CN115242331B - Collaborative spectrum sensing optimization method for cognitive radio in virtual power plant - Google Patents

Abstract

The cooperative spectrum sensing optimization method of the cognitive radio in the virtual power plant comprises the following steps of establishing a cooperative spectrum sensing model of the cognitive radio in the virtual power plant; optimizing a cooperative spectrum sensing model of cognitive radio in the virtual power plant by utilizing an improved bate ray foraging optimization algorithm; determining a mathematical model update location; calculating an optimal value of the updated position fitness and a weight factor value corresponding to the optimal value; the local escape operator introduced into the GBO algorithm improves the batray foraging optimization algorithm to determine the mathematical model update position: when random parameter rand < pr=0.5 is set, a local escape operator in the GBO algorithm is adopted, and each agent position is updated according to the optimal value and the direction appointed by the gradient of the corresponding weight factor value; calculating an updated optimal value after updating the position and a corresponding position: judging whether an optimization result is reached or not according to the iteration times and the maximum iteration times. According to the method, the cognitive radio technology is introduced, so that the frequency spectrum utilization rate and the communication quality are improved, and the economic loss and the communication cost are reduced.

Description

Cooperative spectrum sensing optimization method for cognitive radio in virtual power plant

Technical Field

The present invention relates to the field of electric power communication technology, and in particular to a collaborative spectrum sensing optimization method of cognitive radio in a virtual power plant.

Background technique

Radio communication spectrum is particularly precious today when wireless communication technology is extremely developed. With the rapid development of wireless communication technology, the problem of poor spectrum resources is becoming increasingly serious. In the virtual power plant, a large amount of data is exchanged between distributed energy sources, electric vehicles, energy storage and other equipment, but the allocated power communication spectrum resources are limited.

The existing cooperative spectrum sensing method applied in virtual power plants is a commonly used and effective method to fully utilize spectrum resources. For example, Chinese patent CN103401625A provides a collaborative spectrum sensing optimization method based on particle swarm optimization algorithm, which can obtain better detection performance while reducing the time spent by the algorithm in spectrum sensing; Chinese patent CN103401626B proposes a collaborative spectrum sensing based on genetic algorithm The spectrum sensing optimization method uses probability transition rules to guide its search direction, so that compared with the existing technology, the possibility of searching for the optimal solution is increased; Chinese patent CN108900266B discloses a method based on collaborative node selection Cognitive Internet of Vehicles spectrum sensing method with FCM algorithm dynamically selects collaborative vehicles based on vehicle position and inter-vehicle correlation to ensure spectrum sensing accuracy and save costs; considering the movement of vehicle positions, no prior information such as signal-to-noise ratio is required, improving Detection performance. However, the above-mentioned existing research mostly uses older algorithms, which will fall into local optimal solutions and slow convergence speed for complex problems, and it is difficult to introduce cognitive radio into virtual power plants.

The Manta Ray Foraging Optimization Algorithm MRFO is inspired by the foraging behavior of manta rays. MRFO is formulated by imitating major foraging strategies such as chain, whirlwind, and somersault. However, in some optimization situations, the MRFO algorithm will fall into local solutions, especially in complex and high-dimensional problems. Because each solution updates the current position based on the previous solution, the convergence speed of the algorithm is reduced, and the search space solution is not effectively covered, resulting in premature convergence of the MRFO algorithm. The local escape operator LEO in the GBO algorithm is used to avoid falling into the local optimal solution and also solves the premature convergence. The LEO strategy updates the solution following a robust strategy and randomly selects solutions on the search space to update the solution.

Contents of the invention

The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art and provide a cooperative spectrum sensing optimization method for cognitive radio in a virtual power plant. Based on the cooperative spectrum sensing technology of the improved manta ray foraging algorithm MRFO-GBO, cognitive radio is introduced The virtual power plant communication system optimizes the weighted vector of the fusion center, thereby improving the efficiency and reliability of data communication in the virtual power plant.

The technical solution of the present invention is: a collaborative spectrum sensing optimization method for cognitive radio in a virtual power plant, which includes the following steps:

Step 1: Establish a collaborative spectrum sensing model for cognitive radio in virtual power plants.

Step 2: Use the improved manta ray foraging optimization algorithm to optimize the cooperative spectrum sensing model of cognitive radio in the virtual power plant.

