A relay-based integrated sensing and control method in industrial Internet of Things

By introducing automatic guided vehicles and drones into the industrial Internet of Things, combining communication, perception and control, optimizing the relay transmission model and power distribution, the data transmission stability and energy consumption problems under channel conditions are solved, and high efficiency is achieved data backhaul and system energy efficiency improvement.

CN118678407BActive Publication Date: 2024-11-22DALIAN MARITIME UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411147209.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-11-22
Estimated Expiration
2044-08-21

AI Technical Summary

Technical Problem

In the industrial Internet of Things environment, the prior art is difficult to ensure the accuracy and stability of wireless transmission, especially when the mobility of drones and automatic guide vehicles is strong and the channel conditions change, how to effectively combine communication and perception to reduce interference And improving data backhaul performance is a challenge.

Method used

A method of integrated synestheses and control in industrial Internet of Things is proposed. By introducing automatic guided vehicles and drones into the system, an integrated communication, perception and control relay transmission model divided into two stages is established, and a semi-dual is used. The industrial relay transmission and scheduling method calculates a model that maximizes the total energy efficiency, and uses Dinkelbach algorithm, Taylor expansion and reconstruction rank-one solution methods to optimize communication and perceived power to realize the perceived-control activation strategy.

Benefits of technology

In a harsh industrial Internet of Things environment, increase the total amount of data back-pass communication and reduce system energy consumption, perform perception and motion control according to changes in channel conditions, improve transmission conditions, and improve system data transmission efficiency and energy efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118678407B_ABST
    Figure CN118678407B_ABST
Patent Text Reader

Abstract

A relay-based communication, sensing and control integration method in an industrial Internet of Things comprises the following steps: calculating the division points, communication power and sensing beams of two stages of an industrial Internet of Things system based on integrated communication, sensing and control relay transmission, and establishing a model for maximizing the total energy efficiency of the industrial Internet of Things system based on integrated communication, sensing and control relay transmission; dividing the model of the total energy efficiency of the industrial Internet of Things system into a first-stage system and a second-stage system according to two stages, the first-stage system is converted into a first approximate convex problem by using a Dinkelbach algorithm, a penalty term and a Taylor expansion method, and the second-stage system is converted into a second approximate convex problem by using a quadratic transformation and a method of reconstructing a rank-one solution; solving the first approximate convex problem by using a continuous convex approximation algorithm based on the golden section to obtain the optimal energy efficiency of the first-stage system, and solving the second approximate convex problem by using an interior point method of reconstructing a rank-one solution to obtain the optimal energy efficiency of the second-stage system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a communication, sensing and control integrated relay method in an industrial Internet of Things. Background Art

[0002] The widespread application of industrial Internet of Things technology in the field of automation has promoted the emergence of industrial network systems. Industrial network systems are a new generation of systems that exchange sensing and control information through communication networks. In industrial environments, a large number of IoT devices and wireless sensors are widely deployed to provide valuable data collection. However, due to the harsh communication environment and high mobility of equipment, it is extremely challenging to meet the high stability and reliability requirements of information transmission. Therefore, industrial network systems require not only high-quality wireless connections, but also high-precision sensing capabilities. Researching the fusion scheme of wireless transmission and sensing in industrial network systems is crucial to improving data feedback performance.

[0003] To address this trend, communication-sensing integration (synaesthesia integration) is proposed, which aims to allow perception and communication to share the same spectrum and infrastructure. Synaesthesia integration systems can further benefit from mutual assistance such as perception-assisted communication and communication-assisted perception. Strengthening the tight coupling between the two and reducing the interference between them is an urgent problem to be solved. Most schemes mainly focus on coupling sensing and communication signals using periodic schemes, where the sensing function mainly uses beamforming technology for target detection without considering sensing-assisted communication. In addition, most related works mainly focus on terrestrial networks, which themselves have various limitations such as signal blocking, incomplete coverage, and damaged infrastructure.

[0004] Next generation wireless networks have been seen as a key driver for industrial network systems. In general, communication systems often require sensing assistance, such as sensing signal estimation of channel state information, spectrum sensing in cognitive radio, and communication and sensing resource allocation. In recent years, integrating drones into fifth-generation and more advanced cellular networks has become a promising solution. In addition, in order to support high-speed data and large-scale connections on limited spectrum resources, it is meaningful to study the interference elimination between perception and communication. Drones are used as aerial nodes for relay transmission, especially using horn antennas. How to ensure beam alignment at different stages to ensure data transmission is a problem that needs to be considered, and sensing without adjustment cannot guarantee channel conditions.

[0005] In summary, the problems with existing technologies are: due to the complexity of the industrial wireless transmission environment, the accuracy and stability of data transmission cannot be guaranteed only through sensors; due to the mobility of drones and automated guided vehicles, it is challenging to obtain accurate channel conditions in a timely manner; how to balance the performance of communication and perception in the integrated communication and perception technology, how to couple the two and reduce the interference between them is still a topic that needs to be discussed for the industrial Internet of Things system, and how to operate after sensing the status information of the drone in different situations in the two stages of drone relay to ensure channel conditions still needs to be discussed. In the process of the automated guided vehicle transmitting data back through the drone relay, the combination of communication, perception and control systems also needs to be discussed. Summary of the invention

[0006] According to the technical problems raised above, the present invention provides an integrated relay method for communication, sensing and control in industrial Internet of Things, including the following steps:

[0007] Introduce automated guided vehicles and drones into the industrial Internet of Things system, and establish the channel model, communication model, perception model and control model of the industrial Internet of Things system with integrated relay transmission of communication, perception and control divided into two stages;

[0008] According to the actual situation of half-duplex relay transmission, the industrial Internet of Things system based on integrated relay transmission of communication, perception and control is divided into two stages, and the integrated scheduling method of communication, perception and control is adopted to control data transmission;

[0009] Calculate the division points, communication power and sensor beams of the two phases of the industrial Internet of Things system based on integrated communication, perception and control relay transmission, and establish a model to maximize the total energy efficiency of the industrial Internet of Things system based on integrated communication, perception and control relay transmission;

[0010] The total energy efficiency model of the industrial Internet of Things system is divided into two stages: the first stage system and the second stage system. The first stage system is converted into a first approximate convex problem using the Dinkelbach algorithm, penalty term and Taylor expansion method, and the second stage system is converted into a second approximate convex problem using the quadratic transformation and reconstruction of the rank one solution method.

[0011] The first approximate convex problem is solved by a continuous convex approximation algorithm based on the golden section, and the optimal energy efficiency of the first-stage system is obtained. The second approximate convex problem is solved by an interior point method of reconstructing a rank-one solution, and the optimal energy efficiency of the second-stage system is obtained.

[0012] Furthermore, the two stages of the industrial Internet of Things system with integrated relay transmission of communication, perception and control include:

[0013] Phase 1: Data transmission between the AGV and the UAV; Phase 2: The UAV transmits data to the wireless access point;

[0014] The channel model of the industrial Internet of Things system with integrated communication, perception and control relay transmission mainly includes multipath fading, path loss and misalignment fading. The channel gains of the two stages are expressed as follows:

[0015] h 1 =h p h f h m

[0016] h 2 =h p h f h a

[0017] Where: path loss h p Described by the free space propagation model, the expression is:

[0018]

[0019] Among them, G t and G r are the transmit and receive signal gains, respectively, λ = f / c is the signal wavelength, where c is the speed of light, f is the occupied frequency band, and d is the distance between the transceivers;

[0020] Where: multipath fading h f It is usually described by Rayleigh distribution, whose probability distribution function for:

[0021]

[0022] in: is a fading channel with an RMS value of α = 2, and the probability distribution function of x is The independent variable;

[0023] The first stage misalignment fading h m :Assume that the transmitting beam of the UAV covers an area S with a radius of R, and the radius of the beam coverage area of ​​the automatic guided vehicle is r, and the distance is d, where 0≤R≤R m , R m is the maximum radius of the beam at distance d, and l is the beam center O of the automatic guided vehicle AGV With the drone’s beam center O UAV The misalignment error between m It is expressed as:

[0024]

[0025] The equivalent beam width is expressed as R e Indicates that θ is the angular error of the horn antenna. The misalignment error in the first stage only considers the tangent plane direction of the optimal direction, that is, in is the misalignment error coefficient. In addition, when l = 0, the received power of the drone is expressed as P 0 , obtained by the following formula:

[0026]

[0027] in, erf(·) is the Gaussian error function, then the equivalent beam width R e It is expressed as:

[0028]

[0029] The second stage misalignment fading h a It is expressed as:

[0030]

[0031] in, is the projected area in the best case, and similarly, is the projection area of ​​the drone beam on the wireless access point beam, where and θ 0 =2arctan(r / d) is the UAV beam radius and the UAV beam angle; in addition, the misalignment error L is expressed as:

[0032]

[0033] S14, in the perception channel of integrated relay transmission for communication, perception and control: the wireless access point sends a perception signal through a uniform linear array to perceive the position of the automatic guided vehicle and the UAV. The perception channel is expressed as:

[0034]

[0035] Where β is the channel gain value at 1m away from the wireless access point, m∈M,Z a are the x- and y-axis coordinates of the wireless access point. The position information of the AGV and the UAV are expressed in coordinates as follows: 1 (Z 1 (g x1 ,g y1 ),g z1 ) and G 2 (Z 2 (g x2 ,g y2 ),gz2 ),

[0036] In the first-stage communication model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the signal y received by the drone u It is expressed as:

[0037] y u =h 1 s a +h s x+ω

[0038] Among them, s a represents the communication signal sent by the automatic guided vehicle, x represents the perception signal sent by the wireless access point, and h 1 represents the flat fading wireless channel between the AGV and the UAV, h s represents the wireless channel between the wireless access point and the drone, ω represents the mean value is zero and the variance is N 0 Additive noise;

[0039] UAV receiving achievable rate C R,k It is expressed as:

[0040]

[0041] Among them, P c,k Represents the transmission power of the communication signal, P s,k Indicates the power of the UAV receiving the sensing signal, N 0 is the additive noise power;

