulated value as a threshold for its 1-bit quantifier

k represents the kth node in the area; Δ represents a parametric quantization step size. The accumulated value may show convergence around the unknown parameter θ. Deriving a Maximum Likelihood Estimator (MLE) for use in the fusion center based on the adaptive quantization scheme and deriving an estimate of the unknown parameter θ without clustering

CRLB, i.e. best achievable estimation accuracy of the maximum likelihood estimator.

The maximum likelihood estimator is a non-cluster distributed estimation formula, and the CRLB under the condition that the WSN is not clustered is calculated through the non-cluster distributed estimation formula:

wherein N is the number of nodes;

indicating the normalized frequency, P (tau), of the sensor used as a threshold_n-1K Δ) is the quantization threshold τ of the n-1 th node_n-1Probability at k Δ; p is a radical of_w(x) Is w_kA Probability Density Function (PDF); f_w(x) Is w_kComplementary Cumulative Density Function (CCDF); θ is an estimated environmental physical quantity of the physical world, i.e. a parameter estimated from quantized data received from the UAV; w is a_kIs the observed noise of the kth node in the region;

then calculating the number of traversal nodes required by the region which is closest to and smaller than the given region CRLB by adopting a dichotomy;

(2) for WSN clustering but no fusion data: the calculation of locally estimated CRLB and the required traversal node number of the area is the same as the calculation of the WSN non-clustering;

(3) for WSN clustering and cluster head fusion data:

a clustering adaptive quantization scheme is proposed: the globally estimated CRLB may be represented by a weighted sum of the locally estimated CRLB;

calculating CRLB under the condition that WSN is clustered and cluster heads fuse data through a clustering distributed estimation formula:

wherein,

a weighting coefficient representing a local estimate;

and

distributed estimation for the ith cluster and the jth cluster respectively;

then calculating the number of members in the cluster when the CRLB of a given area is met under the condition of different cluster numbers, wherein the maximum cluster number is the number of solar nodes in the area, and the minimum cluster number is 2;

and comparing the total energy consumption of the WSN under different cluster numbers, and selecting the cluster number and the member number in the cluster under the condition of minimum energy consumption.

And S3, carrying out UAV path planning according to the distribution uniformity of the nodes, the residual energy of the nodes and the solar energy acquisition value to obtain a UAV path planning route corresponding to the 3 schemes.

Wherein, for node distribution uniformity: since the number of traversal nodes is calculated based on the CRLB of the distributed estimation, the UAV does not access all the nodes to collect data, and it is necessary to ensure that the data information repetition rate of the area collected by the data nodes collected by the UAV is low, that is, the nodes are distributed sufficiently and uniformly in the area.

Here, the node distribution uniformity f is calculated by a distribution uniformity potential function_rWherein, the distribution uniformity potential function calculation formula of the r-th node is as follows:

in the formula, n is the total number of nodes in the region; l represents the extension number of other nodes in the area; k represents the kth node in the area; s represents the s-th continuation node of the node in the region; x is the number of_rI.e. represents the r-th node;

to a node x_rThe s-th continuation node of the k-th node in the area;

the uniformity potential function calculation formula of all nodes in the region is as follows:

when the distribution of the nodes in the region is more uniform, the total uniformity potential function value is smaller;

for node residual energy: considering the life cycle of the WSN, the selected traversal nodes not only need to meet the distribution uniformity, but also need to balance the residual energy of the nodes. The nodes are divided into common nodes and solar nodes, and because the solar nodes have an energy collection function, a predicted energy collection value should be added when the weight of the residual energy is calculated.

Here, the ordinary node residual energy weight is calculated by the following calculation formula:

in the formula, E_startResidual energy is the node; e_RestIs the initial energy of the node;

calculating the weight of the residual energy of the solar node by the following calculation formula:

in the formula, E_ForeTo predict the energy harvesting value, the value E can be harvested from solar energy as follows_havEstimation, e.g. may be taken_havZ sin (x) in (a).