Step 3: Update the position using the mathematical model of cyclone and chain foraging behavior.

Step 4: Calculate the optimal value of the fitness value of the updated position and its corresponding weight factor value: calculate the fitness value of each manta ray corresponding to the updated position, compare the fitness values of each two manta rays, whichever is smaller, and then The smaller one is compared with the fitness value of the next manta ray until the optimal value at this time is found, as well as the corresponding weight factor value.

Step 5: Introduce the local escape operator in the GBO algorithm to improve the manta ray foraging optimization algorithm and determine the update position of the mathematical model.

Step 6: Calculate the updated optimal value and the corresponding position after the updated position: Compare the updated position obtained in step 5 with the optimal value in step 4, and take the smaller one as the updated optimal value, and obtain the corresponding position.

Step 7: Determine whether the optimization result is achieved based on the number of iterations and the maximum number of iterations.

A further technical solution of the present invention is that step 1 specifically includes setting the energy equipment in the virtual power plant that sends local sensing statistical data to the fusion center as a cognitive user, and setting a total of M cognitive users.

Energy equipment with a fixed specific frequency band and cooperative sensing priority is set as the primary user, and energy equipment with a non-fixed specific frequency band or without cooperative sensing priority is set as the secondary user; the cooperative spectrum sensing framework of cognitive radio in the virtual power plant includes M Cognitive user and fusion center; then the binary hypothesis test expression of the spectrum sensing signal g _m (k) received by the m cognitive user at the kth moment is:

Among them, s(k) is the signal sent by the main user, and the energy of the signal is h _m is the channel gain between the main user and the mth cognitive user. It is affected by channel shadow, channel loss and fading, etc. Its vector expression is h=[|h ₁ | ² ,|h ₂ | ² ,...,|h _M | ² ] ^Τ ; a _m (k) is the additive Gaussian white noise when the cognitive user perceives the main user, with a mean of 0 and a variance of/> H ₀ indicates that there is no main user, and H ₁ indicates that there is a main user.

For each cognitive user, the summary of the energy of the sample detection received signal of the primary user N times is used as the presidential metric _um . The specific expression is as follows:

Therefore, the expression of the local signal-to-noise ratio R _m of the mth cognitive user is as follows:

In order to realize cooperation among multiple cognitive users, u _m is transmitted to the fusion center through a dedicated control channel. The expression of the output signal y _m perceived by the mth cognitive user obtained at the fusion center is as follows:

ym＝um+nm, m＝1,2,...,M (4)

Among them, n _m is the additive Gaussian white noise sent by the cognitive user to the fusion center, with a mean value of 0 and a variance of The variance is expressed as a vector

Thus, the expression of the global test statistic y _fc at this time is obtained as follows:

Among them, x is the weight factor assigned by the fusion center to each cognitive user statistics. The specific expression is x=[x ₁ ,x ₂ ,...,x _M ] ^T , and y=[y ₁ , y ₂ ,...,y _M ] ^T .

The calculation results of the variance of the global test statistic y _fc in the case of H ₀ are as follows:

in,

The variance calculation results of the global test statistic y _fc in the H ₁ case are as follows:

Among them, ζ = 2Ndiag ² (σ) + diag (δ) + 4E _s diag (h) diag (σ).

There is a threshold γ _fc , which is used to compare the global test statistic y _fc with the energy signal size received by the fusion center, and then determine whether there is a signal sent by the main user. If the energy signal is greater than the threshold, it is judged that the main user sends a signal, otherwise there is no signal.

Therefore, the expression of false alarm probability in cooperative spectrum sensing is as follows:

in,

The expression of detection probability in cooperative spectrum sensing is as follows:

From the expression of P _f in formula (8), the expression of threshold γ _fc is:

Substituting formula (10) into formula (9) we get:

It can be seen from formula (11) that when P _f is given, the maximum value of P _d can be obtained by optimizing x; therefore, the relevant objective functions and constraints of the cooperative spectrum sensing model of cognitive radio in the virtual power plant are as follows: :

Considering that the Q function in formula (11) is a monotonically decreasing function, maximizing P _d is equivalent to minimizing f(x) in formula (12) to find the optimal weight vector.

