Positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty

Abstract

A positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty is provided, which includes following steps: S1, establishing a re

Description

The disclosure relates to the technical field of ocean sensor networks, and in particular to a positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty. Ocean sensor network (OSNs) is one of the important supporting means for the exploitation of marine resources and maritime security. As an important part of the important supporting means, surface wireless sensor networks (SWSNs) can not only assist the application of data transmission and positioning of underwater sensor network (UWSNs), but also be an important intermediate link for building an air-sea network and cross-media communication network. In SWSNs, if there is no coordinate information in the transmitted marine data, it is impossible to accurately determine the location of marine resources, so it is very important to obtain the position information. Considering the cost, not all nodes (buoys, pontoons, etc.) in the network are equipped with GPS receivers. Therefore, how to locate these nodes without GPS to ensure that the information collected by SWSNs nodes has position information is an important scientific issue. However, the aging of equipment and the dynamic ocean environment have brought severe challenges to SWSNs positioning technology. On the one hand, the aging of equipment will lead to the uncertainty of the transmission power (TP) of nodes, which will lead to the deviation between the rated TP and the actual TP of nodes in SWSNs, thus reducing the positioning accuracy. On the other hand, the ocean environment is complex and changeable, and the meteorological conditions are harsh. The time-varying humidity and temperature will make the path loss factor (PLE) in the environment highly uncertain, which will lead to the deviation of RSS observations, thus increasing the positioning error. The existing positioning methods may not guarantee the positioning accuracy in the dynamic and changeable ocean environment, especially under the condition that TP and PLE parameters are uncertain at the same time, the positioning accuracy of SWSNs is low, which seriously affects the functionality of SWSNs. The objective of the disclosure is to provide a positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty, so as to ensure the positioning accuracy of SWSNs. The objective of the disclosure can be achieved by the following technical scheme. The disclosure relates to a positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty, which includes the following steps: SI, establishing a restricted motion model of offshore nodes, setting node positions and anchor node positions of target nodes, and establishing a ranging model of the surface wireless sensor networks; S2, performing a differential operation on the ranging model, establishing a natural constant least squares framework, and solving the natural constant least squares framework by a differential linear unbiased estimation method to obtain estimated values of coarse-grained target nodes; S3, constructing an optimization function with path loss factors as variables based on the estimated values of the coarse-grained target nodes in the S2, and improving a Puma Optimizar Algorithm by combining the quantum computation theory with a good point set method to solve an optimal path loss factor; and S4, based on the optimal path loss factor in the S3 and matrix multipliers, establishing a framework of differential generalized trust region subproblem; combining with Lagrange multipliers, obtaining fine-grained target node positions by a dichotomy method. x = h x f Further, the node positions of the target nodes are L p 2J , an(j a pOSipon of an i-th anchor - . Cl — o 1 11 . i i i n -1 1 node is ' L , and the ranging model ot the surface wireless sensor networks is as follows: x-a =P0-Wa log10 + , 4 p where n represents transmission powers from the target nodes received by the i-th anchor d P node; 0 is reference distances, 0 is transmission powers of the target nodes; a represents the path loss factors, and represents Gaussian distribution noise with a mean value of 0 and a f ¢72 variance of 1 . Further, specific steps of performing the differential operation on the ranging model and establishing the natural constant least squares framework are as follows: performing the differential operation on the ranging model, changing a base number of an obtained differential result, introducing a natural constant e, performing shifting and squaring operations, and then performing linear expansion according to a natural constant Taylor series first-order expansion formula to define a variable [Ji’X2’Z] where constructing the natural constant least squares framework based on an expanded formula and the variable the differential result is: a / = 1 >ogio I—4+%, If-"ill where (7=2.LC ; Further, the natural constant least squares framework is: arg min ||30-T||2, 0 2a21 2p22au 2a22 2p2,\an 2a — 2p31au 2^32 — 2 / ^31^12 M M 2am — 2Pn,Pu 2aN2 — 2pN,\a\2 2_ a where N represents a total number of anchor nodes. Further, specific steps of solving the natural constant least squares framework by the differential linear unbiased estimation method to obtain the estimated values of the coarse-grained target nodes are as follows: solving the natural constant least squares framework to obtain estimated values of initial positions of the target nodes, and linearly expanding v ’ at x close to x by using the differential linear unbiased estimation method, where is the estimated values of the initial positions of the target nodes, X| , *2 and are three elements of 0 respectively, and ^.o r?.0 ~i7 X = x , L J , after expansion, the estimated values of the coarse-grained target nodes are obtained, and the estimated values of the coarse-grained target nodes is: x =£'-(r7373r) 'r^ r ~iT S= r ~iT r= i2,2(x°)t where I is identity matrix. Further, the S3 includes following specific steps: S31, constructing the optimization function with the path loss factors as variables based on the estimated values of the coarse-grained target nodes of the S2, and then initializing a population number, maximum iterations and initial iterations, and taking the initial iterations as current iterations, where population individuals are the path loss factors; S32, generating an initial population by combining the quantum computation theory with the good point set method, and calculating an initial fitness value; S33, judging whether the current iterations are less than or equal to 3, and if the current iterations are less than or equal to 3, executing S34; if the current iterations are more than 3, executing S35; S34, running exploration and exploitation steps, performing inexperienced calculation, calculating a score of the exploration and exploitation steps, updating the current iterations, and returning to the S3 3; S35, judging whether the current iterations are less than or equal to the maximum iterations, if the current iterations are less than or equal to the maximum iterations, performing experienced calculation, and running the exploration or exploitation steps according to the score of the exploration and exploitation steps of a previous iteration; if a new fitness value is less than a fitness value of the previous iteration, updating an optimal population, and then performing S36, and then performing S36; if