[0042] In the first stage, the k-th time code rate R transmitted by the AGV to the UAV R,k It is expressed as:

[0043] R R,k =C R,k

[0044] Assuming that the drone successfully decodes the codeword and stores the corresponding information in its buffer, use Q k represents the amount of data in the relay buffer at the end of the kth time slot, so Q n,k Will change as follows:

[0045]

[0046] In the first-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the key indicator of the perception function is the covariance matrix of the perception beam distribution determined by the transmitted signal. The covariance matrix R ω,k It is expressed as:

[0047]

[0048] Among them, w m,k is the transmit beamforming vector of the sensing signal, and the wireless access point transmits the sensing signal power P t,k For: P t,k =Tr(R ω,k )

[0049] In Directional sensing beam pattern for:

[0050]

[0051] in, represents the steering vector when the uniform linear array is used as an antenna at a wireless access point, λ represents the wavelength, and d 0 Indicates the antenna spacing, N 0 is the number of linear array antennas of the wireless access point;

[0052] Therefore, the wireless access point senses the power P of the drone in the kth time slot s,k It can be expressed as:

[0053]

[0054] Using the position information of the AGV and the UAV, determine the horizontal angle φ of the sensing target relative to the wireless access point m,k for:

[0055]

[0056] In the second-stage communication model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the signal y received by the wireless access point a It is expressed as:

[0057] y a =h 2 s u +δ k h s x e +ω,

[0058] Where ω is the power of the additive noise, s u The signal received by the UAV after decoding and forwarding the communication signal sent by the automatic guided vehicle, x e is the sensing echo signal reflected back to the wireless access point, δ k is the perceptual activation coefficient; channel capacity C T,k It is derived as follows:

[0059]

[0060] Where: P u,k is the communication transmission power of the UAV;

[0061] The kth time slot code rate R of the wireless access point T,k It is expressed as:

[0062] R T,k =min{Q k-1 ,C T,k},

[0063] When R T,k ≤C T,k When , the wireless access point can successfully decode the codeword, and as the drone relays and transmits it, the number of codewords in the buffer area becomes:

[0064]

[0065] In the second-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the covariance matrix R v,k It is expressed as:

[0066]

[0067] Among them, v k is the transmit beamforming vector of the sensing signal; then, the wireless access point transmits the sensing signal power P in the second stage v,k for:

[0068] P v,k =Tr(R v,k )

[0069] In Directional sensing beam pattern for:

[0070]

[0071] In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone motion control model is as follows:

[0072] The discrete time control model of UAV motion control is:

[0073] x k+1 =A d x k +B d u k +w k ,

[0074] Among them, x k is the state of the drone at time k, A d and B dis the system parameter matrix in the discrete-time model, u k is the motion control input, w k is the disturbance caused by additive Gaussian white noise with zero mean and variance W. Assuming that at each time slot k, the sampling state of the wireless access point is perfect, that is, y k =x k , the motion control input of the drone is calculated as:

[0075] u k =Ky k ,

[0076] Where K is the control gain, for a speed of V k The propulsion power consumption of the rotor UAV is P f,k (V k ) is modeled as:

[0077]

[0078] Among them, U tip is the rotor tip speed, v o is the average rotor induced speed, d o is the fuselage drag ratio, P b and P i are blade profile power and induced power, ρ, s and w respectively o are respectively the air density, rotor solidity and blade angular velocity. The speed Z of the UAV in the discrete domain k Calculated by the following formula:

[0079]

[0080] Among them, Δ 0 Indicates a time interval;

[0081] In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone antenna angle control model is as follows: the distance and relative angle between the drone and the automatic guided vehicle can be obtained through perception measurement, the direction of the horn antenna on the drone can be expressed by the pitch angle θ and the azimuth angle φ, and the angle of the horn antenna pointing to the device direction is calculated as θ d ,in:

[0082]

[0083] Then, the error in the horn antenna pointing direction at the kth moment is e θ,k =θ k -θ d,k ,θ k is the antenna pointing angle of the drone at the kth moment, obtained through the drone's built-in sensor, and the controller is used to control the direction of the horn antenna and the output zk It is expressed as:

[0084] z k =K a θ k ,

[0085] Among them, K a To control the gain, the discrete time control model of the UAV horn antenna angle control is:

[0086] θ k+1 =A a θ k +B a z k +w a,k ,

[0087] Among them, A a and B a are the system parameter matrices, w a,k is the disturbance at time k, θ k+1 represents the antenna angle of the UAV at time slot k+1.

[0088] Furthermore, according to the actual situation of half-duplex relay transmission, the industrial Internet of Things system architecture based on integrated relay transmission of communication, perception and control is divided into two stages, and the communication, perception and control integrated scheduling method is adopted to perform data transmission control as follows:

[0089] S21. The first stage of the perception-control cycle triggering method: the automatic guided vehicle sends the collected sensor information to the UAV through the horn antenna. At the same time, the wireless access point sends a perception signal to obtain the position information of the automatic guided vehicle and the UAV. Then, the controller in the wireless access point calculates the misalignment error, which is the key indicator for activating the UAV control system. The controller in the wireless access point makes a judgment based on the misalignment error l related to the UAV beam radius R. When l>R, the wireless access point will issue UAV motion control and antenna angle control instructions, otherwise it will only issue antenna angle control instructions, where the control activation coefficient ε k for:

[0090]

[0091] S22, the second stage of the perception-control event triggering method: when the UAV is transmitting data with the wireless access point, the communication function of the wireless access point is continuously activated to receive the data from the UAV. Only when the triggering condition is met, the perception function will perform a perception, and the controller will maintain the local estimation state. It is expressed as:

[0092]

[0093] When the estimated value of the above formula cannot maintain the control performance and guarantee the communication requirements, the actual state of the UAV is sensed by the wireless access point, and the estimated value of the state at time slot k is:

[0094]

[0095] in: is the state estimate, x k It is the actual state;

[0096] Assuming the drone is hovering on a horizontal plane, determine the error e between the actual state and the estimated state of the drone on the controller k+1 for:

[0097]

[0098] State-to-Noise Ratio From this, the perception-control activation threshold p is derived tr Relationship with state-to-noise ratio;

[0099] The perception-control activation threshold p is obtained tr The relationship between the state noise ratio threshold, that is, Therefore, the rate threshold of perception-control activation is obtained

[0100]

[0101] The second stage of perception mode activation conditions are as follows:

[0102]

[0103] The control activation mode is determined according to the UAV beam projection area. The activation conditions of the second stage control mode are as follows:

[0104]

[0105] Furthermore, the expression of the non-convex problem of maximizing the ratio of the total communication volume to the total power consumption of the industrial Internet of Things system with integrated communication, perception and control relay transmission is as follows:

[0106] S31, joint optimization phase division point, communication power and perception beam vector, to obtain the ratio of the maximum communication volume to the total power consumption P 0 :

[0107]

[0108] Where: C1 and C2 represent the AP sensing accuracy constraints corresponding to the first and second phases, where ψ kis the first stage coefficient, which is 1 when in the first stage and 0 otherwise; P d is the beam pattern compared with the basic beamforming; τ is the minimum mean square error value compared with the basic beamforming;

[0109] C3 and C4 represent the maximum transmit power constraints for beamforming, where P l represents the perceived power upper limit; C5 is the first stage serial interference elimination condition constraint;

[0110] C6 is the communication power constraint of the second-stage UAV;

[0111] C7 is the cache constraint of the drone;

[0112] C8 represents the integer constraint of the time slot division point;

[0113] C9 represents the 0-1 constraint of the perception and control activation factor;

[0114] Constraints C2-C6 are non-convex due to the quadratic form of the covariance matrix.

[0115] Furthermore, the process of converting the first-stage system into a first approximate convex problem using the Dinkelbach algorithm, penalty term and Taylor expansion method is as follows:

[0116] S41, the beamforming and power allocation sub-problem of the first stage is expressed as:

[0117]

[0118] stC1,C3,C5,C7,C9

[0119] The time slot partitioning point, beamforming and power allocation sub-problems are expressed as:

[0120]

[0121] stC2,C4,C6-C9

[0122] S42, P 1 The objective function is a fractional non-convex function, P 1 The objective function is transformed into a concave function maximization problem through the Dinkelbach method, that is:

[0123]

[0124] stC1,C3,C5,C7,C9

[0125] Among them, η k =R R,k / P c,k +P t,k +εk P f,k , and if found Make If it holds, then the optimization problem P 1 The solution is obtained by solving the equivalent problem P 1.1 and get;

[0126] S43. Next, define auxiliary variables in, , rank(W m,k )=1, then P 1.1 Rewritten as:

[0127]

[0128] stC1,C3,C5,C7,C9,

[0129]

[0130] C11:rank(W m,k )=1

[0131] Use eigenvalue decomposition or Gaussian randomization to reconstruct the general solution obtained by the semidefinite relaxation method into a rank-one solution, transform the rank-one constraint into a penalty term in the objective function, and then use a continuous convex approximation algorithm to solve the problem. The rank-one constraint is equivalent to:

[0132]

[0133] Among them, ||·|| * is the nuclear norm, i.e., the sum of the singular values ​​of the matrix, ||·|| 2 is the spectral norm, i.e. the largest singular value of the matrix. Therefore, if the matrix W m,k is a rank-one matrix, then the above equation holds. Otherwise, since W m,k is a semi-positive definite matrix, the sum of the singular values ​​is always greater than the largest singular value, that is, ||W m,k || * -||W m,k || 2 > 0, in order to obtain a rank-one matrix, a penalty term is introduced in the objective function, and the result is:

[0134]

[0135] stC1, C3, C5, C7, C9,

[0136]

[0137] Among them: ξ is the penalty factor, only when it tends to 0, that is When W approaches ∞,m,k The rank of is close to 1;

[0138] S44. In this case, P 1.2 The main non-convexity comes from the second term of the penalty term, which is used at the point The first-order Taylor expansion at is used instead:

[0139]

[0140] in: is with The eigenvector corresponding to the largest eigenvalue, then, P 1.2 The problem is similar to the following one:

[0141]

[0142] stC1,C3,C5,C7,C9,C10

[0143] S45, introduce attenuation coefficient μ 0 , by initializing a large value ξ, and then by ξ = μ 0 ξ gradually reduces it to a sufficiently small value to obtain the overall suboptimal solution, where 0<μ 0 <1, when the penalty term is small enough, that is, ||W m,k || * -||W m,k || 2 <ε 2 , jump out of the iteration.