A further technical solution of the present invention is: the step 2 specifically includes, step 2-1, initializing the parameters of the improved manta ray foraging optimization algorithm: setting the maximum number of iterations to MaxIt; the current number of iterations t=1, randomly generating N Manta ray is used as the initial value. The initial value of manta ray n is x _n = [x _n1 , x _n2 ,..., x _nM ], n = 1, 2,..., N, and must satisfy the constraints.

Step 2-2: Calculate the optimal value of fitness and its corresponding weight factor value: compare the fitness values of each two manta rays to choose the smaller one, and then compare the smaller one with the next manta ray Compare the fitness values until the optimal value at this time is found, as well as the corresponding weight factor value.

A further technical solution of the present invention is: the step 3 specifically includes setting a random number rand with a value between 0 and 1; if rand < 0.5, update the position according to the mathematical model of whirlwind foraging behavior; otherwise, Update positions based on a mathematical model of chain foraging behavior.

The updating of the position according to the mathematical model of whirlwind foraging behavior includes:

Suppose coef=t/MaxIt. When coef<rand, select a randomly generated position in the search space as the reference position to explore the optimal solution; when coef>rand, select the weight factor value corresponding to the current best solution as the reference position to explore the optimal solution. The reference position of the optimal solution.

When coef<rand, the mathematical model of whirlwind foraging behavior is as follows:

where _xn,m (t) is the position of the mth user recognized by the nth manta ray in the tth iteration, is a random position randomly generated in the search space,/> r is a random vector in the range of (0,1); Ub ^m and Lb ^m are the upper and lower limits of the mth cognitive user, respectively. In the cooperative spectrum sensing model, Ub ^m = 1, Lb ^m = 0; β is a weight coefficient, /> Where _r1 is a random number in (0, 1).

When coef>rand, the mathematical model of whirlwind foraging behavior is as follows:

in, is the position of the high concentration of plankton in the t-th iteration, and the weight factor value BestX corresponding to the current best solution BestF is selected as the reference position for exploring the optimal solution.

The updated position according to the mathematical model of chain foraging behavior uses the following formula to update the model:

Among them, α is the weight coefficient,

A further technical solution of the present invention is: the step 5 specifically includes setting the probability parameter pr=0.5. If rand<pr, the local escape operator in the GBO algorithm is used to update the position; otherwise, the position is updated according to the mathematics of somersault foraging. Model update position.

The use of the local escape operator in the GBO algorithm to update the location includes:

Each agent position is updated according to the direction specified by the gradient of the optimal value and the corresponding weight factor value; in order to ensure the balance between exploration of important search space areas and close to the optimal point and the global point, the important parameter ρ ₁ is expressed as follows:

Among them, γ _min and γ _max are 0.2 and 1.2 respectively, t is the number of iterations, MaxIt is the maximum number of iterations, and the calculation formula of the gradient search rule is as follows:

Among them, Δx is the current optimal position x _best and the random position The step size between, parameter θ is used to ensure that Δx changes with the number of iterations, the calculation expression is as follows:

Among them, rand(1:N) is an N-dimensional random number, r1, r2, r3 and r4 (r1≠r2≠r3≠r4≠n) are different integers randomly selected from (1,N), and step is step size, with x _best and Related,/> is following/> Update the new vector generated, position/> Random points are created in the search space specified by the gradient search rule and the moving direction. The moving direction refers to the current vector x _n moving in the x _best -x _n direction. The expression of the moving direction DM is as follows:

DM＝rand·ρ ₁ (x _best -x _n ) (20)

So we get:

Among them, yp _n =y _n +Δx, yq _n =y _n -Δx; y _n is the average value of z _n+1 and x _n , and the expression of z _n+1 is as follows:

Among them, x _n is the current solution vector, randn is a random solution vector with dimension n, x _worst and x _best are the current worst and best solutions respectively. By Corresponds to the current vector in the formula/> Replace with the position of the best vector x _best , the new vector Expressed as follows:

The new position update formula in the next iteration is as follows:

Among them, r _a and r _b are random numbers between (0,1), The calculation formula is as follows:

The mathematical expression of LEO is:

Among them, f ₁ is a random number obeying a uniform distribution within the range of (-1, 1); f ₂ is a random number obeying a normal distribution, with a mean of 0 and a standard deviation of 1; u ₁ , u ₂ and u ₃ are three random number. In order to balance global exploration and local exploration, its parameters are set as follows: The expression is:/> And x _rand =Lb ^m +rand(0,1)·(Ub ^m -Lb ^m ),/> is the randomly selected overall solution (p∈[1,2,...,N]), and μ ₂ is a random number in the range of (0,1).