the new fitness value is not less than the fitness value of the previous iteration, directly performing S36; and S36, updating the score of the exploration or exploitation steps and the current iterations, and returning to S35 until the current iterations are greater than the maximum iterations, and outputting an optimal population corresponding solution at this time, where the optimal population corresponding solution is the optimal path loss factor. Further, the exploration specifically includes: randomly generating a value within a range of based on ran^ function, where if the value is greater than 0.5, then according to: Zn^=R^^Ub-Lb) + Lb, Z obtaining a first parameter ” G; on a contrary, according to: 7 — X +GF + g((F ^F \ + (f ^F 11 ^n,G ^a,G T \\ >'h ^c.d ) T\^c.d ^e,f )) ’ z obtaining a first parameter ” G; where n is a current population number; Dim represents a randomly generated number in a case of uniform distribution of [°’1]; G = 2-rand-l- Ea.b ~ Xa,a~ Xb,G Ub an(j Lb are F = X — Y F = X — X X upper bound and lower bound, respectively; c-d C’G rfG; e-f e’° f’° \ a’G, Xg , Xgg , Xd,a, xe.G an(j x / ,g represent solutions of a whole population; updating relevant solutions after obtaining the first parameter, where specific steps of updating the relevant solutions are as follows: new Z„,G. if m = m^ or rand <U, X„, otherwise NC = 1-U p = nc / m: after updating the relevant solutions, calculating fitness of updated solutions, and then executing: if fit(-^new) <fit(Xn) sahsfied; executing ^a’° , an(j th gn executing U -U + p if fit(X< fit[Xis not satisfied, directly executing td-U + p anj returning to the specific steps of updating the relevant solutions; X ni where m represents individuals and new represents the updated solutions; rand represents randomly generated integers; represents preset parameters between 0 and 1; M is a total number of pumas; is a fitness function, and "represents solutions corresponding to the current population; the exploitation is specifically: randomly generating a value within the range of based on ran^ function, and performing updating according to the value, where an updated solution is: meanfSof ,,) \ total / ‘drand >Q.5,X l + (pc-rand} new = <oth erwi se, if rand >L, Xnew otherwi se,Xnew = (2 x rand) x ( / v / ^ / ,) + / ya- / 0-PumahJ (2 • rand -1 + randn} -pumabest meaxi. (*) Sol Xi^ where ' ' represents an average function; total represents a sum of all solutions; 1 is a randomly selected solution in the whole population; is a number randomly selected from 0 and 1; and L represent optimization parameters preset in advance; Pumab«t js an optimal puma at present; randn represents a Gaussian distribution function with a mean value of 0 and a variance of 1; X> ^ound^M-^rand). R = 2.mnd_1; randn-exp 2 Iter- < 2 Maxlter ) F2 = wx(y)2 • cos((2xrand}-w), where ^ter represents the current iterations; Maxlter represents the maximum iterations; w and v are generated by the Gaussian random distribution function randn respectively. Further, the scores of the exploration and exploitation steps with the inexperienced calculation in the S34 are: Score* = (PF, • fu lEr ) + (PF • fu 2Er), nr y i tji y y l u nr y * = (PF • / ; 1 Ei ) + (PF • / ; 2E1), tsi y i U IS1 / y z, «2 U til / where, ^dfitEi ( = PF- JpEl 1 1 1 ^dfAEr + ^^fitEr + ^^fitEr Seq^+Seq^ + Seq^ = pp ^g^fitEi + ‘^Viitio + ^Viitiii EI 2\Seq\me+Seq2ime+Seq2ame} ^eq&Er = |^^best - fi^Er | ’ ^^fitEr = l-Ahr “ ^Er|’ ^^fitEi = |^^best “ ^Ei | ’ ‘'WfitEi = |^^Ei - . / ^Ei | ’ =|X-X|> f i f i f । where Ju Er and Ju E1 respectively represent a function of using Ju to calculate exploration f 2 f 2 and exploitation in an inexperienced stage; Ju Er and Ju E1 respectively represent a function f 2 pp of using Ju to calculate the exploration and exploitation in the inexperienced stage; 1 and are preset parameters in advance, used to sort priorities of and -^2 functions; -fjp -fjp is set to 1; best is a fitness function value corresponding to an initial solution; and ■ Er are values corresponding to the fitness function in a first to a third calculation of exploration operations; *^E1, -^E1 and are corresponding values of the fitness function in a first to a third calculation of exploitation operations, respectively, and subscripts Er represents exploration and subscripts Ei represents exploitation; a score of experienced exploration and exploitation steps in S3 5 is: = 1E1 + 2B + . k . 3E3, FEr = lEr + 2Er + £Er -Ic-fX, where, f lEi = pp . fi^i J elt 1 1 1 rpEi ( i — / hE1 ,1 + ( 3 — fit -,) + (- — 1iP o) old,l J new, 1 ) \J old,2 J new, 2 J \^ old,3 J new, 3 J qnEl . nrrEl . qnEl + ^ / ,2 -^ / ,3 J e l ( Xi!,., ) + (, / ^,2 ) + (Xld3 rr< Er . t^Ei . t^Ef 1t,\ t,3 if selected, / e3E1 =0, otherwi se, fe 3 E1 + PF3, if selected, / e3fr=0, otherwise, fe^+PR, f 1^1 f I Er where ' and ' are respectively values corresponding to exploitation and exploration f 1 fitE1 iil steps in a 7-th iteration by using Je function in an experienced stage; J old and J old are respectively fitness function values of exploitation and exploration steps calculated under a previous iteration of the experienced stage; ■ new and - new are fitness function values of ^Ei exploitation and exploration steps calculated under a current iteration respectively; * and rEr f 2E1 are iterations not selected in the exploitation and exploration steps respectively, and Je ' 2 Er and 1 are respectively the values corresponding to the exploitation and exploration steps in the f 2 / / tE1 fitE' fitE1 7-th iteration by using function in the experienced stage; 7 old l, oli2 and old’3 respectively represent fitness function values of unoptimized solutions corresponding to exploitation steps in 7-th, -th and ('-2) -th iterations in the experienced stage; fir fir J old,2 anj J oid,3 are respectively fitness function values of the unoptimized solutions corresponding to exploration steps in the 7 -th, -th and -th iterations in the experienced stage; J , ' new-2 and J new-3 respectively represent fitness function values of optimal solutions corresponding to the exploitation steps in the 7-th, -th and -th ^Er -fji^ -fijE: iterations in the experienced stage respectively; ' ', ' new>2 and ' new’3 are are fitness function values of the optimal solutions corresponding to the exploration steps in 7-th, / y*Ei y^Ei y^Ei -th and ' ' -th iterations in the experienced stage respectively; u , t 2 and S '-3 respectively represent iterations not selected by exploitation steps in ^-thto 7-th, 0 / , i\ / , qnEr <F’Er ^Er 1 —I) 1 1 —3) , It — 2) 1 . . 