When the residual energy of the node is more, the weight of the residual energy is smaller;

for solar energy collection values: the solar energy collection can be quantized into a solar energy collection energy model represented by a trigonometric function of different peak values according to the factors of time, weather and position, and a solar energy collection value is calculated through the model. According to the embodiment, the following solar energy collection model is designed according to sunny days and cloudy days, and according to the rising and falling conditions of the sun in China, solar energy collection is started at 6 am and finished at 6 pm.

The solar energy collection energy model is as follows:

E_hav＝z*sin(x)-rand

in the formula, E_havCollecting a value for solar energy; z is the solar energy collection peak value under different conditions (sunny days and cloudy days); x is a conversion value of a solar energy collecting time point,

time is the number of seconds from the current time to 6 am; rand is a random value that simulates factors that affect energy harvesting.

The process of planning the UAV path according to the node distribution uniformity, the node residual energy, and the solar energy collection value is as follows:

s31, selecting a traversal node of the UAV based on a minimized cost equation, wherein the minimized cost equation is as follows:

in the formula, N_iThe number of the nodes selected in the ith cluster is obtained; w is a_ijThe residual energy weight of the jth node in the ith cluster is obtained; f. of_ijThe uniformity potential function value of the jth node in the ith cluster is obtained;

when the remaining energy of a node is more, the smaller the weight coefficient is, the smaller f is, and the greater the probability of selecting the node is.

S32, considering the UAV path planning Problem of the traversal node obtained by considering the minimum value of the cost function obtained by the simulated annealing algorithm, may be regarded as a Traveling Salesman Problem (TSP), that is, a shortest path planning Problem of starting from an initial city, traversing all other cities, and finally returning to the initial city, where the solving of the Traveling Salesman Problem is to traverse all nodes except the initial node once in a connected graph, and plan a Hamilton loop except the shortest node. Here, the mathematical model defining the traveler problem is as follows:

given a connectivity graph P ═ (C)_P,L_P) Wherein, C_PFor a set of node (city) numbers, L_PIs a collection of edges (distances) between nodes; set node set C_P1,2_P＝{(r,j,w_rj)|r,j∈C_P,w_rj∈R⁺}，w_rjIs the distance between the r-th node and the j-th node, the Hamilton loop is a set of edges that contains all the nodes

Including the start node twice (since other nodes start from the start node and finally return to the start node), the rest once, m_r,m_j∈C_PAnd r ≠ j;

for edge set L_PThe selection of the element(s) is represented as follows:

the objective function of the Hamilton loop is then expressed as follows:

assuming that the previous state of the system is x (n), the system changes to x (n)) +1 according to a set index (such as gradient descent), the energy of the system correspondingly changes from E (n) to E (n +1), and the acceptance probability P of the system changing from x (n) to x (n)) +1 is defined as:

the system therein needs to be different according to different scenes, such as global area search and local intra-cluster search adopted in this embodiment.

S33, solving the minimum value of the minimized cost equation in the step S31 by adopting a simulated annealing algorithm (SA), wherein the smaller the cost function f of the node is, the greater the probability of selecting the node is, and thus the selected traversal node in the region is solved;

the simulated annealing algorithm is derived from the solid state annealing principle and comprises two parts, namely the Metropolis algorithm and the annealing process. The Metropolis algorithm is how to make a jump out under the condition of local optimal solution, and is the basis of annealing. The process of solving the minimized cost equation by the simulated annealing algorithm is as follows:

s331, in order to make the cooling process faster, firstly determining a better initial solution w through a Monte Carlo algorithm, and calculating an appropriate value f (w);

s332, considering that the selected nodes are discrete, selecting any one element in the random updating solution to generate a new solution w ', and calculating a new fitness f (w') to ensure that a global optimum is obtained;

s333, then calculating an adaptation increment Δ f ═ f (w ') -f (w), and determining whether Δ f is greater than 0, if Δ f is less than or equal to 0, accepting a new solution according to Metropolis criterion, and if Δ f is greater than 0, accepting a new solution w ═ w ', f (w) ═ f (w ');

s334, slowly cooling, judging whether a termination condition is met, if not, returning to the step S332 again; the termination condition can set the iteration times or the precision of the solution required to be solved, and the stop is realized when the iteration times or the precision requirement is met.