The updating position according to the somersault foraging mathematical model is performed by using the following formula to update the model:

Among them, S is the somersault coefficient that determines the manta ray's somersault range, S=2; r ₂ and r ₃ are two random numbers in (0,1).

A further technical solution of the present invention is: the step 7 specifically includes judging whether the number of iterations t reaches the maximum number of iterations MaxIt. If it reaches the maximum number of iterations, calculate and output the detection probability _Pd ; otherwise, the number of iterations t=t+1, repeat Steps 2 to 7 until the maximum number of iterations MaxIt is reached.

Compared with the prior art, the present invention has the following characteristics:

(1) The present invention introduces cognitive radio technology into the existing virtual power plant, which can effectively improve the communication quality of a large amount of data collected, monitored and controlled in the virtual power plant and its scheduling optimization management level while promoting the sustainable development of the power system and ensuring the safe and stable operation of the power system.

(2) By using the improved MRFO algorithm, the present invention reduces calculation costs, improves calculation accuracy, and reduces the probability of power system interruption, thereby reducing economic losses and reducing communication costs.

(3) The present invention optimizes the weighted vector of the fusion center by adopting an improved MRFO algorithm, thereby maximizing the detection probability, thereby reducing the false alarm probability, and improving the spectrum utilization as well as the reliability and efficiency of data communication in the virtual power plant.

The detailed structure of the present invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

Description of drawings

Figure 1 is a flow chart of the cooperative spectrum sensing optimization method of the present invention;

FIG2 is a schematic diagram of establishing a collaborative spectrum sensing model according to the present invention;

Figure 3 is a specific algorithm flow chart of the cooperative spectrum sensing optimization method of the present invention;

Figure 4 shows the detection probabilities of the method of the present invention and other algorithms at different iteration speeds;

Figure 5 shows the detection probabilities of the method of the present invention and other algorithms under different false alarm probabilities;

FIG6 shows the detection probability of the method of the present invention and the manta ray foraging optimization algorithm under different numbers of cognitive users;

Figure 7 shows the detection probabilities of the method of the present invention and the manta ray foraging optimization algorithm under different signal-to-noise ratios.

Detailed ways

Embodiment, as shown in Figure 1, a collaborative spectrum sensing optimization method for cognitive radio in a virtual power plant includes the following steps:

Step 1: Establish a collaborative spectrum sensing model for cognitive radio in the virtual power plant.

The energy equipment that sends local sensing statistical data to the fusion center in the virtual power plant is set as a cognitive user. The energy equipment includes photovoltaic power generation, electric vehicles, energy storage, wind power generation, etc. In this embodiment, a total of M cognitive users are set. Know the user.

Set energy devices with fixed specific frequency bands and cooperative sensing priorities as primary users, and energy devices with non-fixed specific frequency bands or without cooperative sensing priorities as secondary users. The basis of cognitive radio networks is spectrum sensing, and high-quality spectrum sensing capabilities are the prerequisite for spectrum access management and spectrum allocation. The cooperative spectrum sensing framework of cognitive radio in virtual power plant includes M cognitive users and fusion center. Then the binary hypothesis test expression of the spectrum sensing signal g _m (k) received by the m cognitive user at the k time is:

Among them, s(k) is the signal sent by the primary user, and the energy of the signal is h _m is the channel gain between the primary user and the mth cognitive user, which is affected by channel shadowing, channel loss and fading, and its vector expression is h = [|h ₁ | ² ,|h ₂ | ² ,...,|h _M | ² ] ^Τ ; a _m (k) is the additive Gaussian white noise of the cognitive user when perceiving the primary user, with a mean of 0 and a variance of/> H ₀ means that there is no primary user, and H ₁ means that there is a primary user.