7 , 7-, , 7 • 1 -th to v 7 -th and v 7 -th to K 7 -th iterations; M , M and respectively 1 • . , ( / -1) , , ( / -2) , represent iterations not selected by exploration steps in the v 7-th to -th, v -th to ( / -1) ( / -3) ( / -2) f3E1 f 3Er v 7 -th and 7 7 -th to v 7 -th iterations, and Je ‘ and Je 1 are values f 3 corresponding to exploitation and exploration steps in the 7-th iteration with function in pp pE' pEl the experienced stage; 3 is an advanced parameter, ' and ‘ respectively represent scores corresponding exploitation and exploration in the 1 -th iteration; and * and ‘ respectively represent parameters corresponding the exploitation and exploration in the z-th iteration. Further, each individual of the initial population occupies two positions in a search space as follows: mx = —(cos 0+ l)(Ub-Lb) + Lb, my (sin 0 + V)(Ub -Lb) + Lb. Ttl for a position 3, a corresponding position of another superposition state is: mx = (cos 0* + l)(Ub -Lb) + Lb. n* ■ I ~Lb^ 0 = arcsin ---------- Lb Lh 111 where a value domain of , m . \Ub,Lb] and 3 is L J. Further, the fine-grained target node positions are: where, 2a^ - 2-105“ «2r M 105“ -1 2<r ^2-10¾ M 105“ M II2.105« « =^10a*log: zy where is the optimal path loss factor. Compared with the prior art, the disclosure has the following beneficial effects. The disclosure is divided into three stages. In the first stage, the uncertainty of TP is eliminated by differential operation, and an NC-LSE framework is established to solve the coarse-grained position of the target node, thus realizing the positioning under the uncertainty of TP. In the second stage, the target optimization function with PLE as a variable is established, and the quality of the initial population is improved by combining quantum computation theory and good point set method, and then the PO algorithm is improved to solve PLE, thus eliminating the uncertainty of PLE. In the third stage, the PLE is substituted, and a fine-grained estimation framework based on DGTRS is established to solve the position of the target node. By using the third-order coarse-grained to fine-grained estimation method, the influence of two-parameter uncertainty is eliminated, and the accurate estimation of the position of the target node from coarse to fine is realized, which provides a feasible solution for the positioning problem of SWSNs in a highly dynamic and uncertain ocean environment. Figure lisa flow chart of the present disclosure. Figure 2 is a restricted schematic diagram of offshore nodes of the present disclosure. Figure 3 is a flow chart of PLE estimation of path loss factor in the second stage of the improved puma algorithm based on quantum computation of the present disclosure. Figure 4 is a comparison diagram of relative positioning errors between the present disclosure and other schemes under different anchor node numbers. Figure 5 is a comparison diagram of relative positioning errors between the present disclosure and other schemes under different environmental noises. The present disclosure will be described in detail with the attached drawings and specific embodiments. This embodiment is implemented on the premise of the technical scheme of the present disclosure, and the detailed implementation and specific operation process are given, but the protection scope of the present disclosure is not limited to the following embodiments. In order to improve the positioning accuracy of SWSNs under the condition of two-parameter uncertainty, the disclosure provides a positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty, and the third-order positioning from coarse-grained to fine-grained integrates the quantum calculation. The disclosure introduces the difference idea, eliminates the influence of TP uncertainty, constructs a natural constant-least square estimation (NC-LSE) framework, proposes a differential linear unbiased estimation to obtain the coarse-grained position, and based on this, establishes a target optimization function with PLE as a variable. The initial population quality is optimized by using quantum computation theory and good point set method, and the Puma Optimizar (PO) Algorithm is improved to solve the PLE in the target function. Then, the optimized PLE is substituted, the matrix multiplier is introduced, a differential generalized trust region subproblem (DGTRS) framework is established, and the fine-grained position is estimated. The flow chart of the positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty proposed by the disclosure is shown in Figure 1, and the method includes the following steps: SI, establishing a restricted motion model of offshore nodes, setting node positions and anchor node positions of target nodes, and establishing a ranging model of the surface wireless sensor networks; S2, performing a differential operation on the ranging model, establishing a natural constant least squares framework, and solving the natural constant least squares framework by a differential linear unbiased estimation method to obtain estimated values of coarse-grained target nodes; S3, constructing an optimization function with path loss factors as variables based on the estimated values of the coarse-grained target nodes in the S2, and improving the Puma Optimizar Algorithm by combining the quantum computation theory with a good point set method to solve an optimal path loss factor; and S4, based on the optimal path loss factor in the S3 and matrix multipliers, establishing a framework of differential generalized trust region subproblem; combining with Lagrange multipliers, obtaining fine-grained target node positions by a dichotomy method. The specific steps for SI to establish the model are as follows: Nodes deployed on the water surface are fixed to a certain area on the water surface by using anchor chains, and the nodes on the water surface will move with the movement of wind, waves and currents. Due to the fixation by anchor chains, the movement range on the water surface will be limited, as shown in Figure 2, which is usually a circle with radius L . If the length of the yy y^ __ I j anchor chain is L and the water depth is L, then £ V £ l Considering that there are anchor nodes and one target node to be located in the SWSNs network, the position of the i-th anchor node is ' L / 2J , where f represents transposition; the position of the target a- = [a- a I7 node to be located is L i’ 2j Assuming that the anchor node and the target node to be located exchange information through radio signals, the received signal strength (RSS) by the anchor node from the target node is expressed as: ||x-a II P: = P0-10alog10-----+ ^0 (1) p where n means that the z-th anchor node receives the transmission power from the target d P node; 0 is the reference distance, usually Im; 0 is the transmission power of the target node; a represents the path loss factor, and the experienced value is usually UH represents second-order norm; 7' represents Gaussian distribution noise with a mean value of 0 and a r cr2 variance ot 1 . The first stage estimation is performed in S2. First, the idea of GPS differential technology is introduced, the first anchor node is selected as the reference point, and the model in formula (1) is subjected to differential operation, and then: ^0 = -10a log10 -J----J + %, Ir "JI (2) where = (> = 2’L ’ ); = 71. It can be seen from formula (2) that the differential operation is capable of eliminating the TP uncertainty. Subsequently, the base number of logarithm of formula (2) is changed to obtain: —7¾ InlO = -aIn—+ — M InlO, 10 V1 d 10 1 P) where The natural constant e is introduced, and the formula (3) is shifted and squared to obtain: Q Ct — 7 k J (4) According to the natural constant Taylor series first-order expansion formula, linear expansion is performed on the formula (4): where (5) p = e a a Let , then formula (5) is transformed into: Pj,A « dj. 0 = 1 Y X yl y = IMP variable L *’ 2,XJ is defined, where z II II , and the original positioning problem is further transformed into NC-LSE framework: argmin||3^-'P||2, e (7) in the formula, 2n21 - 2puan 2a22 -2p2XaX2 p2, -1 INI2-^2.1 hl2 2a^ ^2z% ,47,, ^-^31 Aril 11 3 = M -c,\i ~^Pn 2a22 —2p3Xal2 p33 — 1 M M 2ci.:x — 2pNXaX2 p. \ T= IINN / NNI2 M [IN II2 -pn^ hl2. Through the differential linear LS estimation method of formula (9), the estimated value of initial position of the target node can be obtained: 0==(3'3) ' 3^, Although the influence of TP uncertainty is eliminated by differential operation, the uncertainty of PLE will still make the estimated value of formula (9) have a large error. Therefore, a linear unbiased estimation method is proposed to improve the accuracy of initial position estimation. y y y / J Let 1 , 2 and 3 be solved by formula (9) respectively to obtain three elements of , and . Using the Taylor series first-order expansion formula, linear expansion is performed on the when x is close to x : where - ; I is identity matrix. The revised coarse-grained estimated value Xc is expressed as: .Hr'nsri'ryjs (]1) In S3, the second stage estimation is carried out, and the corresponding estimated distances d — X 4 / ¢ / = llx — tt II 3 11 711 and 1 H c 'll are calculated by using the first-order coarse-grained estimated value and according to the Euclidean geometric distance formula. According to formula (6), with a as a variable, an optimization function is constructed: n a = arg min (p.