If so, the operation is ended, and the final solution is the minimum value of the minimization cost equation.

And S34, calculating the minimum value of the objective function of the Hamilton loop according to the selected traversal nodes in the area to obtain the shortest path of the UAV.

And S4, calculating node energy consumption corresponding to the 3 routes, and selecting an optimal scheme through the energy consumption and the UAV flight distance, wherein the optimal scheme is the final data acquisition track of the UAV.

The node energy consumption models of the UAV trajectory planning route corresponding to the 3 schemes are respectively as follows:

(1) based on non-cluster distributed estimation, the WSN is not clustered:

energy consumption of each node:

wherein,

energy consumption to receive UAV information for a node;

sending data to the UAV energy consumption for the node; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; r_cFor communication radius, to ensure full connectivity of the network, R_c＝2*R_S，R_SIs the sensing radius;

(2) based on non-cluster distributed estimation, the WSN forms clusters but the cluster heads do not fuse the energy consumption model of data:

let the number of cluster members of the ith cluster be m_iEnergy consumption of cluster head node

The method comprises the following steps:

(a) receiving UAV clustering informationEnergy consumption

(b) Energy consumption for sending cluster information to cluster members

(c) Energy consumption for receiving cluster member data

(d) Energy consumption for sending data to UAV

Energy consumption of cluster members

The method comprises the following steps:

(e) energy consumption for receiving cluster head clustering information

(f) Energy consumption for receiving other node broadcast quantized data

(g) Energy consumption for sending data to cluster head

Finally, the energy consumption of the cluster head nodes and the energy consumption of the cluster members are respectively as follows:

(3) based on clustered distributed estimation, the WSN is clustered and the cluster heads fuse the energy consumption model of the data:

energy consumption of cluster head node

Energy consumption of member of Hezhou cluster

Respectively as follows:

wherein m is_i*E_daFusing data energy consumption for cluster heads;

the optimal scheme is selected through energy consumption and UAV flight distance, and specifically comprises the following steps:

calculating the total energy consumption E of the network based on the node energy consumption model_Total；

Then compare the total energy consumption E of the 3 schemes_TotalAnd selecting an optimal scheme according to the UAV flight distance Len corresponding to the UAV trajectory planning route, wherein the scheme selection function is as follows:

min f＝a₁E_Total+a₂Len

in the formula, a₁Is the energy consumption weight; a is₂Is a path length weight; wherein a is₁And a₂According to the actual situation, if the network life cycle proportion is larger, the setting a is₁The larger, and similarly, the more the UAV path length is constrained, the setting a₂The larger. In this embodiment, a is set in consideration of the larger lifetime of the network₁＝100，a₂＝0.01。

In order to verify the algorithm of the embodiment, the embodiment also adopts MATLAB R2019a application software for simulation, and a simulation experiment firstly simulates modeling of a farmland environment and node deployment. As can be seen from fig. 2, the field environment is defined as a rectangular area of 200m × 200m, divided into 4 field areas, and the unmanned aerial vehicle (rectangular portion) is disposed at the middle position of the entire rectangular area, ignoring the distance between the areas, where "Δ" represents a solar node.

TABLE 1

E_elecFor transmission of energy per unit length, E_fsEnergy consumption parameters for free space energy consumption models, E_mpEnergy consumption parameters for multipath fading models, E_daTo fuse the energy loss per bit.

And then analyzing the execution conditions of the node selection module and the UAV trajectory planning module of the 3 schemes through the same given CRLB, and then analyzing the UAV-WSN energy consumption condition of running multiple rounds.