For each cognitive user, the summary of the energy of the sample detection received signal of the primary user N times is used as the presidential metric _um . The specific expression is as follows:

Therefore, the expression of the local signal-to-noise ratio R _m of the mth cognitive user is as follows:

In order to realize cooperation among multiple cognitive users, u _m is transmitted to the fusion center through a dedicated control channel. The expression of the output signal y _m perceived by the mth cognitive user obtained at the fusion center is as follows:

y _m ＝ _um +n _m , m＝1,2,...,M (4)

Among them, n _m is the additive Gaussian white noise sent by the cognitive user to the fusion center, with a mean value of 0 and a variance of The variance is expressed as a vector

Thus, the expression of the global test statistic y _fc at this time is obtained as follows:

Among them, x is the weight factor assigned by the fusion center to each cognitive user statistics. The specific expression is x=[x ₁ ,x ₂ ,...,x _M ] ^T , and y=[y ₁ , y ₂ ,...,y _M ] ^T . The size of the weight vector reflects the contribution of a certain cognitive user's signal to the global decision-making.

The calculation results of the variance of the global test statistic y _fc in the case of H ₀ are as follows:

in,

The variance calculation result of the global test statistic y _fc in the H ₁ case is as follows:

Among them, ζ = 2Ndiag ² (σ) + diag (δ) + 4E _s diag (h) diag (σ).

There is a threshold γ _fc , which is used to compare the global test statistic y _fc with the energy signal size received by the fusion center, and then determine whether there is a signal sent by the main user. If the energy signal is greater than the threshold, it is judged that the main user sends a signal, otherwise there is no signal.

Therefore, the expression of false alarm probability in cooperative spectrum sensing is as follows:

in,

The expression of detection probability in cooperative spectrum sensing is as follows:

From the expression of P _f in formula (8), the expression of threshold γ _fc is:

Substituting formula (10) into formula (9) we get:

It can be seen from formula (11) that when P _f is given, the maximum value of P _d can be obtained by optimizing x. Therefore, the relevant objective functions and constraints of the cooperative spectrum sensing model of cognitive radio in virtual power plants are obtained as follows:

Considering that the Q function in formula (11) is a monotonically decreasing function, maximizing P _d is equivalent to minimizing f(x) in formula (12) to find the optimal weight vector.

Step 2: Use the improved manta ray foraging optimization algorithm (MRFO-GBO) to optimize the collaborative spectrum sensing model of cognitive radio in virtual power plants.

Step 2-1, initialize the parameters of the improved manta ray foraging optimization algorithm (MRFO-GBO): set the maximum number of iterations to MaxIt; the current iteration number t = 1, randomly generate N manta rays as the initial value, and record the initial value of manta ray n as _xn = [ _xn1 , _xn2 , ..., _xnM ], n = 1, 2, ..., N, and must satisfy the constraints

Step 2-2, calculate the optimal value of fitness BestF and its corresponding weight factor value BestX: Compare the fitness values of each two manta rays by comparing the smaller one, and then compare the smaller one with the next one The fitness values of the manta ray are compared until the optimal value BestF at this time is found, as well as the corresponding weight factor value BestX.

Step 3: Use the mathematical model of whirlwind and chain foraging behavior to update the position: there is a random number rand, whose value is between 0 and 1; if rand < 0.5, update the position according to the mathematical model of whirlwind foraging behavior; otherwise , update the position according to the mathematical model of chain foraging behavior.

The updating of the position according to the mathematical model of whirlwind foraging behavior includes:

Let coef=t/MaxIt. When coef＜rand, select a randomly generated position in the search space as the reference position to explore the optimal solution; when coef＞rand, select the weight factor value BestX corresponding to the current best solution BestF as Explore the reference position of the optimal solution;

When coef<rand, the mathematical model of whirlwind foraging behavior is as follows:

Among them, x _n,m (t) is the position of the n-th manta ray in the t-th iteration of the m-th cognitive user, is a random position randomly generated in the search space,/> r is a random vector in the range of (0,1); Ub ^m and Lb ^m are the upper and lower limits of the mth cognitive user respectively. Under the cooperative spectrum sensing model, Ub ^m =1, Lb ^m =0; β is Weight coefficient,/> where r ₁ is a random number in (0, 1).