[0144] Furthermore, the process of solving the first approximate convex problem by using the continuous convex approximation algorithm based on the golden section to obtain the optimal energy efficiency of the first-stage system is as follows:

[0145] 1) Initialization: feasible golden search interval (η l ,η h ), iterative threshold κ, golden section coefficient And the golden section search upper and lower bound update formula υ l =η h -τ(η h -η l ), h =η l +τ(η h -η l ), a feasible Taylor expansion point Reduction factor μ 0 , iterative threshold ε 1 and ε 2 , n = 1;

[0146] 2) When |η h-η l When |>κ, iterate;

[0147] 3) Input η according to the continuous convex approximation method l and η h Perform external iterations separately;

[0148]

[0149] 8) If f(η l )≤f(η h ), f(η h ) is assigned to f(η l ),υ l The value of η is assigned to l , h The value of is assigned to υ l , η l +τ(η h -η l ) is assigned to υ h ;

[0150] 9) When |η h -η l When |≤κ, the iteration ends;

[0151] 10) Set (η h +η l ) / 2 is assigned to According to the penalty-based double-layer continuous convex approximation method, the input Solving for P 1.4 , get the optimal communication power of the automatic guided vehicle Optimal sensing beamforming vector with wireless access point

[0152] Furthermore, the process of converting the non-convex problem into the second approximate convex problem by using the quadratic transformation and reconstruction rank-one solution method in the second stage system is as follows:

[0153] S51, P 2 The problem is rephrased as follows:

[0154]

[0155] stC2,C4,C6-C9

[0156] S52, P 2.1 The optimization objective function has a multi-ratio fractional form. First, a quadratic transformation is applied to decouple the objective function into:

[0157]

[0158] in, Then, the original problem P 2.1 The equivalent conversion is:

[0159]

[0160] stC2,C4,C6-C9

[0161] And, if and only if When the optimization problem P 2.2 Equivalent to the optimization problem P 2.1 ,in is x k The maximum value of

[0162] S53, convert the optimization problem of each time slot into two forms:

[0163] S531: When the perception-control system is not activated and the wireless access point does not need to perform perception control operations on the drone:

[0164]

[0165] stC6-C9

[0166] Question P 2.3.1 is a concave function that can be solved efficiently using the CVX toolbox and the MOSEK solver;

[0167] S532: When the perception-control system is activated and the wireless access point needs to perform perception control operations on the drone:

[0168]

[0169] stC2,C4,C6-C9

[0170] Question P 2.3.2 is a non-convex function. The main non-convexity is R in the objective function. T ,k and v k Next, define the auxiliary variables in, rank(V k )=1, then P 2.3.2 Rewritten as:

[0171]

[0172] stC2,C4,C6-C9,

[0173]

[0174] C13:rank(V k )=1

[0175] P 2.4 The main non-convex constraint comes from the rank-one constraint of C13. For the problem P 2.4 , there is always an optimal solution of rank one:

[0176]

[0177] Therefore, the newly constructed solution Satisfy all the constraints of the original problem and ensure Finally, P 2.4 Rephrased as:

[0178]

[0179] stC2,C4,C6-C9,C12

[0180] Question P 2.5 It is a quadratic semidefinite programming problem, which is solved efficiently using the CVX toolbox and the MOSEK solver.

[0181] Furthermore, the process of solving the second approximate convex problem by the interior point method of reconstructing the rank-one solution and obtaining the optimal energy efficiency of the second-stage system is as follows:

[0182] 1) Initialization: feasible auxiliary variable χ k 、UAV communication power P c,k , the beamforming vector V of the wireless access point k ;

[0183] 2) When χk does not converge, iterate;

[0184] 3) According to Find the best χ k ;

[0185] 4) If δ′ k = 0, by solving P 2.3.1 Get the communication power P of the drone c,k ;

[0186] 5) Otherwise, by solving P 2.5 Get the communication power P of the drone c,k With the beamforming vector V of the wireless access point k ;

[0187] 6) When the iteration ends, if δ′ k =0, based on the iteration Solve for the value of P 2.3.1 Get the optimal communication power of the drone

[0188] 7) Otherwise, based on the iteration Solve for the value of P 2.5 Get the optimal communication power of the drone Optimal beamforming vector with wireless access point

[0189] Compared with the prior art, the present invention has the following advantages:

[0190] The present invention provides an integrated relay method for communication, sensing and control in an industrial Internet of Things, which can ensure data feedback and reduce system energy consumption in harsh industrial Internet of Things environments, and adopt different schemes to couple the sensing-control scheme in stages according to the relay mode for data transmission. It can sense the drone when the communication rate decreases due to changes in channel conditions to achieve motion control or antenna angle control to improve transmission conditions, effectively increasing the total amount of communication for data feedback in the industrial Internet of Things system while reducing the total power consumption of the system.

[0191] Based on the above reasons, the present invention can be widely promoted in fields such as wireless communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0192] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0193] Figure 1 is a flow chart of the method of the present invention;

[0194] Figure 2 A scene graph used in an embodiment of the present invention;

[0195] Figure 3 A time slot division diagram used in an embodiment of the present invention;

[0196] Figure 4 A diagram of the misalignment situation division in the first stage provided by an embodiment of the present invention, in the best case;

[0197] Figure 5 A diagram for dividing misalignment conditions in the first stage provided by an embodiment of the present invention, when there is a small misalignment condition;

[0198] Figure 6 When the misalignment situation division diagram of the first stage provided by the embodiment of the present invention is greatly misaligned;

[0199] Figure 7 A diagram for dividing misalignment conditions in the second stage provided by an embodiment of the present invention;

[0200] Figure 8A flow chart of the first stage perception-control activation scheme provided by an embodiment of the present invention;

[0201] Fig. 9 A flow chart of the second stage perception-control activation scheme provided by an embodiment of the present invention;

[0202] Fig.10 A comparison diagram of the objective function value versus the number of external iterations for the continuous convex approximation algorithm based on the golden section provided by an embodiment of the present invention under different search interval upper limits;

[0203] Fig.11 A comparison diagram of the objective function value versus the number of inner iterations under different reduction coefficients for the continuous convex approximation algorithm based on the golden section provided in an embodiment of the present invention;

[0204] Fig.12 A comparison diagram of the change of penalty item value with the number of inner iterations under different reduction coefficients for the continuous convex approximation algorithm based on the golden section provided in an embodiment of the present invention;

[0205] Fig.13 A comparison diagram of the objective function value changing with the number of iterations under different search initial values ​​for the reconstruction rank-one method based on quadratic changes provided in an embodiment of the present invention;

[0206] Fig.14 A comparison diagram of the total amount of communication within a single synaesthesia integration frame as a function of the stage division point under different solutions provided by the embodiments of the present invention;

[0207] Fig.15 A comparison diagram of the total amount of communication within a single interawareness integrated frame as a function of the upper limit of the UAV communication power under different solutions provided in the embodiments of the present invention;

[0208] Fig.16 A comparison diagram of the total energy efficiency within a single synaesthesia integrated frame as a function of the upper limit of the UAV communication power under different solutions provided in the embodiments of the present invention;

[0209] Fig.17 A comparison diagram of the total amount of communication within a single synaesthesia integrated frame as the flight altitude of the drone changes under different solutions provided in the embodiments of the present invention;

[0210] Fig.18 A comparison diagram of the total energy efficiency of a single synaesthesia integrated frame as it changes with the flight altitude of the drone under different solutions provided in the embodiments of the present invention;

[0211] Fig.19 A comparison diagram of beam patterns solved by a continuous convex approximation algorithm based on the golden section at different angles provided by an embodiment of the present invention;

[0212] Fig. 20A comparison diagram of beam patterns solved by the reconstruction rank-one method based on quadratic changes at different angles provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0213] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0214] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0215] like Figure 1 As shown, an embodiment of the present invention provides a method for integrating communication, sensing and control in an industrial Internet of Things, including the following steps:

[0216] An integrated relay method for communication, sensing and control in industrial Internet of Things, comprising the following steps:

[0217] An integrated relay method for communication, sensing and control in industrial Internet of Things, characterized by comprising the following steps:

[0218] S1: Introduce automated guided vehicles and drones into the industrial Internet of Things system, and establish the channel model, communication model, perception model and control model of the industrial Internet of Things system with integrated relay transmission of communication, perception and control divided into two stages;

[0219] S2: According to the actual situation of half-duplex relay transmission, the industrial Internet of Things system based on integrated communication, perception and control relay transmission is divided into two stages, and the communication, perception and control integrated scheduling method is adopted to control data transmission;

[0220] S3: Calculate the division points, communication power and sensor beams of the two phases of the industrial Internet of Things system based on integrated communication, perception and control relay transmission, and establish a model to maximize the total energy efficiency of the industrial Internet of Things system based on integrated communication, perception and control relay transmission;

[0221] S4: The total energy efficiency model of the industrial Internet of Things system is divided into a first-stage system and a second-stage system according to two stages. The first-stage system is converted into a first approximate convex problem using the Dinkelbach algorithm based on the golden section, penalty terms and Taylor expansion method, and the second-stage system is converted into a second approximate convex problem using a quadratic transformation and a rank-one reconstruction method.

[0222] S5: The continuous convex approximation algorithm is used to solve the first approximate convex problem to obtain the optimal energy efficiency of the first-stage system. The interior point method of reconstructing the rank-one solution is used to solve the second approximate convex problem to obtain the optimal energy efficiency of the second-stage system.