Wherein, the parameter table of the simulated annealing algorithm adopted by the track planning part is as shown in the following table 2:

TABLE 2

(1) Regions were not clustered (scheme 1):

fig. 3 is a plan view of the UAV trajectory planning circuit in scheme 1, and it can be seen from fig. 3 that the nodes represented by black solid points are uniformly distributed in the area where the nodes are located. The calculated path length of the wheel UAV is 1200.17m, and the total energy consumption E of the network_Total0.0049J. When the number of nodes to traverse is small, the path length of the UAV is short, and in this scheme, the nodes only need to communicate with the UAV, and the overall energy consumption of the network is minimal, so the scheme is generally adopted when considering that the network lifetime is heavier than a certain value.

Fig. 4 shows energy consumption of Nodes in 4 farmland areas under the condition of the 100 th round of operation scheme 1, where in fig. 4, the abscissa is a node (Nodes), and the ordinate is a node residual energy (energy rest), where a gray column represents a solar node residual energy, and energy collection of a solar node is considered, so that the residual energy of the solar node is not considered in calculating network energy consumption balance, and then a standard deviation matrix of the current common node residual energy is as follows:

Std＝[4.3636e-04,7.9359e-04；8.3618e-04,4.6303e-04]

the maximum and minimum matrix of the residual energy of the common nodes in the region is as follows:

Min/Max＝[6.2e-03/7.9e-03,6.3e-03/8.9e-03；5.3e-03/8.5e-03,7.2e-03/9.0e-03]

by comparing the value of the standard deviation matrix for removing the residual energy of the solar nodes with the value of the maximum and minimum matrix for removing the residual energy of the common nodes, the energy consumption balance of the network in each area is better.

(2) Area clustering, cluster head only forwards un-fused data (scheme 2): calculated, the path length of the UAV in the round is 813.5m, and the total energy consumption E of the network_Total0.0213J. Fig. 5 is a schematic diagram of a UAV trajectory planning route according to scheme 2, and it can be seen from fig. 5 that the UAV traverses the cluster heads for data acquisition, and the path length of the UAV depends on the number of cluster heads having intra-cluster selection nodes, so that even when the number of traversal nodes is large enough, the path length of the UAV is substantially unchanged, and the total energy consumption of the nodes has a direct influence on the number of cluster heads having intra-cluster selection nodes and the number of intra-cluster selection nodes. From this wheel, it can be seen that the UAV path length for scenario 2 is reduced by about 1/3 compared to scenario 1, but the network total energy consumption is more than 4 times that of scenario 1.

Figure 6 shows the node energy consumption for the case of run 2 run 60 for 4 farmland areas. The standard deviation matrix of the residual energy of the common nodes in the area without the solar nodes is as follows:

Std＝[7.3593e-04,6.3812e-04；5.8002e-04,5.3920e-04]

the maximum and minimum matrix of the residual energy of the common nodes in the region is as follows:

Min/Max＝[4.6e-03/8.6e-03,6.2e-03/8.7e-03；4.3e-03/6.8e-03,7.1e-03/9.1e-03]

and comparing the value of the standard deviation matrix for removing the residual energy of the solar nodes with the value of the maximum and minimum matrix for removing the residual energy of the common nodes, so that the energy consumption balance of the network in each area is better.

(3) Region clustering, cluster head fusion data (scheme 3): different from the two previous schemes in which the selection of the nodes is directly performed in the whole area, under the condition of the scheme, the number of formed clusters and the number of traversal nodes in the clusters directly influence the CRLB of the distributed estimation, so the scheme adopts a traversal algorithm to calculate the number of formed clusters and the number of traversal nodes in the clusters. And selecting traversal nodes in each cluster based on the calculated cluster forming number and the calculated traversal node number in the cluster.

Calculated, the path length of the UAV in this round is 535.5m, the total energy consumption E of the network_Total0.0374J. Fig. 7 is a UAV trajectory planning circuit diagram of the scheme 3, and it can be seen from the diagram that the path length for UAV trajectory planning based on clustered distributed estimation is smaller than those of the first two schemes, but the node distribution uniformity is worse than that of the first two schemes, and the total energy consumption of the network is increased by a lot.