^dy d] . Q By introducing the theory of good point set, * indicates 5 is unit cube in 5 z*= ( f f L r i g G -dimensional euclidean space, and if K then: PM (k) = • kj,• k},L J / ' • A j), 1 <k <Mj, (13) J / - ' (M) , where ' ” is the fractional part of ’ r = cos(2^ / p),\ <k <s} where P is the smallest prime number and satisfies 5<(p-3) / 2 and its deviation satisfies: C I F S I where £ represents any positive integer; v ’ 7 represents constant, which is only related to P (k} r and s. Then M ' ' is a good point set and r is a good point. After generating a uniform population according to formula (13) and Conditional formula (14), introducing quantum computation theory, an individual m in the population is represented by two superposition states of qubits: where represents a qubit with a qubit length of where |^) = G70|0) + 57i |1) and ^2 represent the probability amplitudes of the quantum state and satisfy the normalization ,.. |ztT0|2+IGTJ2 =1 condition Illi Qubits are represented in sine and cosine matrix based on probability amplitudes; |C / ) = [cos£z,sin6(f, (16) where cos = s’n = 571 Then the quantum state superimposed by individuals m in the population of formula (15) is expressed by probability amplitude: cos (6() cos (02) L cos (6() sin^) sin(^2) L sin(^z) (17) Q j Q |0 27TI where ” ’ 1 are randomly generated in 1 ’ J. Each individual occupies two positions in the search space, and each position represents an optimal solution, which is decomposed into: m.. =(cos(0),cos(6(hL ,cos(0)). ^=(sin(6(),sin(6Q,L ,sin(^ Therefore, the mapping relationship between quantum space and solution space can be constructed by linear change. When a qubit [cos sin 1 has the value domain of t and »»> tn ITTh T hl the value domain of the corresponding solution x and y is L J , then the transformation may be obtained: mx = — (cos 6 + l)(Ub -Lb) + Lb, 2 (20) m = —(sin 0 + l)(Ub - Lb) + Lb. 2 (21). th m Because the initial population ( x or has been obtained through the good point set in the early stage, it is equivalent to the collapse of the corresponding quantum state to a certain state after one observation. In order to restore the superposition state, assuming that the current state is sinusoidal position and the generated population is y , the corresponding inverse transformation may be made according to formula 21: 0* = arcsin t Ub-Lb (22) Calculating the corresponding position of another state according to the obtained rotation angle / / / = — (cost?* + l)(Ub - Lb) + Lb. 2 (23) Combining the good point set theory and quantum computation method, based on a set of qubits, more initial populations are obtained, which further improves the population diversity and the distribution quality in the search space and enhances the global search ability. After the high-quality population is generated, the search iteration is carried out, in which the first three iterations are inexperienced, and the exploration (search) and exploitation (hunting) steps are run at the same time, which simulates the problem that puma is unfamiliar with the living space and has unclear prey area in the early stage. After the fourth iteration, it is an experienced stage, and the corresponding mechanism is switched, and only exploration or exploitation operations are carried out each time according to the conditions, so as to simulate the hunting behavior of puma after it is familiar with the environment. The flow chart of the second stage PLE estimation path loss factor of the improved puma algorithm integrating quantum computation based on S3 is shown in Figure 3. The specific search iteration steps include: (1) Exploration steps Pumas randomly jump into the search space to find food, and the whole population is sorted in [°’1] by ascending order. In the exploration steps, if the number randomly generated among rand functiOn is greater than 0.5, the value is updated by formula (24), otherwise formula (25) is adopted. Z^^R^^Ub-Lb^ + Lb, (24) Z. (25) where n is the current population number; Dim represents a randomly generated number under the condition of uniform distribution of 2 • rand -1 ■ b’G ■, Ub anc[ Lb are Upper bound and lower bound respectively; represent the solutions of the whole population. Then, according to the condition of formula (26), the relevant solution is updated, and the corresponding fitness function value is calculated in combination with formula (12). When Conditional formula (29) is satisfied, the new solution is used to replace the old solution, that is, formula (30) is executed, otherwise, formulas (26) to (29) are continuously executed. new !rand or rand <U, otherwise (26) (27) p = NC / M (28) (29) n,new Y in where new represents the updated solution; rand represents a randomly generated integer; represents a preset parameter between 0 and 1; A / is the total number of pumas; is a fitness function, namely formula (12); n represents the solution corresponding to the current population. (2) Exploitation steps In order to further simulate the hunting behavior of puma running and ambush strategy, the optimal solution of the target is obtained. The conditional formula (31) is constructed, and when the generated random number is greater than 0.5, the running strategy simulated by the first line of formula (31) is adopted; when the generated random number is smaller or equal to 0.5, ambush strategy is adopted, in which the second line of formula in formula (31) is ambush strategy 1, which is used to simulate the short jump of puma to other pumas; the third line of formula (31) is ambush strategy 2, which is used to simulate the long jump process of puma to the optimal puma. ifra^>0.5,Xnew = Pumabest +(2- rand) • exp (randn) • X[ - Xn = <otherwise, if rand >L, Xnew , , (F-R-X(n) + R-(l-^j-Puma^) otherwise,X = (2 x rand) x ------------------------------ - Puma neB V 7 (2rand-\ + randn) (31) where mean0 represents average function; ^°^otai represents the sum of all solutions; is a randomly selected solution in the whole population; is a number randomly selected from 0 and 1; K and £ represent optimization parameters preset in advance; ^umabest is the best puma at present; randn represents a Gaussian distribution function with a mean value of 0 and c , A' • 1 <j- x round (\ + (M-1)-rand) rnn / 1 , R a variance of 1; 2 is solved according to v 7 7; K-^ tana 1- i p and 2 are solved according to formulas (32) and (33) respectively: I ( 2 randn-exp 2-Iter-\------- I \MaxIter F2=wx (v)2 • cos((2 x rand (32) (33) where ^ter represents the current iterations; Maxi ter represents the maximum iterations; and v are generated by Gaussian random distribution function randn respectively. (3) Inexperienced stage The inexperienced stage is usually the first three iterations, and the exploration and exploitation steps are run at the same time, so as to realize the goal of getting familiar with the environment and defining the prey range of puma. In this stage, the scores of corresponding exploration and fl f 2 exploitation steps are calculated by using Ju and Ju functions. As shown in formula (34) to formula (39): fl =PF- Ju 1Er 1 1 1 f lc. = PF ■ J u Ei 1 ^gfffitEr ^^fitEl s^qtime (34) (35) ( fuK=PF.