[0223] S1 / S2 / S3S4 / S5 are executed sequentially;

[0224] Figure 2 A scene graph used in an embodiment of the present invention;

[0225] Figure 3 The time slot division diagram used in the embodiment of the present invention; further, the automatic guided vehicle and the unmanned aerial vehicle are introduced into the industrial Internet of Things system, and the industrial Internet of Things system architecture of the integrated relay transmission of communication, perception and control divided into two stages is established as follows:

[0226] First, an AGV is used as a mobile data collector to approach IoT devices and collect data, thereby minimizing the transmission energy consumption of IoT devices. Next, the AGV receives information from sensors along a predetermined path, and deploys drones to collaborate with the AGV, relaying information using a half-duplex decoding and forwarding working mode. The wireless access point with perception and control functions is responsible for receiving the signals forwarded by the drone, and perceives and controls the drone and AGV according to the set mode. However, since the drone may be interfered by the sensor signals of the wireless access point, non-orthogonal multiple access technology is introduced to alleviate this problem.

[0227] The interawareness integrated frame structure of the industrial Internet of Things system with integrated communication, perception and control relay transmission mainly includes two stages:

[0228] S121. Phase 1: Data transmission between the AGV and the UAV. Considering their maneuverability and the use of horn antennas, accurate beam alignment becomes very important. The wireless access point periodically senses the position of the AGV and the UAV and performs control operations to ensure optimal beam alignment. This ensures that the communication rate requirements are met.

[0229] S122, Phase 2: The drone transmits data to the wireless access point. Since the wireless access point receiving antenna is an omnidirectional antenna, the drone beam can be aligned in a general direction to ensure the communication rate between the two. The perception-control function of the wireless access point works in an event-triggered manner. Only when the rate is lower than the set threshold will it sense and obtain the drone's location information, and issue control commands through the controller to ensure communication requirements.

[0230] The channel model of the industrial Internet of Things system with integrated communication, perception and control relay transmission mainly includes multipath fading, path loss and misalignment fading. The channel gains of the two stages are expressed as follows:

[0231] h 1 =h p h f h m

[0232] h 2 =h p h f h a

[0233] S131, path loss h p : The path loss is described by the free space propagation model, and its expression is:

[0234]

[0235] Among them, G t and G r are the transmit and receive signal gains, respectively, λ = f / c is the signal wavelength, where c is the speed of light, f is the occupied frequency band, and d is the distance between the transceivers;

[0236] S132, multipath fading h f :Multipath fading is usually described by Rayleigh distribution, and its probability distribution function is for:

[0237]

[0238] in, is a fading channel with an RMS value of α = 2 m , x is the independent variable of the function;

[0239]

[0240] S133, first stage misalignment fading h m :Assume that the transmitting beam of the UAV covers an area S with a radius of R, and the radius of the beam coverage area of ​​the automatic guided vehicle is r, and the distance is d, where 0≤R≤R m, R m is the maximum radius of the beam at distance d, and l is the beam center O of the automatic guided vehicle AGV With the drone’s beam center O UAV According to the relationship between l and R, the misalignment can be divided into three types: the best case, small misalignment and large misalignment.

[0241] Figure 4 A diagram of the misalignment situation division in the first stage provided by an embodiment of the present invention, in the best case;

[0242] Figure 5 The first stage misalignment situation division diagram provided by the embodiment of the present invention, when the misalignment situation is small;

[0243] Figure 6 When the misalignment situation division diagram of the first stage provided by the embodiment of the present invention is greatly misaligned;

[0244] Misalignment fading is expressed as:

[0245]

[0246] The equivalent beam width is expressed as R e Indicates that θ is the angular error of the horn antenna. The misalignment error in the first stage only considers the tangent plane direction of the optimal direction, that is, in is the misalignment error coefficient. In addition, when l = 0, the received power of the UAV can be expressed as P 0 , obtained by the following formula:

[0247]

[0248] in, erf(·) is the Gaussian error function, then the equivalent beam width R e It is expressed as:

[0249]

[0250] S134, second stage misalignment fading h a : Compared with the first stage, the wireless access point uses an omnidirectional antenna, and the misalignment error is affected by the alignment of the drone horn antenna. In addition, the projection area of ​​the drone horn antenna beam on the wireless access point omnidirectional antenna beam is also a factor affecting the alignment. The misalignment fading in the second stage is expressed as:

[0251]

[0252] in, is the projected area in the best case. Similarly, is the projection area of ​​the drone beam on the wireless access point beam, where and θ 0 =2arctan(r / d) is the drone beam radius and drone beam angle. In addition, the misalignment error L can be expressed as:

[0253]

[0254] The perception channel of the industrial Internet of Things system with integrated communication, perception and control relay transmission: the wireless access point sends a perception signal through a uniform linear array to perceive the position of the automatic guided vehicle and the drone. The perception channel is expressed as:

[0255]

[0256] Where β is the channel gain value at 1m away from the wireless access point, m∈M,Z a are the x- and y-axis coordinates of the wireless access point. The position information of the AGV and the UAV are expressed in coordinates as follows: 1 (Z 1 (g x1 ,g y1 ),g z1 ) and G 2 (Z 2 (g x2 ,g y2 ),g z2 ).

[0257] S151, in the first-stage communication model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the signal y received by the drone u It is expressed as:

[0258] y u =h 1 s a +h s x+ω

[0259] Among them, s a represents the communication signal sent by the automatic guided vehicle, x represents the perception signal sent by the wireless access point, and h 1 represents the flat fading wireless channel between the AGV and the UAV, h s represents the wireless channel between the wireless access point and the drone, ω represents the mean value is zero and the variance is N 0 Additive noise;

[0260] In wireless transmission based on non-orthogonal multiple access technology, serial interference cancellation technology is generally used to eliminate the signal of weak users from the received signal of strong users to improve the signal-to-noise ratio.

[0261]

[0262] The receiving power varies, and the UAV can receive the achievable rate C R,k It is expressed as:

[0263] Among them, P c,k Represents the transmission power of the communication signal, P s,k Indicates the power of the UAV receiving the perception signal. When the serial interference elimination condition is met, the interference of the perception signal on the communication signal is eliminated; N 0 is the additive noise power, however, if the serial interference cancellation condition is not met, the perceived signal will still affect the data transmission rate as interference.

[0264] Since the UAV uses a single antenna, we use a half-duplex decoding and forwarding relay mode. In the first stage, the codeword rate at the kth moment transmitted by the automatic guided vehicle to the UAV can be expressed as:

[0265] R R,k =C R,k .

[0266] Assume that the drone successfully decodes the codeword and stores the corresponding information in its buffer. n,k represents the amount of data in the relay buffer at the end of time slot k in the kth synaesthesia integration frame. n,k Will change as follows:

[0267]

[0268] S152. In the first-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the key indicator of the perception function is the covariance matrix of the perception beam distribution determined by the transmitted signal, and the covariance matrix is ​​expressed as:

[0269]

[0270] Among them, w m,k is the transmit beamforming vector of the sensing signal. Then, the wireless access point transmits the sensing signal power as:

[0271] P t,k =Tr(R ω,k )

[0272] In The perception beam diagram in the direction is:

[0273]

[0274] in, represents the steering vector when a uniform linear array is used as an antenna at a wireless access point. represents the wavelength, d0 represents the antenna spacing, and N0 is the number of linear array antennas at the wireless access point; therefore, the power of the wireless access point sensing the UAV in the kth time slot can be expressed as:

[0275]

[0276] Using the position information of the AGV and the UAV, the horizontal angle of the sensing target relative to the wireless access point can be determined as:

[0277]

[0278] S161. In the second-stage communication model of the industrial Internet of Things system with integrated relay transmission for communication, perception and control, the signal received by the wireless access point is expressed as:

[0279] y a =h 2 s u +δ k h s x e +ω,

[0280] Where ω is the power of the additive noise, s u The signal received by the UAV after decoding and forwarding the communication signal sent by the automatic guided vehicle, x e is the sensing echo signal reflected back to the wireless access point, δ k is the sensing activation coefficient. However, in real scenarios, the echo interference from the sensing signal of the wireless access point is very small and can be treated as noise. Its interference is often ignored when calculating the rate, which doubles the noise power. Therefore, the channel capacity can be derived as:

[0281]

[0282] Where: P u,k is the communication transmission power of the UAV;

[0283] The k-th time slot code rate of the wireless access point can be expressed as:

[0284] R T,k =min{Q n,k-1 ,C T,k},

[0285] The minimum value in the above formula is due to the relay buffer limitation, that is, it cannot transmit more information than what is stored in the buffer. T,k ≤C T,kWhen , the wireless access point can successfully decode the codeword. As the drone relays and transmits, the number of codewords in the buffer becomes:

[0286]

[0287] S162. In the second-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the covariance matrix is ​​expressed as:

[0288]

[0289] Among them, v m,k is the transmit beamforming vector of the sensing signal. Then, the power of the sensing signal transmitted by the wireless access point in the second phase is:

[0290] P v,k =Tr(R v,k )

[0291] In The perception beam diagram in the direction is:

[0292]

[0293] S171. In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone motion control model is as follows:

[0294] The discrete time control model of UAV motion control is:

[0295] x k+1 =A d x k +B d u k +w k ,

[0296] Among them, x k is the state of the drone at time k, A d and B d is the system parameter matrix in the discrete-time model, u k is the motion control input, w k is the disturbance caused by additive Gaussian white noise with zero mean and variance W. Assuming that at each time slot k, the sampling state of the wireless access point is perfect, that is, y k =x k , the controller calculates the control input of the drone as:

[0297] u k =Ky k ,

[0298] Where K is the control gain, for a speed of Vk The propulsion power consumption of the rotorcraft is modeled as:

[0299]

[0300] Among them, U tip is the rotor tip speed, v o is the average rotor induced speed, d o is the fuselage drag ratio, P b and P i are blade profile power and induced power, ρ, s and w respectively o are respectively the air density, rotor solidity and blade angular velocity. The speed V of the UAV in the discrete domain k Calculated by the following formula:

[0301]

[0302] Where: Δ 0 Indicates a time interval;

[0303] S171. In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone antenna angle control model is as follows: the distance and relative angle between the drone and the automatic guided vehicle can be obtained through perception measurement, and the direction of the horn antenna on the drone can be represented by the pitch angle θ and the azimuth angle φ. Since the drone and the automatic guided vehicle maintain relative motion, the antenna control model is simplified, and it is considered that the azimuth angle between them remains optimally constant. According to the knowledge of trigonometric functions, the angle of the horn antenna pointing to the device direction is calculated as θ d ,in:

[0304]

[0305] Then, the error in the horn antenna pointing direction at the kth moment is e θ,k =θ k -θ d,k ,θ k is the antenna pointing angle of the drone at the kth moment, which is obtained through the built-in sensor of the drone. The controller can be used to control the direction of the horn antenna. The control output is expressed as:

[0306] z k =K a θ k ,

[0307] Among them, K a To control the gain, the discrete time control model of the UAV horn antenna angle control is:

[0308] θ k+1 =A a θ k +Ba z k +w a,k ,

[0309] Among them, A a and B a are the system parameter matrices, w a,k is the disturbance at time k, θ k+1 represents the antenna angle of the UAV in time slot k+1;

[0310] Since the power consumption of UAV motion control is relatively high, UAV antenna angle control is introduced, and then an integrated scheduling method of communication, perception and control is proposed. The appropriate control method of the UAV is selected based on the error conditions to obtain a higher data transmission rate for the industrial Internet of Things system.

[0311] Figure 7 A diagram for dividing misalignment conditions in the second stage provided by an embodiment of the present invention;

[0312] Figure 8 A flow chart of the first stage perception-control activation scheme provided by an embodiment of the present invention;

[0313] Fig. 9 The flowchart of the second-stage perception-control activation scheme provided by the embodiment of the present invention; further, according to the actual situation of half-duplex relay transmission, the industrial Internet of Things system architecture based on integrated relay transmission of communication, perception and control is divided into two stages, and the process of data transmission control is performed by adopting the integrated scheduling method of communication, perception and control:

[0314] S21. The first stage of the perception-control cycle triggering method: The automatic guided vehicle sends the collected sensor information to the drone through the horn antenna. At the same time, the wireless access point sends a perception signal to obtain the position information of the automatic guided vehicle and the drone. Then, the controller in the wireless access point calculates the misalignment error, which is the key indicator for activating the drone control system. The controller in the wireless access point makes a judgment based on the misalignment error l related to the drone beam radius R. When l>R, the wireless access point will issue drone motion control and antenna angle control instructions, otherwise it will only issue antenna angle control instructions, where the control activation coefficient ε k for.

[0315]

[0316] S22. The second stage of perception-control event triggering method: Given that the UAV uses a horn antenna and the wireless access point uses an omnidirectional antenna, the alignment of the UAV beam with the wireless access point direction within a specific range can ensure the transmission of backhaul data. To this end, an event triggering method is introduced. In the second stage, when the UAV transmits data to the wireless access point, the communication function of the wireless access point is continuously activated to receive data from the UAV. The perception function will only perform a perception when the trigger condition is met. The purpose of this is to maximize the transmission of backhaul data. From a control perspective, the event triggering method can reduce the number of perception-control times and allocate sufficient time resources for data transmission on the backhaul link. In this method, the controller should generally maintain a local estimation state. The estimated state can be expressed as:

[0317]

[0318] When the estimated value of the above formula cannot maintain the control performance and guarantee the communication requirements, the actual state of the UAV is sensed by the wireless access point. Then the estimated value of the state at time slot k is:

[0319]

[0320] For simplicity, it is assumed that the UAV is hovering on a horizontal plane, thus simplifying the data transmission of the backhaul link. The error between the actual state and the estimated state of the UAV on the controller can be determined as:

[0321]

[0322] Analogous to the signal-to-noise ratio in communication, this application proposes the concept of state-to-noise ratio, that is, From this, the perception-control activation threshold p is derived tr The relationship between the state noise ratio and the perception-control activation threshold p tr The relationship between and the state-to-noise ratio threshold is, therefore, the rate threshold of perception-control activation can be obtained:

[0323]

[0324] Therefore, the second stage of perceptual mode activation conditions are as follows:

[0325]

[0326] The control activation mode is determined based on the drone beam projection area. The activation conditions for the second stage control mode are as follows:

[0327]

[0328] Furthermore, the expression of the non-convex problem of maximizing the ratio of the total communication volume to the total power consumption of the industrial Internet of Things system based on integrated relay transmission of communication, perception and control is as follows:

[0329] S31, the division point, communication power and sensing beam vector in the joint optimization stage are used to obtain the ratio of the maximum communication volume to the total power consumption:

[0330]

[0331] Where: C1 and C2 represent the AP sensing accuracy constraints corresponding to the first and second phases, where ψ k is the first stage coefficient, which is 1 when in the first stage and 0 otherwise; P d is the beam pattern compared with the basic beamforming; τ is the minimum mean square error value compared with the basic beamforming;

[0332] C3 and C4 represent the maximum transmit power constraints for beamforming, where P l Indicates the perceived power upper limit;

[0333] C5 is the first stage serial interference elimination condition constraint;

[0334] C6 is the communication power constraint of the second-stage UAV;

[0335] C7 is the cache constraint of the drone;

[0336] C8 represents the integer constraint of the time slot division point;

[0337] C9 represents the 0-1 constraint of the perception and control activation factors. Since the quadratic form of the covariance matrix makes the constraints C2-C6 non-convex, we need to obtain the problem P 0 The global optimal solution is challenging.

[0338] It should be noted that P 0 is a mixed integer nonlinear programming problem because it involves integer variables k th and the perceptual beamforming vector w m,k ,v k And the automatic guided vehicle or UAV communication transmission power P c,k ,P u,k Since the nonlinear objective function cannot be clearly expressed using optimization variables, we can directly solve P 0 It is very difficult, so an effective decomposition and iteration method is used to effectively solve P 0 .

[0339] Furthermore, the process of converting the first-stage system into an approximate convex problem using the golden section-based Dinkelbach algorithm, penalty term and Taylor expansion method is as follows:

[0340] S41. Due to the quadratic form of the covariance matrix and P 0 The fractional form of the objective function makes it challenging to efficiently solve this mixed integer nonlinear problem. In addition, the perception beamforming vector changes with the change of UAVs and AGVs, and affects the communication rate and perception power. Therefore, in order to effectively solve this problem, it is decomposed into two sub-problems according to the two stages of relay transmission. One is the beamforming and power allocation sub-problem in the first stage, and the other is the time slot division, beamforming and power allocation sub-problem in the second stage. Specifically, in the kth time slot, the beamforming and power allocation sub-problem of the first stage is expressed as:

[0341]

[0342] stC1,C3,C5,C7,C9.

[0343] The time slot partitioning point, beamforming and power allocation sub-problems are expressed as:

[0344]

[0345] stC2,C4,C6-C9.

[0346] S42, due to P 1 The objective function of is a fractional non-convex function, so P 1 It is a non-convex problem. And the numerator is the automatic guided vehicle communication transmission power P c is a concave function of , and the denominator is a convex function of power. Therefore, P 1 It can be transformed into a concave function maximization problem through the Dinkelbach method, namely:

[0347]

[0348] stC1,C3,C5,C7,C9.

[0349] Among them, η k =R R,k / P c,k +P t,k +ε k P f,k And if we find Make If the optimization problem P 1 The solution can be obtained by solving the equivalent problem P 1.1 And get.

[0350] S43. Next, define auxiliary variables in, rank(Wm,k )=1, then P 1.1 Rewritten as:

[0351]

[0352] stC1, C3, C5, C7, C9,

[0353]

[0354] C11:rank(W m,k )=1.

[0355] The semidefinite relaxation method is considered to be an effective method for solving the rank-one constraint C11. The general solution obtained by the semidefinite relaxation method is reconstructed into a rank-one solution using eigenvalue decomposition or Gaussian randomization method, and the rank-one constraint is converted into a penalty term in the objective function. Then, the continuous convex approximation algorithm is used to solve the problem. Then, the rank-one constraint is equivalent to:

[0356]

[0357] Among them, ||·|| * is the nuclear norm, i.e., the sum of the singular values ​​of the matrix, ||·|| 2 is the spectral norm, i.e. the largest singular value of the matrix. Therefore, if the matrix W m,k is a rank-one matrix, then the above equation holds. Otherwise, since W m,k is a semi-positive definite matrix, the sum of the singular values ​​is always greater than the largest singular value, that is, ||W m,k || * -||W m,k || 2 > 0, in order to obtain a rank-one matrix, a penalty term is introduced in the objective function, and the result is:

[0358]

[0359] stC1, C3, C5, C7, C9,

[0360]

[0361] Among them: ξ is the penalty factor, only when it tends to 0, that is When W approaches ∞, m,k The rank of is close to 1;

[0362] S44. In this case, P 1.2 The main non-convexity comes from the second term of the penalty term, which is used at the point The first-order Taylor expansion at is used instead:

[0363]

[0364] in: is with The eigenvector corresponding to the largest eigenvalue, then, P 1.2 The problem is similar to the following one:

[0365]

[0366] stC1,C3,C5,C7,C9,C10.