Figure 8 shows the node energy consumption for the 10 th run scenario 3 for 4 farmland areas. The standard deviation matrix of the residual energy of the common nodes in the area without the solar nodes is as follows:

Std＝[9.4807e-04,7.8176e-04；2.8133e-04,9.2428e-04]

the maximum and minimum matrix of the residual energy of the common nodes in the region is as follows:

Min/Max＝[5.7e-03/9.5e-03,6.4e-03/8.4e-03；6.6e-03/7.8e-03,5.8e-03/9.0e-03]

when the cluster number and the member number in the cluster are calculated, if the precision of the data needing to be acquired is too high, namely the given CRLB is too small, the cluster number and the member number in the cluster are larger, and the member number in the cluster of some clusters cannot meet the obtained member number in the cluster, so the method is slightly inferior to the first two schemes in the aspect of balance.

A comparison table of the indexes of the 3 schemes is shown in Table 3:

TABLE 3

Simulation results show that: under the same given CRLB, the region is clustered, the cluster head fusion data scheme is optimal in shortening the path length of the UAV, and the total network energy consumption of the region non-clustering scheme is minimum. Therefore, in selecting a scheme, the weight relationship between the network life cycle and the UAV path length can be weighted by means of a scheme selection function.

The techniques described herein may be implemented by various means. For example, these techniques may be implemented in hardware, firmware, software, or a combination thereof. For a hardware implementation, the processing modules may be implemented within one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Programmable Logic Devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, micro-controllers, electronic devices, other electronic units designed to perform the functions described herein, or a combination thereof.

For a firmware and/or software implementation, the techniques may be implemented with modules (e.g., procedures, steps, flows, and so on) that perform the functions described herein. The firmware and/or software codes may be stored in a memory and executed by a processor. The memory may be implemented within the processor or external to the processor.

Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.

The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (2)

1. A UAV data acquisition track algorithm based on a solar power supply type agricultural Internet of things is characterized by comprising the following steps:

s1, aiming at a farmland UAV-WSN, selecting a solar node as a cluster head, then determining a clustering center by adopting an FCM clustering algorithm based on the optimal cluster number, and mapping the cluster head node by combining the clustering center to deploy the solar node;

s2, providing 3 different UAV trajectory design schemes of WSN non-clustering, WSN clustering but no cluster head fusion data and WSN clustering and cluster head fusion data, and then calculating the number of traversal nodes and the number of clusters required by the area under the 3 schemes according to the given CRLB;

s3, planning UAV paths according to the distribution uniformity of the nodes, the residual energy of the nodes and the solar energy collection value to obtain UAV path planning routes corresponding to the 3 schemes;

s4, calculating node energy consumption corresponding to the 3 routes, and selecting an optimal scheme through the energy consumption and the UAV flight distance, wherein the optimal scheme is the final data acquisition track of the UAV;

in step S1, determining a clustering center by using an FCM clustering algorithm based on the optimal cluster number, wherein the process specifically includes:

s11, calculating a clustering number K according to the optimal clustering number formula:

wherein K is equal to the number m of cluster heads, and

n is the number of common nodes; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; c is the region side length d₂Is the distance between the cluster head and the UAV; e_mpEnergy consumption parameters of the multi-path attenuation model; e_elecEnergy consumption is transmitted for unit length; e_fsEnergy consumption parameters of the free space energy consumption model;

s12, initializing a membership degree matrix u_ij：

Giving a clustering number K and a total number N of nodes, considering the minimum objective function in the cluster of each clustering group, and adopting a fuzzy set to set a membership set of each node to each cluster as U, namely the membership of the jth node belonging to the ith cluster is U_ijThe center of the group is c_iAnd is and

the jth node, i.e., node x_j；

In order to make the distance between each node and the center of the group to which the node belongs effective when determining the node to which the node belongs and the distance between the node and the center of the group to which the node does not belong ineffective when determining the node to which the node belongs, the importance of controlling the membership degree by considering a fuzzy number q is considered, so that the determination x_jThe calculated values belonging to group i are:

when node x_jThe smaller the degree of membership to the i-th group, even if the distance is large, the fuzzy number q makes it possible to reduce the number of the groups