- ^^fltEr + ^g^fitEr + ^^fitEr Seq}me+Seq^me+Seqlme { _ pp . + ^fftitEi ^g^fitEi Score... = {PF ){PF • / „2e ), ntr y i ** tsr f y + *' u rtr (36) (37) (38) ScoreEi =(PFpfulEi)+(PF2-fu2Ei), .tsi y i u tsi q y j, j u tsi t 2 yj \ fl fl where Ju Er and Ju E1 respectively represent the function of calculating exploration and fl f 2 f 2 exploitation by Ju in the inexperienced stage; Ju Er and Ju E1 respectively represent the f 2 PF function of calculating exploration and exploitation by Ju in the inexperienced stage; 1 PF fl / 2 and 2 are preset parameters in advance, which are used to prioritize J and J functions. ^eqtime is usually set to 1; 6egfitEr, Seq&Ei Seqfwp Seq^ and Seqt^ may be expressed as formulas (40) to (45) respectively. =|Mst -M|> ^^fitEr = |^^Er “ fFr | ’ ^^fitEr = |^^Er ~ fi^i | ’ SeqltE = I fiti. — fill-1, / litli |J Ei J Ei|’ (44) = |. / ¾ — . / ^ri | ’ 0 where is the fitness function corresponding to the initial solution; ^?Er and / / r-are the values corresponding to the fitness function in the first to third calculation of exploration operations, respectively; -^Ei, and are the values corresponding to the fitness function in the first to third calculation of exploitation operations, respectively. (4) Experienced stage After the first three iterations, the experienced stage is entered. This stage is different from the inexperienced stage, and it is not necessary to calculate the exploration and exploitation steps at the same time. It is only necessary to determine the stage to enter according to formula (38) and formula (39). If >ScoreEi , expioratjon stepS are entered; if Score< ScoreEi the f 2 f 3 exploitation steps are entered, and the three functions of -,e , "■ and -,e will be used for scoring selection respectively. Je is mainly used to calculate the priority level of exploration and exploitation steps, as shown in formula (46) and formula (47): rE1 'new ■r^Ei t (46) (47) f 1E1 f lEr where Je * and Je * are respectively the values corresponding to the exploitation and f 1 fit^ exploration steps in the ^-th iteration by using function in the experienced stage; ■ old and J old are respectively the fitness function values of exploitation and exploration steps fitEl calculated under the previous iteration of the experienced stage; J new and ■ new are respectively the fitness function values of exploitation and exploration steps calculated under the ry-’Er current iteration; >and ' are the unselected iterations in the exploitation and exploration steps, respectively. fl f 2 Compared with Je function, Je function emphasizes the resonance component more, that is, emphasizes the fitness function relationship with the solutions of the previous two iterations in the process, as shown in formulas (48) and (49): x = pk- ,)+(^9- 9) + (^,- / Z / Er J old,l J new,l J old,2 new,2 J old,3 -7 new,3 J rrfEl / TrEf / TrEr ' ^ / ,2 + ^ / ,3 (48) (49) / * 2E1 2Er where ■ ‘ and ' t are respectively the values corresponding to the exploitation and exploration steps in the / -th iteration by using Je function in the experienced stage; fit™ fit^ 0142 and old3 respectively represent fitness function values of unoptimized solutions corresponding to exploitation steps in t -th, -th and -th iterations in the fifEr / / / experienced stage; ' 01d>1, ' old’2 and ' are respectively fitness function values of the unoptimized solutions corresponding to exploration steps in the / -th, ' ’ -th and ' 7-th fit^1 fit^1 fit511 iterations in the experienced stage; "e"J, new>2 and ! respectively represent fitness function values of optimal solutions corresponding to the exploitation steps in the / -th. -th and fit Er fit Er fit Er -th iterations in the experienced stage respectively; ‘ new>1 , ‘ new’2 and ' new>3 are are fitness function values of the optimal solutions corresponding to the exploration steps in C-l) ( / -2) TE1 rE1 / -th, ' 7-th and ' 7-th iterations in the experienced stage respectively; w , / >2 and S rE1 ( / -1) M respectively represent iterations not selected by exploitation steps in ' 7-th to / -th, ( / -2) , ( / -1) , , ( / -3) , ( / -2) ... T* TP , 7)* . . ' 7 -th to v 7 -th and v 7 -th to 7 -th iterations; r i , and ' ’ respectively 1 • • , ( / -1) , , ( / -2) , represent iterations not selected by exploration steps in the v 7-th to / -th, v 7-th to ( / -1) . , ( / -3) , ( / -2) . . v 7 -th and v 7 -th to v 7 -th iterations. f 3 f 3 For the third function Je in the experienced stage, Je pays more attention to the diversity of choices, so that the unselected parts in the previous iterative stage have the choice opportunity again, so as to prevent the solution from falling into the local optimum, as shown in formulas (50) and (51): / 3Ei = / ifselected’   Za'=0’ [otherwise, / 3' + PF3, a _ if selected, fe3f = 0, [otherwise, fe3f+PF3, f 3E1 f 3Er where Je ! and Je * are the values corresponding to the exploitation and exploration steps in the ^-th iteration by using function in the experienced stage respectively; 3 is an advanced parameter, and the range is The closer this parameter is to 1, the lower the corresponding score and the greater the chance of being selected. Then, the total cost fraction of exploitation and exploration steps is calculated, as shown in formula (52) and formula (53): = + + -lc ■ , (52) FEr = 7rEV lEr + X-Er / 2Er + JEr -lc-f 3Er, t t J e t t J e t t Jet' ( S-4 ) pEi ^Er where ' and * respectively represent the cost fraction of corresponding exploitation and exploration under the t -th iteration; ' and ‘ respectively represent the parameters corresponding to exploitation and exploration in the / -th iteration (the range of values is rEr which change with the number of iterations. When ' and 1 are infinitely close to 1, the function corresponding to the resonance component is given priority and updated according to formula (54). For example, if the score of exploration step is greater than that of exploitation step, Ei the ' of exploitation step will be punished linearly with 0.01, and the same punished procedure would be carried out when the score of exploration step is lower than that of exploitation step. ' “ ' ; ' t - lc represents a set of difference values of fitness calculated from the exploitation and exploration steps with the optimal solution, as shown in formula (55). / cEr,E1 ifFEl >FEr,^E1 = 0.99,Fr =[a-''-0.01,0.01], otherwise, A;Er = 0.99,a:E1 = ]a;E1 -0.01,0.01], ic={i^oid - r, i Aid - r}, o ¢ / c. (54) (55) In S4, the third stage estimation is carried out, and after substituting the a calculated in the second stage into formula (2), obtaining: F = -10alogKi~+% 5 d> (56) The term of (56) is shifted, and Taylor series first-order expansion is performed on the exponential term: A 1010“ d- ® Subsequently, the differential LS framework is further constructed: 10 arg min 10 5“ d^ - d~ (58) Q = <||< introducing matrix multiplier: (59) where * is a matrix in which all elements in the matrix are -1. The DGTRS framework is obtained by squaring each item in formula (58): 15 arg min Q (HHEf2Mj-(HHEfM2, s.t. QrDQ+2BrQ=0, (60) where D l / 2’®2xiAx2’0] an(j 0 are matrices with all elements of 0; B [°21’ °'5l; 7% 2< -2-101¾ M 7¾ 'T' _ A * 'P 2«f-2-105“ aTN =(HHrf / 2^ ^=(HHT^ 0>=QrDQ+2BrQ f , , Let v 7 ; ' ' v m m • formula (60) may be solved by introducing Lagrange multiplier and constructing function formula (62): £(q;2) = (M) + where is Lagrange multiplier. In the function of formula (62) , Q is taken as a variable, and after solving the partial derivative, obtaining: J3''i 'llkVl?' ■0'3B 0Q ' (63) If formula (63) approaches 0, the Lagrange multiplier estimation equation can be obtained: Q (2) = ((^)^+20 j ((^)^+26^. where represents a function for . Since is unknown, taking as a variable, the equation in formula (64) is equivalent to 0, and then the is solved by formula (65): 2= / mm(®,0), where 7^(^0) represents the function expression when ® is 0, where 0 =Q (2)^ D Q (2)+2BrQ (2) However, the Lagrange multiplier obtained by formula (65) is not optimal. Therefore, defining the interval I : ' is a strictly decreasing function in the interval I , and the optimal Lagrangian multiplier may be found by dichotomy method. Then substituting , and finding the fine-grained position according to formula (67): (67) X where f is the fine-grained estimated position; 112 represents the first two terms of the corresponding vector. The following is the actual simulation experiment: In order to verify the effectiveness of the method Coarse-to-fine target localization(CFTL)) proposed in this paper, the experiment was carried out on the platform of Matlab R2022b, and the simulation set is: the water depth m and the anchor chain length A - 25 m, the movement range of the node on the water surface was within the circle with the radius of R = 20 m w L . Because the wind, waves and currents at every moment are unknown, the rana function is used to randomly generate the positions of anchor nodes and target nodes in the restricted area, and the root mean square error (RMSE) is used as the benchmark to evaluate the positioning accuracy: RMSE = V MC (68) where ^6' is the total number of Monte Carlo experiments, set to 1000; is the current — X number of times; x is a fine-grained position estimated value, and f. P = ^55 dBm _ 7 s In the simulation, the real 0 and a - 3 of the nodes are set, and then the noise is combined to simulate the RSS value received by the anchor node under real conditions. Then, according to the obtained RSS value, the third-order positioning is carried out. Although PLE is uncertain, its experienced range is usually ^A] Therefore, the value is taken according to the experienced range at each moment, and the coarse-grained estimation is obtained in the first-order calculation stage by this way, and then the second-order calculation step is entered based on this. In the experiment, the fixed parameters of the second-order PLE estimation are set as follows: P^=0'5, PF2=^, PF3 K=1, £=0.67, M = 30,and Maxlter = 100 In order to further explore the effectiveness of the method proposed in this paper when the number of anchor nodes changes, simulation experiments are carried out, a 3 dB js set, and a simple source localization method (ASSL), advanced best linear unbiased estimation positioning method (Ablue), ratio and search positioning method (RAS), active set method (ASM) and the coarse-to-fine target localization (CFTL) method proposed by the disclosure are compared respectively. The relative positioning error is shown in Figure 4. As the number of anchor nodes increases, the information available for location increases. Therefore, when N increases, the positioning error of each algorithm decreases gradually. As can be seen from Figure 4, the positioning accuracy of ASM and RAS is close when the number of anchor nodes is small, but with the increase of N, the positioning errors are different, among which RAS obtains the estimated PLE by linear search in the process, so the positioning error is smaller than ASM when N is large. Compared with other methods, Ablue has no additional means to estimate PLE, but still presents better positioning accuracy, which shows it has good robustness under two-parameter uncertainty. Due to the addition of PO and fine-grained estimation stage, the positioning accuracy of the method proposed in this disclosure has been improved, which enables to achieve better positioning effect when the number of anchor nodes changes and the two parameters are uncertain. The increase of environmental noise will often adversely affect the positioning accuracy. In order to further explore the positioning effect of the method proposed in this paper in the change of environmental noise, A = 8 is set. The relative positioning error is shown in Figure 5. As can be seen from Figure 5, although the positioning error of CFTL increases with the increase of noise, the rate of change of the error increase is decreasing. The corresponding situation also appeared in Ablue, and the positioning error increased slightly compared with CFTL. Compared with CFTL and Ablue, the positioning error of other methods is obviously increased under the ,. . „ , _ . ,, . . , CT2 = 1 dB 1 ■ • • ay ■ condition of noise change. Especially ASM, when ' , the positioning effect is even better than Ablue because of the set constraints, but with the increase of noise, its positioning accuracy has declined to a great extent. CFTL shows a good positioning effect when the noise increases, and the performance is higher than other methods. The disclosure is a third-order estimation method from coarse-grained to fine-grained. In the first stage, the TP uncertainty is eliminated by differential operation, and an NC-LSE framework is established to solve the coarse-grained position of the target node, thus realizing the positioning under the TP uncertainty. In the second stage, the target optimization function with PLE as a variable is established, and the quality of the initial population is improved by combining quantum computation theory and good point set method, and then the PO algorithm is improved to solve PLE, thus eliminating the uncertainty of PLE. In the third stage, the PLE is substituted, and a fine-grained estimation framework based on DGTRS is established to solve the position of the target node. Compared with the existing SWSNs positioning technology, the method provided by the disclosure eliminates the influence of two-parameter uncertainty by using a third-order estimation method from coarse-grained to fine-grained, and simultaneously realizes the accurate estimation of the target node position from coarse to fine, thus providing a feasible solution to the SWSNs positioning problem in a highly dynamic and uncertain ocean environment. The preferred embodiments of the present disclosure have been described in detail above. It should be understood that those skilled in the art can make many modifications and changes according to the concept of the present disclosure without creative work. Therefore, any technical scheme that can be obtained by a person skilled in the technical field through logical analysis, reasoning or limited experiments on the basis of the existing technology according to the concept of the present disclosure should be within the protection scope determined by the claims.