[0367] S45. Only when ξ is small enough, the penalty term is closer to the rank-one constraint, but this will cause a new problem: when ξ is small enough, the penalty term will approach infinity. At this time, the penalty term will become the dominant term of the objective function, and the true objective function value cannot be obtained. Therefore, the attenuation coefficient μ is introduced. 0 , by initializing a large value ξ, and then by ξ = μ 0 ξ gradually reduces it to a sufficiently small value to obtain the overall suboptimal solution, where 0<μ 0 <1, when the penalty term is small enough, that is, ||W m,k || * -||W m,k || 2 <ε 2 , jump out of the iteration,

[0368] Furthermore, the process of solving the problem using the continuous convex approximation method based on the golden section is as follows:

[0369] 1) Initialization: feasible golden search interval (η l ,η h ), iterative threshold κ, golden section coefficient And the golden section search upper and lower bound update formula υ l =η h -τ(η h -η l ),υ h =η l +τ(η h -η l ), a feasible Taylor expansion point Reduction factor μ 0 , iterative threshold ε 1 and ε 2 , n = 1;

[0370] 2) When |η h -η l When |>κ, iterate;

[0371] 3) Input η according to the continuous convex approximation method l and ηh Perform external iterations separately;

[0372]

[0373] 8) If f(η l )≤f(η h ), f(η h ) is assigned to f(η l ), l The value of η is assigned to l , h The value of is assigned to υ l , η l +τ(η h -η l ) is assigned to υ h ;

[0374] 9) When |η h -η l When |≤κ, the iteration ends;

[0375] 10) Set (η h +η l ) / 2 is assigned to According to the penalty-based double-layer continuous convex approximation method, the input Solving for P 1.4 , get the optimal communication power of the automatic guided vehicle Optimal sensing beamforming vector with wireless access point

[0376] Furthermore, the process of introducing the reconstruction rank-one solution method based on quadratic transformation in stages to transform the non-convex problem into the second approximate convex problem is as follows:

[0377] S51, in P 2 In the equation, since there is an integer variable k th , direct solution is hindered. As we all know, after determining the total time slot length, the choice of the dividing point between the two stages is limited, and a search algorithm can be used to determine the most suitable boundary point. Therefore, in solving P 2 When we fix k th In addition, in the second stage, the power consumption and communication volume of the first stage have been determined. Therefore, P 2 The problem statement can be rewritten as follows:

[0378]

[0379] stC2,C4,C6-C9.

[0380] S52, P 2.1The optimization objective function of has a multi-ratio fractional form. We first apply a quadratic transformation to decouple the objective function into:

[0381]

[0382] in, Then, the original problem P 2.1 The equivalent conversion is:

[0383]

[0384] stC2,C4,C6-C9.

[0385] And, if and only if When the optimization problem P 2.2 Equivalent to the optimization problem P 2.1 ,in is x k The maximum value of .

[0386] S53. In the second stage of relay, the sensor-control system adopts event-triggered activation mode. Only when the communication rate from the UAV to the wireless access point is lower than the set threshold, the perception-control system will be activated to obtain the location information of the UAV, and according to the misalignment error judgment, the control command of the wireless access point is used to control the UAV to adjust the antenna angle or motion control, and finally ensure data transmission. Based on this, when the perception-control system is not started, there is no need to calculate the perception beamforming. Therefore, we transform the optimization problem of each time slot into two forms:

[0387] S531: When the perception-control system is not activated and the wireless access point does not need to perform perception control operations on the drone:

[0388]

[0389] stC6-C9.

[0390] Question P 2.3.1 is a concave function that can be solved efficiently using the CVX toolbox and the MOSEK solver.

[0391] S532: When the perception-control system is activated and the wireless access point needs to perform perception control operations on the drone:

[0392]

[0393] stC2,C4,C6-C9.

[0394] Question P 2.3.2 is a non-convex function. The main non-convexity is R in the objective function. T ,k and v kNext, define the auxiliary variables in, rank(V k )=1, then P 2.3.2 Rewritten as:

[0395]

[0396] stC2,C4,C6-C9,

[0397]

[0398] C13:rank(V k )=1.

[0399] P 2.4 The main non-convex constraint of comes from the rank-one constraint of C13. Compared with the first stage, in the second stage, the wireless access point perceives that the target is only the drone, so it is feasible to choose an algorithm with lower complexity. For this type of problem, the common practice is to ignore the rank-one constraint, find a matrix that does not satisfy the rank-one constraint, and use eigenvalue decomposition to recover a suboptimal solution that satisfies the rank-one constraint. However, doing so will lose some performance and may cause the new solution to not satisfy the original constraints. For problem P 2.4 , there is always an optimal solution of rank one:

[0400]

[0401] Therefore, the newly constructed solution Satisfy all the constraints of the original problem and ensure Finally, P 2.4 can be restated as:

[0402]

[0403] stC2,C4,C6-C9,C12.

[0404] Question P 2.5 It is a quadratic semidefinite programming problem, which can also be solved efficiently by CVX toolbox and MOSEK solver.

[0405] Furthermore, the process of solving the second approximate convex problem by the interior point method of reconstructing the rank-one solution and obtaining the optimal energy efficiency of the second-stage system is as follows:

[0406] 1) Initialization: feasible auxiliary variable χ k 、UAV communication power P c,k , the beamforming vector V of the wireless access point k ;

[0407] 2) When xk When it does not converge, iterate;

[0408] 3) According to Find the best χ k ;

[0409] 4) If δ′ k = 0, by solving P 2.3.1 Get the communication power P of the drone c,k ;

[0410] 5) Otherwise, by solving P 2.5 Get the communication power P of the drone c,k With the beamforming vector V of the wireless access point k ;

[0411] 6) When the iteration ends, if δ′ k =0, based on the iteration Solve for the value of P 2.3.1 Get the optimal communication power of the drone

[0412] 7) Otherwise, based on the iteration Solve for the value of P 2.5 Get the optimal communication power of the drone Optimal beamforming vector with wireless access point

[0413] In order to verify the effectiveness of the method of the present invention, the application effect of the present invention is described in detail below in combination with simulation.

[0414] Simulation conditions

[0415] In the simulation scenario, the wireless access point is initially configured with N = 8 antennas. Perceived power interval P d =5. The beam radii of the UAV and the AGV at distance d are r = 0.65 and R = 0.85 respectively. Initial value of UAV power consumption calculation: Tip speed of rotor blade U tip =120; average rotor induced speed v o =4.03; fuselage drag ratio d o =0.6; blade profile power and induced power are P b =79.9 and P i =88.6; air density, rotor solidity and blade angular velocity are ρ=1.225, s=0.05 and w respectively o =0.503. Event triggering scheme activation probability p tr =0.4, the initial value of the iterative simulation in the first stage is η l =1,ηh =50,κ=0.1,μ 0 =0.4,ε 1 =0.1,ε 2 =10 -6 . Initial position Q 1 (-20,34.6,0) and Q 2 (20,34.6,10).

[0416] In the simulation, the communication data volume is defined as the total communication data volume in a single synaesthesia integration frame. The total energy efficiency is defined as the ratio of the total communication data volume of the system in a single synaesthesia integration frame to the total power consumption of the system.

[0417] There are four other comparison methods. The first is a UAV antenna angle sensing-control scheme with a period of N = 5. The communication and sensing models are consistent with the proposed method. The second is a UAV antenna angle sensing-control scheme with a period of N = 10. The third is an event-triggered UAV motion sensing-control scheme. The activation condition setting is consistent with the proposed method. The fourth is a communication method that does not perform any sensing-control operations.

[0418] Simulation content and result analysis

[0419] Simulation 1: Comparative analysis of the objective function value and penalty term value of the first-stage method under different initial values ​​of iterations and reduction coefficients as the number of iterations changes.

[0420] from Fig.10 It can be seen that the search interval of the proposed continuous convex approximation algorithm based on the golden section decreases with the increase of the number of iterations, and when the number of external iterations is 10, the search interval converges to 0. Moreover, the smaller the upper limit of the search interval, the faster the convergence speed and the smaller the convergence point.

[0421] from Fig.11 It can be seen that the objective function value of the continuous convex approximation algorithm based on the golden section converges rapidly with the number of internal iterations at different reduction coefficients. Moreover, the smaller the reduction coefficient, the greater the convergence value of the objective function. Fig.12 It can be seen that the corresponding penalty item values ​​converge to a value close to 0, and the smaller the reduction coefficient, the faster the convergence speed and the smaller the convergence value.

[0422] Simulation 2: Comparative analysis of how the objective function of the second-stage method changes with the number of iterations under different search initial values.

[0423] from Fig.13 It can be seen that under different initial values ​​of iteration, the objective function value of the reconstruction rank-one method based on quadratic change converges rapidly with the increase of the number of iterations, and under different initial values ​​of iteration, the convergence values ​​of different auxiliary variables are similar.

[0424] Simulation 3: Comparative analysis of the total amount of communication within a single synaesthesia integrated frame under different schemes as the stage division point changes.

[0425] from Fig.14 It can be seen that as the stage division point increases from 30 to 50, the data volume of a single synaesthesia integration frame increases first and then decreases. In addition, the proposed method shows a larger data volume compared with other comparison methods. Moreover, the maximum data volume appears at the stage division point of 39. This is because in a single synaesthesia integration, the distance between the automatic guided vehicle and the UAV is smaller than the distance between the UAV and the wireless access point. When the communication power upper limit is fixed, the communication rate of the first stage will be slightly higher than that of the second stage, and the total communication data volume of the system is closely related to the second stage. Therefore, in the subsequent simulation, we set the stage division point of the first and second stages to 39.

[0426] Simulation 4: Comparative analysis of the total amount of communication within a single synaesthesia integrated frame and the total energy efficiency under different schemes as the upper limit of drone communication power changes.

[0427] from Fig.15 It can be seen that in the three cases where the communication power of the automated guided vehicle is determined, the communication data volume of the five methods gradually increases with the increase of the upper limit of the UAV communication power, and finally reaches convergence. This is because the amount of transmitted data is determined by the total amount of data that can be transmitted in the first stage, and is therefore affected by the communication rate in the first stage, that is, the upper limit of the communication power of the automated guided vehicle. It can be seen that the larger the upper limit of the communication power of the automated guided vehicle, the larger the communication data volume of the system. It can also be seen that the proposed scheme is significantly better than the other four schemes.

[0428] from Fig.16 It can be seen that in a single synaesthesia integration frame, the total energy efficiency of the system increases with the increase of the UAV communication power limit and then tends to be stable. In addition, the total energy efficiency is mainly determined by the UAV's cache and total energy consumption in the first stage, where the energy consumption is further determined by the power. In addition, compared with the periodic antenna angle control method, the proposed method has a certain performance gain.