The distance is therefore equivalent to invalid when determining the position to which it belongs;

taking into account the initial degree of membership u_ijIntroducing a logistic chaotic model to carry out membership u on the influence of the final clustering effect of the FCM algorithm_ijThe mathematical model of the logistic chaotic model is as follows:

a_n+1＝μ*a_n*(1-a_n)

wherein, a_nValue generated for the nth iteration, a₁Is a random value in the range of (0-1); a is_n+1A value generated for the (n +1) th iteration; mu is an element of [0 to 4 ∈]；a_nThe value generated by iteration is in a chaotic state and has ergodicity;

s13, adopting Lagrange multiplicationSub-type parallel and parallel to u_ijAnd c_iAnd (3) calculating a partial derivative according to the following calculation formula:

wherein, c_lIs the cluster center of the first cluster;

s14, for all nodes and all clusters, the objective function is:

solving the condition of satisfying the constraint

Objective function of time J^(t)Minimum value of (1), J^(t)Representing the objective function value at the t iteration;

s15, if J^(t)-J^(t-1)I < epsilon, where epsilon is the minimum deviation value, J^(t-1)Representing the objective function value at the t-1 th iteration, the algorithm ends, cluster center c_iNamely the clustering center; otherwise, returning to the step S13;

in step S2, the design scheme of UAV trajectory without clustering by WSN specifically includes: all the traversal nodes are in direct communication with the UAV, the UAV sends information to the traversal nodes, and the traversal nodes send data to the UAV;

the design scheme of the UAV trajectory with WSN clustering and no cluster head fused with data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, and the cluster head nodes forward the data to the UAV;

the design scheme of the UAV trajectory of the WSN clustered and cluster head fused data specifically includes: the UAV only communicates with the cluster head, the UAV sends cluster member information to the cluster head, the cluster head broadcasts the cluster member information to the cluster members, then the cluster members send data to the cluster head, the cluster head fuses the data, and then the data are forwarded to the UAV;

in step S2, the number of traversal nodes and the number of clusters required for the area under 3 schemes are calculated from a given CRLB as follows:

(1) for WSNs not clustered:

a cluster-free adaptive quantization scheme is proposed:

calculating CRLB under the condition that WSN is not clustered through a cluster-free distributed estimation formula:

wherein N is the number of nodes;

indicating the normalized frequency, P (tau), of the sensor used as a threshold_n-1K Δ) is the quantization threshold τ of the n-1 th node_n-1Probability at k Δ; p is a radical of_w(x) Is w_kA Probability Density Function (PDF); f_w(x) Is w_kComplementary Cumulative Density Function (CCDF); θ is an estimated environmental physical quantity of the physical world, i.e. a parameter estimated from quantized data received from the UAV;

is an estimate of the parameter θ; w is a_kIs the observed noise of the kth node in the region; Δ represents a parametric quantization step; k represents the kth node in the area; tau is_kAccumulating early transmission data for a node from other sensor nodes and using the accumulated value as a threshold for its 1-bit quantifier;

then calculating the number of traversal nodes required by the region which is closest to and smaller than the given region CRLB by adopting a dichotomy;

(2) for WSN clustering but no fusion data: the calculation of locally estimated CRLB and the required traversal node number of the area is the same as the calculation of the WSN non-clustering;