When coef>rand, the mathematical model of whirlwind foraging behavior is as follows:

in, is the position of the high concentration of plankton in the t-th iteration, and the weight factor value BestX corresponding to the current best solution BestF is selected as the reference position for exploring the optimal solution.

The chain foraging behavior mathematical model is used to update the position using the following formula to update the model:

Among them, α is the weight coefficient,

Step 4: Calculate the optimal value newPopF of the updated position fitness and its corresponding weight factor value newPopP. Calculate the fitness value of each manta ray corresponding to the updated position, compare the fitness values of every two manta rays and take the smaller one, then compare the smaller one with the fitness value of the next manta ray, until the optimal value newPopF and the corresponding weight factor value newPopP are found.

Step 5: Introduce the local escaping operator (LEO) in the GBO algorithm to improve the manta ray foraging optimization algorithm MRFO to determine the mathematical model update position.

Set the probability parameter pr = 0.5. If rand < pr, the local escape operator (LEO) in the GBO algorithm is used to update the position; otherwise, the position is updated according to the somersault foraging mathematical model.

The updated location using the local escaping operator (Local Escaping, LEO) in the GBO algorithm includes:

Each agent position is updated according to the direction specified by the gradient of the optimal value newPopF and the corresponding weight factor value newPopP. In order to ensure the balance between the exploration of important search space areas and the closeness to the optimal point and the global point, the important parameter ρ ₁ is expressed as follows:

Among them, γ _min and γ _max are 0.2 and 1.2 respectively, t is the number of iterations, and MaxIt is the maximum number of iterations. ρ ₁ has a larger value in early iterations to enhance population diversity, and its value decreases in the later stages as the number of iterations increases to accelerate convergence.

By increasing the diversity of the population, the proposed algorithm can avoid falling into the local optimal solution. The calculation formula of GSR (Gradient-based optimizer) is as follows:

Provide more randomness to the GBO algorithm through the concept of GSR, thereby enhancing exploration behavior and avoiding local optima. In the above formula, Δx is the current optimal position x _best and the random position The step size between, parameter θ is used to ensure that Δx changes with the number of iterations, the specific expression is as follows:

Among them, rand(1:N) is an N-dimensional random number, r1, r2, r3 and r4 (r1≠r2≠r3≠r4≠n) are different integers randomly selected from (1,N), and step is step size, with x _best and Related. /> is following/> Update the new vector generated, position/> Created from random points specified by GSR and DM (Direction moving, moving direction) in the search space, where DM refers to the current vector (x _n ) moving in the (x _best -x _n ) direction. This process provides a suitable local search trend to improve the convergence speed of the algorithm. The expression of DM is as follows:

DM＝rand·ρ ₁ (x _best -x _n ) (20)

So we get:

Among them, yp _n =y _n +Δx, yq _n =y _n -Δx; y _n is the average value of z _n+1 and x _n , and the expression of z _n+1 is as follows:

Among them, x _n is the current solution vector, randn is a random solution vector with dimension n, x _worst and x _best are the current worst and best solutions respectively. by adding Corresponds to the current vector in the formula/> Replace with the position of the best vector x _best , the new vector Expressed as follows:

The GBO algorithm uses formula (21) to enhance the global search in the exploration phase, while formula (23) is used to enhance the local search capability in the development phase.

The new position update formula in the next iteration is as follows:

Among them, r _a and r _b are random numbers between (0,1), The calculation formula is as follows:

The introduction of LEO enhances the performance of optimization algorithms for solving complex problems. The LEO operator can effectively update the position of the solution, thereby helping the algorithm jump out of the local optimal point and speeding up the convergence of the optimization algorithm.

The mathematical expression of LEO is:

Among them, f ₁ is a random number obeying a uniform distribution in the range of (-1, 1); f ₂ is a random number obeying a normal distribution, with a mean of 0 and a standard deviation of 1. u ₁ , u ₂ and u ₃ are three random numbers. In order to balance global exploration and local exploration, their parameters are set as follows: The expression is:/> And x _rand =Lb ^m +rand(0,1)·(Ub ^m -Lb ^m ),/> is the randomly selected overall solution (p∈[1,2,...,N]), and μ ₂ is a random number in the range of (0,1).

The updated position according to the somersault foraging mathematical model uses the following formula to update the model:

Among them, S is the somersault coefficient that determines the manta ray's somersault range, S=2. r ₂ and r ₃ are two random numbers in (0, 1).