[0429] Simulation 5: Comparative analysis of the total amount of communication and total energy efficiency within a single synaesthesia integrated frame under different schemes as the UAV's flight altitude changes.

[0430] from Fig.17 It can be seen that as the flying altitude of the drone increases, the communication rate of both the first and second phases decreases. Therefore, the amount of data that can be received in a single synaesthesia integrated frame also decreases. In addition, compared with the other four methods, the communication data volume of the proposed method is larger. Among them, compared with the other four methods, the performance improvement of the proposed method is 9.32%, 20.47%, 7.97% and 336.81% respectively.

[0431] from Fig.18 It can be seen that as the flight altitude of the drone increases, the total energy efficiency of the system within a single synaesthesia integration frame gradually decreases. In addition, since the drone motion control method consumes a lot of energy for motion control, the energy efficiency of the event-triggered drone motion sensing-control scheme is lower than that of our proposed method and the periodic antenna angle control method. Among them, the energy efficiency of the method proposed in this paper is improved by 12.27%, 24.2%, 69.18% and 115.02% respectively compared with the other four methods. Simulation 6: Comparative analysis of the beam pattern performance of the methods at different stages at different angles.

[0432] from Fig.19 It can be seen that in the first stage, the wireless access point perceives both the AGV and the UAV. However, in the second stage, the wireless access point only perceives the UAV as a target. In the first stage, in order to perform beamforming, our proposed double-layer continuous convex approximation method based on penalty terms has two main lobes with roughly the same power in different target directions.

[0433] from Fig. 20 It can be seen that in the second stage, it can be observed that the beamforming based on the quadratic change reconstruction rank one method can achieve a large power peak in the target direction. In addition, the proposed method has obvious power gain in the target direction and less power leakage in the non-target direction, which further verifies the feasibility and efficiency of the proposed method.

[0434] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A relay-based integrated method for communication and control in industrial Internet of Things, characterized in that: The steps include: Introduce automated guided vehicles and drones into the industrial Internet of Things system, and establish the channel model, communication model, perception model and control model of the industrial Internet of Things system with integrated relay transmission of communication, perception and control divided into two stages; According to the actual situation of half-duplex relay transmission, the industrial Internet of Things system based on integrated relay transmission of communication, perception and control is divided into two stages, and the integrated scheduling method of communication, perception and control is adopted to control data transmission; Calculate the division points, communication power and sensor beams of the two phases of the industrial Internet of Things system based on integrated communication, perception and control relay transmission, and establish a model to maximize the total energy efficiency of the industrial Internet of Things system based on integrated communication, perception and control relay transmission; The total energy efficiency model of the industrial Internet of Things system is divided into two stages: the first stage system and the second stage system. The first stage system is converted into a first approximate convex problem using the Dinkelbach algorithm, penalty term and Taylor expansion method, and the second stage system is converted into a second approximate convex problem using the quadratic transformation and reconstruction of the rank one solution method. The first approximate convex problem is solved by a continuous convex approximation algorithm based on the golden section, and the optimal energy efficiency of the first-stage system is obtained. The second approximate convex problem is solved by an interior point method of reconstructing a rank-one solution, and the optimal energy efficiency of the second-stage system is obtained. The two stages of the industrial Internet of Things system with integrated communication, perception and control relay transmission include: Phase 1: Data transmission between the AGV and the UAV; Phase 2: The UAV transmits data to the wireless access point; The channel model of the industrial Internet of Things system with integrated communication, perception and control relay transmission mainly includes multipath fading, path loss and misalignment fading. The channel gains of the two stages are expressed as follows: h1=h p h f h m h2=h p h f h a Where: path loss h p Described by the free space propagation model, the expression is: Among them, G t and G r are the transmit and receive signal gains, respectively, λ = f / c is the signal wavelength, where c is the speed of light, f is the occupied frequency band, and d is the distance between the transceivers; Where: multipath fading h f It is usually described by Rayleigh distribution, whose probability distribution function for: in: is a fading channel with an RMS value of α = 2, and the probability distribution function of x is The independent variable; The first stage misalignment fading h m :Assume that the transmitting beam of the UAV covers an area S with a radius of R, and the radius of the beam coverage area of ​​the automatic guided vehicle is r, and the distance is d, where 0≤R≤R m , R m is the maximum radius of the beam at distance d, and l is the beam center O of the automatic guided vehicle AGV With the drone’s beam center O UAV The misalignment error between m It is expressed as: The equivalent beam width is expressed as R e Indicates that θ is the angular error of the horn antenna. The misalignment error in the first stage only considers the tangent plane direction of the optimal direction, that is, in is the misalignment error coefficient. In addition, when l = 0, the received power of the drone is expressed as P0, which is obtained by the following formula: in, erf(·) is the Gaussian error function, then the equivalent beam width R e It is expressed as: The second stage misalignment fading h a It is expressed as: in, is the projected area in the best case, and similarly, is the projection area of ​​the drone beam on the wireless access point beam, where and θ0 = 2arctan(r / d) are the UAV beam radius and the UAV beam angle; in addition, the misalignment error L is expressed as: In the perception channel of integrated relay transmission for communication, perception and control: the wireless access point sends perception signals through a uniform linear array to perceive the positions of the automatic guided vehicle and the UAV. The perception channel is expressed as: Where β is the channel gain value at 1m away from the wireless access point, m∈M,Z a are the x- and y-axis coordinates of the wireless access point. The position information of the AGV and the UAV are expressed in coordinates as follows: G1(Z1(g x1 ,g y1 ),g z1 ) and G2(Z2(g x2 ,g y2 ),g z2 ), In the first-stage communication model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the signal y received by the drone u It is expressed as: y u =h1s a +h s x+ω Among them, s a represents the communication signal sent by the automatic guided vehicle, x represents the sensing signal sent by the wireless access point, h1 represents the flat fading wireless channel between the automatic guided vehicle and the UAV, and h s represents the wireless channel between the wireless access point and the drone, ω represents the additive noise with zero mean and variance N0; UAV receiving achievable rate C R,k It is expressed as: Among them, P c,k Represents the transmission power of the communication signal, P s,k represents the power of the sensing signal received by the drone, and N0 is the additive noise power; In the first stage, the k-th time code rate R transmitted by the AGV to the UAV R,k It is expressed as: R R,k =C R,k Assuming that the drone successfully decodes the codeword and stores the corresponding information in its buffer, use Q k represents the amount of data in the relay buffer at the end of the kth time slot, so Q n,k Will change as follows: In the first-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the key indicator of the perception function is the covariance matrix of the perception beam distribution determined by the transmitted signal. The covariance matrix R ω,k It is expressed as: Among them, w m,k is the transmit beamforming vector of the sensing signal, and the wireless access point transmits the sensing signal power P t,k For: P t,k =Tr(R ω,k ) In Directional sensing beam pattern for: in, represents the steering vector when a uniform linear array is used as an antenna at a wireless access point, λ represents the wavelength, d0 represents the antenna spacing, and N0 is the number of linear array antennas at the wireless access point; Therefore, the wireless access point senses the power P of the drone in the kth time slot s,k It can be expressed as: Using the position information of the AGV and the UAV, determine the horizontal angle φ of the sensing target relative to the wireless access point m,k for: In the second-stage communication model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the signal y received by the wireless access point a It is expressed as: y a =h2s u +d k h s x e +oh, Where ω is the power of the additive noise, s u The signal received by the UAV after decoding and forwarding the communication signal sent by the automatic guided vehicle, x e is the sensing echo signal reflected back to the wireless access point, δ k is the perceptual activation coefficient; channel capacity C T,k It is derived as follows: Where: P u,k is the communication transmission power of the UAV; The kth time slot code rate R of the wireless access point T,k It is expressed as: R T,k =min{Q k-1 ,C T,k }, When R T,k ≤C T,k When , the wireless access point can successfully decode the codeword, and as the drone relays and transmits it, the number of codewords in the buffer becomes: In the second-stage perception model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the covariance matrix R v,k It is expressed as: Among them, v k is the transmit beamforming vector of the sensing signal; then, the wireless access point transmits the sensing signal power P in the second stage v,k for: P v,k =Tr(R v,k ) In Directional sensing beam pattern for: In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone motion control model is as follows: The discrete time control model of UAV motion control is: x k+1 =A d x k +B d u k +w k , Among them, x k is the state of the drone at time k, A d and B d is the system parameter matrix in the discrete-time model, u k is the motion control input, w k is the disturbance caused by additive Gaussian white noise with zero mean and variance W. Assuming that at each time slot k, the sampling state of the wireless access point is perfect, that is, y k =x k , the motion control input of the drone is calculated as: you k =Ky k , Where K is the control gain, for a speed of V k The propulsion power consumption of the rotor UAV is P f,k (V k ) is modeled as: Among them, U tip is the rotor tip speed, v o is the average rotor induced speed, d o is the fuselage drag ratio, P b and P i are blade profile power and induced power, ρ, s and w respectively o are respectively the air density, rotor solidity and blade angular velocity. The speed Z of the UAV in the discrete domain k Calculated by the following formula: Among them, Δ0 represents the time interval; In the control model of the industrial Internet of Things system with integrated communication, perception and control relay transmission, the drone antenna angle control model is as follows: the distance and relative angle between the drone and the automatic guided vehicle can be obtained through perception measurement, the direction of the horn antenna on the drone can be expressed by the pitch angle θ and the azimuth angle φ, and the angle of the horn antenna pointing to the device direction is calculated as θ d ,in: Then, the error in the horn antenna pointing direction at the kth moment is e θ,k =θ k -θ d,k ,θ k is the antenna pointing angle of the drone at the kth moment, obtained through the drone's built-in sensor, and the controller is used to control the direction of the horn antenna and the output z k It is expressed as: With k =K a θ k , Among them, K a To control the gain, the discrete time control model of the UAV horn antenna angle control is: i k+1 =A a i k +B a z k +w a,k , Among them, A a and B a are the system parameter matrices, w a,k is the disturbance at time k, θ k+1 represents the antenna angle of the UAV in time slot k+1; According to the actual situation of half-duplex relay transmission, the industrial Internet of Things system architecture based on integrated relay transmission of communication, perception and control is divided into two stages, and the process of data transmission control is carried out by adopting an integrated scheduling method of communication, perception and control as follows: S21. The first stage of the perception-control cycle triggering method: the automatic guided vehicle sends the collected sensor information to the UAV through the horn antenna. At the same time, the wireless access point sends a perception signal to obtain the position information of the automatic guided vehicle and the UAV. Then, the controller in the wireless access point calculates the misalignment error, which is the key indicator for activating the UAV control system. The controller in the wireless access point makes a judgment based on the misalignment error l related to the UAV beam radius R. When l>R, the wireless access point will issue UAV motion control and antenna angle control instructions, otherwise it will only issue antenna angle control instructions, where the control activation coefficient ε k for: S22, the second stage of perception-control event triggering method: when the UAV is transmitting data to the wireless access point, the communication function of the wireless access point is continuously activated to receive the data from the UAV. Only when the trigger condition is met, the perception function will perform a perception, and the controller will maintain the local estimation state. It is expressed as: When the estimated value of the above formula cannot maintain the control performance and guarantee the communication requirements, the actual state of the UAV is sensed by the wireless access point, and the estimated value of the state at time slot k is: in: is the state estimate, x k It is the actual state; Assuming the drone is hovering on a horizontal plane, determine the error e between the actual state and the estimated state of the drone on the controller k+1 for: State-to-Noise Ratio From this, the perception-control activation threshold p is derived tr Relationship with state-to-noise ratio; The perception-control activation threshold p is obtained tr The relationship between the state noise ratio threshold, that is, Therefore, the rate threshold of perception-control activation is obtained The second stage of perception mode activation conditions are as follows: The control activation mode is determined according to the UAV beam projection area. The activation conditions of the second stage control mode are as follows: The expression of the non-convex problem of maximizing the ratio of the total communication volume to the total power consumption of the industrial Internet of Things system with integrated communication, perception and control relay transmission is as follows: S31, the division point, communication power and sensing beam vector in the joint optimization stage are used to obtain the ratio of the maximum communication volume to the total power consumption P0: Where: C1 and C2 represent the AP sensing accuracy constraints corresponding to the first and second phases, where ψ k is the first stage coefficient, which is 1 when in the first stage and 0 otherwise; P d is the beam pattern compared with the basic beamforming; τ is the minimum mean square error value compared with the basic beamforming; C3 and C4 represent the maximum transmit power constraints for beamforming, where P l Indicates the perceived power upper limit; C5 is the first stage serial interference elimination condition constraint; C6 is the communication power constraint of the second-stage UAV; C7 is the cache constraint of the drone; C8 represents the integer constraint of the time slot division point; C9 represents the 0-1 constraint of the perception and control activation factor; Constraints C2-C6 are non-convex due to the quadratic form of the covariance matrix.