(3) for WSN clustering and cluster head fusion data:

a clustering adaptive quantization scheme is proposed: the globally estimated CRLB may be represented by a weighted sum of the locally estimated CRLB; calculating CRLB under the condition that WSN is clustered and cluster heads fuse data through a clustering distributed estimation formula:

wherein,

a weighting coefficient representing a local estimate;

and

distributed estimation for the ith cluster and the jth cluster respectively;

then calculating the number of members in the cluster when the CRLB of a given area is met under the condition of different cluster numbers, wherein the maximum cluster number is the number of solar nodes in the area, and the minimum cluster number is 2;

comparing the total energy consumption of the WSN under different cluster numbers, and selecting the cluster number and the member number in the cluster under the condition of minimum energy consumption;

for node distribution uniformity:

calculating the distribution uniformity f of the nodes through a distribution uniformity potential function_rWherein, the distribution uniformity potential function calculation formula of the r-th node is as follows:

in the formula, n is the total number of nodes in the region; l represents the extension number of other nodes in the area; k represents the kth node in the area; s represents the inner segment of the regionThe s-th continuation node of the point; x is the number of_rI.e. represents the r-th node;

to a node x_rThe s-th continuation node of the k-th node in the area;

the uniformity potential function calculation formula of all nodes in the region is as follows:

when the distribution of the nodes in the region is more uniform, the total uniformity potential function value is smaller;

for node residual energy:

calculating the weight of the residual energy of the common node by the following calculation formula:

in the formula, E_startResidual energy is the node; e_RestIs the initial energy of the node;

calculating the weight of the residual energy of the solar node by the following calculation formula:

in the formula, E_ForeTo predict the energy harvesting value, the value E can be harvested from solar energy as follows_havEstimating;

when the residual energy of the node is more, the weight of the residual energy is smaller;

for solar energy collection values:

the solar energy collection can be quantized into a solar energy collection energy model represented by a trigonometric function of different peak values according to factors such as time, weather and position, a solar energy collection value is calculated through the model, and according to the rising and falling conditions of the sun, the solar energy collection is started at 6 am and finished at 6 pm;

the solar energy collection energy model is as follows:

E_hav＝z*sin(x)-rand

in the formula, E_havCollecting a value for solar energy; z is the solar energy collection peak value under different conditions of sunny days and cloudy days; x is a conversion value of a solar energy collecting time point,

time is the number of seconds from the current time to 6 am; rand is a random value simulating factors influencing energy collection;

the process of planning the UAV path according to the node distribution uniformity, the node residual energy, and the solar energy collection value is as follows:

s31, selecting a traversal node of the UAV based on a minimized cost equation, wherein the minimized cost equation is as follows:

in the formula, N_iThe number of the nodes selected in the ith cluster is obtained; w is a_ijThe residual energy weight of the jth node in the ith cluster is obtained; f. of_ijThe uniformity potential function value of the jth node in the ith cluster is obtained;

s32, considering the UAV path planning problem of the traversal node obtained by considering the minimum value of the cost function obtained by the simulated annealing algorithm, which may be regarded as a traveling quotient problem, that is, a shortest path planning problem that starts from an initial city, traverses all other cities, and finally returns to the initial city, and solving the traveling quotient problem means traversing all nodes except the initial node once in a connected graph, and planning a Hamilton loop except the shortest one, where a mathematical model defining the traveling quotient problem is as follows:

given a connectivity graph P ═ (C)_P,L_P) Wherein, C_PFor a set of node numbers, L_PIs a collection of edges between nodes;set node set C_P1,2_P＝{(r,j,w_rj)|r,j∈C_P,w_rj∈R⁺}，w_rjIs the distance between the r-th node and the j-th node, the Hamilton loop is a set of edges that contains all the nodes