Step 6: Calculate the updated optimal value Best_PopF and the corresponding position Best_PopP after the updated position: Compare the updated position obtained in step 5 with the newPopF in step 4, whichever is smaller is Best_PopF, and obtain the corresponding position Best_PopP.

Step 7: Determine whether the number of iterations t reaches the maximum number of iterations MaxIt. If it does, calculate and output the detection probability P _d ; otherwise, the number of iterations t=t+1, repeat steps 2 to 7 until the maximum number of iterations MaxIt is reached.

In order to better describe the effect of the present invention, the collaborative spectrum sensing optimization method for cognitive radio in virtual power plants (MRFO-GBO) is compared with other meta-heuristic algorithms, including the manta ray foraging optimization algorithm ( MRFO), Gradient Optimization Algorithm (GBO) and Particle Swarm Optimization (PSO). The parameter settings of each algorithm during calculation are shown in Table 1 below.

Table 1 Parameter setting table for calculation of each algorithm

Figure 4 shows the detection probabilities of MRFO-GBO and other metaheuristic algorithms at different iteration speeds. The other parameters of MRFO-GBO are set as follows: the maximum value of the independent variable Ub ^m =1, and the minimum value of the independent variable Lb ^m = 0, T = 30, M = 6, N = 20, δ ₆ = σ ₆ = (1, 1, 1, 1, 1, 1), false alarm probability P _f = 0.1, different signal-to-noise ratio R ₆ =[-9.3,17.8,-4.6,-9.6,15.5,-4.2] ^T . As can be seen from Figure 4, MRFO-GBO can reach a detection probability of 0.75115 when the number of iterations is equal to 7, compared with MRFO reaching 0.74833 after 14 iterations, GBO reaching 0.74896 after 22 iterations, and PSO after 14 iterations. It reaches 0.71433 after times, which not only has the smallest number of iterations when reaching the optimal solution, but also has the highest detection probability and the fastest convergence speed.

Figure 5 shows the detection probabilities of MRFO-GBO and other meta-heuristic algorithms under different false alarm probabilities, where other parameters of MRFO-GBO are set to: M=6, N=10, P _f = (0.05, 1), signal-to-noise ratio R=[6,-5.9,12.1,5.5,14.2,-5.2] ^T . It can be seen from Figure 5 that MRFO-GBO has a significantly higher detection probability than other algorithms when the false alarm probabilities are equal.

In order to consider the impact of the cooperative spectrum sensing optimization method of cognitive radio in virtual power plants on the detection probability under different numbers of cognitive users and different signal-to-noise ratios. When the number of cognitive users is selected as M=6, 8, 10, and 12 respectively and other parameters are the same, the detection probabilities of MRFO-GBO and manta ray foraging optimization algorithm (MRFO) under different false alarm probabilities are calculated. The results As shown in Figure 6. It can be seen from Figure 6 that the detection probability increases with the increase of the M value. This is because as the number of co-sensing primary users increases, the optimal solution can be better searched among feasible solutions. Therefore, all algorithms The performance is better when the M value is higher; and when the M value is the same, the MRFO-GBO algorithm has a greater ability to find the maximum detection probability than the MRFO algorithm.

When the signal-to-noise ratios are selected as case1=8.33dB, case2=-0.2dB, and case3=-8.15dB, calculate the detection probabilities of MRFO-GBO and manta ray foraging optimization algorithm (MRFO) under different signal-to-noise ratios, and M =6, N=20, P _f =(0.05,1), the results are shown in Figure 7. It can be seen from Figure 7 that the greater the signal-to-noise ratio, the better the detection probability; and under the same signal-to-noise ratio, the MRFO-GBO algorithm can achieve a greater detection probability than the MRFO algorithm.