2. According to the relay-based integrated method for communication and control in industrial Internet of Things in claim 1, it is characterized in that: The process of converting the first-stage system into a first approximate convex problem using the Dinkelbach algorithm, penalty term and Taylor expansion method is as follows: S41, the beamforming and power allocation sub-problem of the first stage is expressed as: stC1,C3,C5,C7,C9 The time slot partitioning point, beamforming and power allocation sub-problems are expressed as: stC2,C4,C6-C9 The objective function of S42 and P1 is a fractional non-convex function. P1 transforms the objective function into a concave function maximization problem through the Dinkelbach method, that is: stC1,C3,C5,C7,C9 Among them, η k =R R,k / P c,k +P t,k +ε k P f,k , and if found Make If true, then the solution to the optimization problem P1 is obtained by solving the equivalent problem P 1.1 and get; S43. Next, define auxiliary variables Among them, W m,k ≥0, rank(W m,k )=1, then P 1.1 Rewritten as: stC1,C3,C5,C7,C9, C11:rank(W m,k )=1 Use eigenvalue decomposition or Gaussian randomization to reconstruct the general solution obtained by the semidefinite relaxation method into a rank-one solution, transform the rank-one constraint into a penalty term in the objective function, and then use a continuous convex approximation algorithm to solve the problem. The rank-one constraint is equivalent to: Among them, ||·θ|| * is the nuclear norm, i.e. the sum of the singular values ​​of the matrix, and ||·||2 is the spectral norm, i.e. the largest singular value of the matrix. Therefore, if the matrix W m,k is a rank-one matrix, then the above equation holds. Otherwise, since W m,k is a semi-positive definite matrix, the sum of the singular values ​​is always greater than the largest singular value, that is, ||W m,k || * -||W m,k ||2>0, in order to obtain a rank-one matrix, a penalty term is introduced in the objective function, and the result is: stC1,C3,C5,C7,C9, Among them: ξ is the penalty factor, only when it tends to 0, that is When W approaches ∞, m,k The rank is close to 1; S44. In this case, P 1.2 The main non-convexity comes from the second term of the penalty term, which is used at the point The first-order Taylor expansion at is used instead: in: is with The eigenvector corresponding to the largest eigenvalue, then, P 1.2 The problem is similar to the following one: stC1,C3,C5,C7,C9,C10 S45, introduce the attenuation coefficient μ0, initialize a large value ξ, and then gradually reduce it to a sufficiently small value through ξ=μ0ξ, and obtain the overall suboptimal solution, where 0<μ0<1. When the penalty term is small enough, that is, ||W m,k || * -||W m,k ||2<ε2, jump out of iteration.

3. The method for integrated communication and control based on relay in industrial Internet of Things according to claim 1 is characterized in that: The process of using the continuous convex approximation algorithm based on the golden section to solve the first approximate convex problem and obtain the optimal energy efficiency of the first stage system is as follows: 1) Initialization: feasible golden search interval (η l ,η h ), iterative threshold κ, golden section coefficient And the golden section search upper and lower bound update formula υ l =η h -τ(η h -η l ), h =η l +τ(η h -η l ), a feasible Taylor expansion point Reduction coefficient μ0, iteration thresholds ε1 and ε2, n = 1; 2) When |η h -η l When |>κ, iterate; 3) Input η according to the continuous convex approximation method l and η h Perform external iterations separately; 4) When || W m,k || * -||W m,k When ||2≥ε2, perform inner iteration; 5) When When, according to Solving for P 1.4 , and based on Update Taylor expansion points n=n+1; 6) When When , jump out of the inner iteration, The value of ξ=μ0ξ; 7) When || W m,k || * -||W m,k ||2<ε2, jump out of the external iteration and calculate f(η l ) and f(η h ); 8) If f(η l )≤f(η h ), f(η h ) is assigned to f(η l ), l The value of η is assigned to l , h The value of is assigned to υ l , η l +τ(η h -η l ) is assigned to υ h ; 9) When |η h -η l When |≤κ, the iteration ends; 10) Set (η h +η l ) / 2 is assigned to According to the penalty-based double-layer continuous convex approximation method, the input Solving for P 1.4 , get the optimal communication power of the automatic guided vehicle Optimal sensing beamforming vector with wireless access point 4. The method for integrated communication and control based on relay in industrial Internet of Things according to claim 1 is characterized in that: The process of converting the non-convex problem into the second approximate convex problem by using the quadratic transformation and reconstruction rank-one solution method in the second stage system is as follows: The statements of S51 and P2 are rewritten as follows: stC2,C4,C6-C9 S52, P 2.1 The optimization objective function has a multi-ratio fractional form. First, a quadratic transformation is applied to decouple the objective function into: in, Then, the original problem P 2.1 The equivalent conversion is: stC2,C4,C6-C9 And, if and only if When the optimization problem P 2.2 Equivalent to the optimization problem P 2.1 ,in is x k The maximum value of S53, convert the optimization problem of each time slot into two forms: S531: When the perception-control system is not activated and the wireless access point does not need to perform perception control operations on the drone: stC6-C9 Question P 2.3.1 is a concave function that can be solved efficiently using the CVX toolbox and the MOSEK solver; S532: When the perception-control system is activated and the wireless access point needs to perform perception control operations on the drone: stC2,C4,C6-C9 Question P 2.3.2 is a non-convex function. The main non-convexity is R in the objective function. T ,k and v k Next, define the auxiliary variables Among them, V k ≥0, rank(V k )=1, then P 2.3.2 Rewritten as: stC2,C4,C6-C9, C13:rank(V k )=1 P 2.4 The main non-convex constraint comes from the rank-one constraint of C13. For the problem P 2.4 , there is always an optimal solution of rank one: Therefore, the newly constructed solution Satisfy all the constraints of the original problem and ensure Finally, P 2.4 Rephrased as: stC2,C4,C6-C9,C12 Question P 2.5 It is a quadratic semidefinite programming problem, which is solved efficiently using the CVX toolbox and the MOSEK solver.

5. According to the relay-based integrated method for inter-sensory control in industrial Internet of Things in claim 1, the process of solving the second approximate convex problem by using the interior point method of reconstructing the rank-one solution to obtain the optimal energy efficiency of the second-stage system is as follows: 1) Initialization: feasible auxiliary variable χ k 、UAV communication power P c,k , the beamforming vector V of the wireless access point k ; 2) When x k When it does not converge, iterate; 3) According to Find the best χ k ; 4) If δ k '=0, by solving P 2.3.1 Get the communication power P of the drone c,k ; 5) Otherwise, by solving P 2.5 Get the communication power P of the drone c,k With the beamforming vector V of the wireless access point k ; 6) When the iteration ends, if δ′ k =0, based on the iteration Solve for the value of P 2.3.1 Get the optimal communication power of the drone 7) Otherwise, based on the iteration Solve for the value of P 2.5 Get the optimal communication power of the drone Optimal beamforming vector with wireless access point