Claims

1. A positioning method of surface wireless sensor networks integrating quantum computation under condition of two-parameter uncertainty, comprising following steps:SI, establishing a restricted motion model of offshore nodes, setting node positions and anchor node positions of target nodes, and establishing a ranging model of the surface wireless sensor networks;S2, performing a differential operation on the ranging model, establishing a natural constant least squares framework, and solving the natural constant least squares framework by a differential linear unbiased estimation method to obtain estimated values of coarse-grained target nodes;S3, constructing an optimization function with path loss factors as variables based on the estimated values of the coarse-grained target nodes in the S2, and improving a Puma Optimizar Algorithm by combining the quantum computation theory with a good point set method to solve an optimal path loss factor; andS4, based on the optimal path loss factor in the S3 and matrix multipliers, establishing a framework of differential generalized trust region subproblem; combining with Lagrange multipliers, obtaining fine-grained target node positions by a dichotomy method.

2. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 1, wherein the node positions of the target nodes are x = and a position of an z-th anchor node isa = [c^a], and the ranging model of the surface wireless sensor networks is as follows:||x-a IIpn ^-IQoHogJ1 +pwherein n represents transmission powers from the target nodes received by the z-th anchor node; d0 is reference distances, Po is transmission powers of the target nodes; a represents the path loss factors, and represents Gaussian distribution noise with a mean value of 0 and a variance of cr2.

3. The positioning method of the surface wireless sensor networks integrating the quantum30computation under the condition of two-parameter uncertainty according to claim 1, wherein specific steps of performing the differential operation on the ranging model and establishing the natural constant least squares framework are as follows:performing the differential operation on the ranging model, changing a base number of an obtained differential result, introducing a natural constant e, performing shifting and squaring operations, and then performing linear expansion according to a natural constant Taylor series first-order expansion formula to define a variable = , wherein Z = ||x||2 ,constructing the natural constant least squares framework based on an expanded formula and the variable 0 = [v^x,,^]7;the differential result is:^=~wlog»|74+^wherein ( j = 2,L ,N ) ;‘Ji1 ‘J 11 ' J J 14. The positioning method of the surface wireless sensor networks integrating the quantumcomputation under the condition of two-parameter uncertainty according to claim 3, wherein thenatural constant least squares framework is:arg mm e2a2] ^2 / ?21an 2a:: - Pu 1 II": 3 = 2ii3i — 2 / ?32rzn ^■^31 ~'^p3ya\3 P3,\ — 1 ,T = II": M M M M ~^PN,ian 2^2 ~^Pt<1 Pn;i ~ 1 .11"» 1 awherein j = 2,L ,N represents a total number of anchor nodes.

5. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 4, wherein specific steps of solving the natural constant least squares framework by the differential linear unbiased estimation method to obtain the estimated values of the coarse-grained target nodes are as follows:solving the natural constant least squares framework to obtain estimated values of initial positions of the target nodes, and linearly expanding at x° close to x by using thedifferential linear unbiased estimation method, wherein 0 is the estimated values of the initialpositions of the target nodes, xf, x2 and x° are three elements of 0 respectively, and x° , after expansion, the estimated values of the coarse-grained target nodes areobtained, and the estimated values of the coarse-grained target nodes is:xc = A- ^(r3;'3r) r 37 3E.3 = 02z1,||x°||2^x° r = [ / 2,2(x°)rJ wherein I is identity matrix.

6. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 3, wherein the S3 comprises following specific steps:S31, constructing the optimization function with the path loss factors as variables based on the estimated values of the coarse-grained target nodes of the S2, and then initializing a population number, maximum iterations and initial iterations, and taking the initial iterations as current iterations, wherein population individuals are the path loss factors;S32, generating an initial population by combining the quantum computation theory with the good point set method, and calculating an initial fitness value;S33, judging whether the current iterations are less than or equal to 3, and if the current iterations are less than or equal to 3, executing S34;if the current iterations are more than 3, executing S35;S34, running exploration and exploitation steps, performing inexperienced calculation, calculating a score of the exploration and exploitation steps, updating the current iterations, and returning to the S3 3;S35, judging whether the current iterations are less than or equal to the maximum iterations, if the current iterations are less than or equal to the maximum iterations, performing experienced calculation, and running the exploration or exploitation steps according to the score of the exploration and exploitation steps of a previous iteration; if a new fitness value is less than a fitness value of the previous iteration, updating an optimal population, and then performing S36, if the new fitness value is not less than the fitness value of the previous iteration, directly performing S36; andS36, updating the score of the exploration or exploitation steps and the current iterations, and returning to S35 until the current iterations are greater than the maximum iterations, and outputting an optimal population corresponding solution at this time, wherein the optimal population corresponding solution is the optimal path loss factor.

7. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 6, wherein the exploration specifically comprises:randomly generating a value within a range of [0,1] based on rand function, wherein if the value is greater than 0.5, then according to:Z,hG=R^(Ub-Lb) + Lb,obtaining a first parameter Zn a \on a contrary, according to:7 -X +CrF + g((F ^F \ + (F ^F U ^n.G ^a.G^ ^^a.b T \\G^ ^c.d ) T \^c,d ^e.f ) j ’obtaining a first parameter Zn G \wherein n is a current population number; 7?Dim represents a randomly generated number in a case of uniform distribution of [0,1]; G rand \\ EaJ>= Xa_G-Xb>a', Ub and Lb are upper bound and lower bound, respectively; Ec d = XcG -XdG ; Ee f =XeG-XfG; XaGand Xf G represent solutions of a whole population;updating relevant solutions after obtaining the first parameter, wherein specific steps of updatingthe relevant solutions are as follows:Z^G, if m in. or rand <U , XaO, otherwiseNC=1-U p = NC / Mafter updating the relevant solutions, calculating fitness of updated solutions, and then executing:if <fit(Xn) is satisfied; executing Xa>G = Xnew, and then executing U=U + p',if fd(Xnew}< fit^X^ is not satisfied, directly executing U =U + p and returning to thespecific steps of updating the relevant solutions;wherein m represents individuals and Xnew represents the updated solutions; mnni represents randomly generated integers; U represents preset parameters between 0 and I; M is a total number of pumas; is a fitness function, and Xn represents solutions corresponding to the current population;the exploitation is specifically:randomly generating a value within the range of [0.1] based on rand function, andperforming updating according to the value, wherein an updated solution is:< otherwise, if rand >L, X\ :. = PumaK, +(2- rand) • exp [randn) • X\ - Xnotherwise^. = (2 x rand) x 2---------—----—----2 - Pumabe„(2 • rand -1 + randn)wherein rneanQ represents an average function; Soltotal represents a sum of all solutions;X{ is a randomly selected solution in the whole population; is a number randomly selectedfrom 0 and 1; K and L represent optimization parameters preset in advance; Pumabest is anoptimal puma at present; randn represents a Gaussian distribution function with a mean value of 0 and a variance of 1; Xr2 = round (\ + (Mrand} R = 2-rand-1;f f 2E = randn-exp 2-Iter A------- , 1 {Maxlter J)F2 = wx(v)2-cos^xrazz^-w),wherein Iter represents the current iterations; Maxlter represents the maximum iterations; w and v are generated by the Gaussian random distribution function randn respectively.

8. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 7, wherein the scores of the exploration and exploitation steps with the inexperienced calculation in the S34 are: Score,, = (^-. / .1,,) + (^- / ,2,,).Se-vre,. = (PF,-fX,) + (P / -f.2a),wherein,^^fitEi — |^E ^Ei|’=|M-A2i|>wherein fulEr and / ,1E1 respectively represent a function of using fu\ to calculateexploration and exploitation in an inexperienced stage; / u2Er and / u2Ei respectively representa function of using fu2 to calculate the exploration and exploitation in the inexperienced stage;PF} and PF2 are preset parameters in advance, used to sort priorities of / 1 and / 2functions; Seq^ is set to 1; ■ best is a fitness function value corresponding to an initial solution; fit^, fit^ and fit\x are values corresponding to the fitness function in a first to a third calculation of exploration operations; fit^ , and fit^ are corresponding values of the fitness function in a first to a third calculation of exploitation operations, respectively, and subscripts Er represents exploration and subscripts Ei represents exploitation;a score of experienced exploration and exploitation steps in S3 5 is:K = Kf tY + Kt t 2t + ot -lc- j 3, , t t J e t t J e t t Jet’jyEr Er r i Er , Er / ?^Er . cEr i e yErK —K. / L + r, / 2, +a 'lc-t3f,t I J e t t J e t t Jet’wherein,Er new+^r +cif selected, / ,3^=0, otherwise, f^f+PF^Er f if selected, / X=°>‘ otherwise, fe3Er+PF3,wherein / el^‘ and / ^r are respectively values corresponding to exploitation and exploration steps in a / -th iteration by using f\ function in an experienced stage; fit^ and are respectively fitness function values of exploitation and exploration steps calculated under a previous iteration of the experienced stage; fit^evi and Jit^ are fitness function values of exploitation and exploration steps calculated under a current iteration respectively; Tp and are iterations not selected in the exploitation and exploration steps respectively, and felf and fe 2^ are respectively the values corresponding to the exploitation and exploration steps in the t-th iteration by using fe2 function in the experienced stage; fit^2 and ^^3respectively represent fitness function values of unoptimized solutions corresponding to exploitation steps in / -th, ( / l)-th and ( / -2) -th iterations in the experienced stage; , and are respectively fitness function values of the unoptimized solutions corresponding to exploration steps in the / -th, ( / -1)-th and ( / -2) -th iterations in the experienced stage; , and fi^,3 respectively represent fitness function values of optimal solutions corresponding to the exploitation steps in the t -th, ( / -1) -th and ( / - 2) -th iterations in the experienced stage respectively; and are are fitnessfunction values of the optimal solutions corresponding to the exploration steps in / -th, ( / -1) -th and ( / - 2) -th iterations in the experienced stage respectively; , and S respectively represent iterations not selected by exploitation steps in ( / -1)-th to / -th, ( / -2) -th to ( / -1)-th and ( / -3)-th to ( / -2)-th iterations; 7,'( , TE2 and TEl respectively represent iterations not selected by exploration steps in the ( / -1)-th to / -th, ( / -2)-th to ( / -1) -th and ( / -3) -th to ( / -2) -th iterations, and fs3E' and fe3f are valuescorresponding to exploitation and exploration steps in the / -th iteration with fe3 function in the experienced stage; PF3 is an advanced parameter, FE1 and FEt respectively represent scores corresponding exploitation and exploration in the / -th iteration; and ref1 and kEt respectively represent parameters corresponding the exploitation and exploration in the / -th iteration.

9. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 8, wherein each individual of the initial population occupies two positions in a search space as follows:mx —(cosO\)P:hI.b)J.b,for a position my, a corresponding position of another superposition state is:6* = arcsin(2(my-Lb)Ub-Lbwherein a value domain of mx and my is [ / / / >,£ / >].

10. The positioning method of the surface wireless sensor networks integrating the quantum computation under the condition of two-parameter uncertainty according to claim 1, wherein the fine-grained target node positions are:X, -to)'          *+rB)Lwherein,■(-V-l)xW’l(W-l)xW ’~. d 10a log10+*wherein de is the optimal path loss factor.

Images