Wherein the method comprises the steps of starting the node twice, remaining the node once, and m_r,m_j∈C_PAnd r ≠ j;

for edge set L_PThe selection of the element(s) is represented as follows:

the objective function of the Hamilton loop is then expressed as follows:

assuming that the previous state of the system is x (n), the state of the system is changed into x (n)) +1 according to the set index, the energy of the system is correspondingly changed from E (n) to E (n +1), and the receiving probability P of the system from x (n) to x (n)) +1 is defined as:

the system is different according to different scene requirements, and global area search and local cluster search are carried out in the system;

s33, solving the minimum value of the minimized cost equation in the step S31 by adopting a simulated annealing algorithm, wherein the smaller the cost function f of the node is, the greater the probability of selecting the node is, and thus the selected traversal node in the region is solved;

s34, calculating the minimum value of the objective function of the Hamilton loop according to the selected traversal nodes in the area to obtain the shortest path of the UAV;

the process of step S33 is as follows:

s331, in order to make the cooling process faster, firstly determining a better initial solution w through a Monte Carlo algorithm, and calculating an appropriate value f (w);

s332, considering that the selected nodes are discrete, selecting any one element in the random updating solution to generate a new solution w 'to calculate a new proper value f (w');

s333, then calculating an adaptation increment Δ f ═ f (w ') -f (w), and determining whether Δ f is greater than 0, if Δ f is less than or equal to 0, accepting a new solution according to Metropolis criterion, and if Δ f is greater than 0, accepting a new solution w ═ w ', f (w) ═ f (w ');

s334, slowly cooling, judging whether a termination condition is met, if not, returning to the step S332 again;

if so, finishing the operation, and obtaining the final solution which is the minimum value of the minimized cost equation;

in step S4, the node energy consumption models of the UAV trajectory planning route corresponding to the 3 schemes are as follows:

(1) based on non-cluster distributed estimation, the WSN is not clustered:

energy consumption of each node:

wherein,

energy consumption to receive UAV information for a node;

sending data to the UAV energy consumption for the node; k is a radical of₁Is the cluster information packet length; k is a radical of₂The length of the data packet is collected; r_cFor communication radius, to ensure full connectivity of the network, R_c＝2*R_S，R_SIs the sensing radius;

(2) based on non-cluster distributed estimation, the WSN forms clusters but the cluster heads do not fuse the energy consumption model of data:

let the number of cluster members of the ith cluster be m_iEnergy consumption of cluster head node

The method comprises the following steps:

(a) receiving UAV clustering information energy consumption

(b) Energy consumption for sending cluster information to cluster members

(c) Energy consumption for receiving cluster member data

(d) Energy consumption for sending data to UAV

Energy consumption of cluster members

The method comprises the following steps:

(e) energy consumption for receiving cluster head clustering information

(f) Energy consumption for receiving other node broadcast quantized data

(g) Energy consumption for sending data to cluster head

Finally, the energy consumption of the cluster head nodes and the energy consumption of the cluster members are respectively as follows:

(3) based on clustered distributed estimation, the WSN is clustered and the cluster heads fuse the energy consumption model of the data:

energy consumption of cluster head node

Energy consumption of member of Hezhou cluster

Respectively as follows:

wherein m is_i*E_daFusing data energy consumption for cluster heads;

the optimal scheme is selected through energy consumption and UAV flight distance, and specifically comprises the following steps:

calculating the total energy consumption E of the network based on the node energy consumption model_Total；

Then compare the total energy consumption E of the 3 schemes_TotalAnd selecting an optimal scheme according to the UAV flight distance Len corresponding to the UAV trajectory planning route, wherein the scheme selection function is as follows:

minf＝a₁E_Total+a₂Len

in the formula, a₁Is the energy consumption weight; a is₂Is the path length weightWeighing; wherein a is₁And a₂According to the actual situation, if the network life cycle proportion is larger, the setting a is₁The larger, and similarly, the more the UAV path length is constrained, the setting a₂The larger.

2. The UAV data acquisition trajectory algorithm based on the solar powered agricultural Internet of things as claimed in claim 1, wherein in step S1, deployment of solar nodes is performed in combination with cluster center mapping cluster head nodes, specifically:

selecting a node as a clustering center by adopting nearest neighbor mapping, and taking c_iDenotes the cluster center of the ith cluster, x_ijRepresenting the jth intra-cluster node of the ith cluster, and selecting the node as follows:

min||c_i-x_ij||

and after the position mapping is completed, selecting a mapping node to deploy for the solar node.

Images