Claims (2)

1. A method for optimizing the collaborative spectrum sensing of cognitive radio in a virtual power plant, characterized in that it comprises the following steps: Step 1), establish a collaborative spectrum sensing model of cognitive radio in the virtual power plant: set the energy equipment that sends local sensing statistical data to the fusion center in the virtual power plant as a cognitive user, and set a total of M cognitive users; Energy devices with fixed specific frequency bands and cooperative sensing priority are set as primary users, and energy devices with non-fixed specific frequency bands or no cooperative sensing priority are set as secondary users. The cooperative spectrum sensing framework of cognitive radio in the virtual power plant includes M cognitive users and a fusion center. Then the binary hypothesis test expression of the spectrum sensing signal g _m (k) received by the mth cognitive user at the kth time is: Among them, s(k) is the signal sent by the main user, and the energy of the signal is h _m is the channel gain between the main user and the mth cognitive user. It is affected by channel shadow, channel loss and fading, etc. Its vector expression is h=[|h ₁ | ² ,|h ₂ | ² ,...,|h _M | ² ] ^Τ ; a _m (k) is the additive Gaussian white noise when the cognitive user perceives the main user, with a mean of 0 and a variance of/> H ₀ means there is no main user, H ₁ means there is a main user; For each cognitive user, the summary of the energy of the sample detection received signal of the primary user N times is used as the presidential metric _um . The specific expression is as follows: Therefore, the expression of the local signal-to-noise ratio R _m of the mth cognitive user is as follows: In order to realize cooperation among multiple cognitive users, u _m is transmitted to the fusion center through a dedicated control channel. The expression of the output signal y _m perceived by the mth cognitive user obtained at the fusion center is as follows: y _m ＝ _um +n _m ,m＝1,2,...,M (4) Among them, n _m is the additive Gaussian white noise sent by the cognitive user to the fusion center, with a mean value of 0 and a variance of The variance is expressed as a vector/> Thus, the expression of the global test statistic y _fc at this time is obtained as follows: Among them, x is the weight factor assigned by the fusion center to each cognitive user statistics. The specific expression is x=[x ₁ ,x ₂ ,...,x _M ] ^T , and y=[y ₁ ,y ₂ ,...,y _M ] ^T ; The calculation results of the variance of the global test statistic y _fc in the case of H ₀ are as follows: Where, ξ＝2Ndiag ² (σ)+diag(δ), The variance calculation results of the global test statistic y _fc in the H ₁ case are as follows: Among them, ζ=2Ndiag ² (σ)+diag(δ)+4E _s diag(h)diag(σ); There is a threshold γ _fc , which is used to compare the global test statistic y _fc with the energy signal received by the fusion center, and then determine whether there is a signal sent by the main user. If the energy signal is greater than the threshold, it is judged that the main user sends a signal, otherwise there is no signal; Therefore, the expression of false alarm probability in cooperative spectrum sensing is as follows: in, The expression of detection probability in cooperative spectrum sensing is as follows: The expression of the threshold γ _fc is obtained from the expression of P _f in formula (8): Substituting formula (10) into formula (9) we get: It can be seen from formula (11) that when P _f is given, the maximum value of P _d can be obtained by optimizing x; therefore, the relevant objective functions and constraints of the cooperative spectrum sensing model of cognitive radio in the virtual power plant are as follows : Considering that the Q function in formula (11) is a monotonically decreasing function, maximizing P _d is equivalent to minimizing f(x) in formula (12) to find the optimal weight vector; Step 2), use the improved manta ray foraging optimization algorithm to optimize the cooperative spectrum sensing model of cognitive radio in the virtual power plant: Initialize the parameters of the improved manta ray foraging optimization algorithm: set the maximum number of iterations to MaxIt; the current iteration number t=1, randomly generate N manta rays as initial values, record the initial value of manta ray n as x _n =[x _n1 , x _n2 ,...,x _nM ],n＝1,2,...,N, and must satisfy the constraints Calculate the optimal value of fitness and its corresponding weight factor value: compare the fitness values of each two manta rays to choose the smaller one, and then compare the smaller one with the fitness value of the next manta ray. Compare until you find the optimal value at this time and the corresponding weight factor value; The mathematical model of whirlwind and chain foraging behavior is used to update the position, which specifically includes setting a random number rand, with a value between 0 and 1; if rand < 0.5, the position is updated according to the mathematical model of whirlwind foraging behavior; otherwise, Update positions based on a mathematical model of chain foraging behavior; The updating of the position according to the mathematical model of whirlwind foraging behavior includes: Let coef = t/MaxIt. When coef < rand, select a randomly generated position in the search space as a reference position for exploring the optimal solution. When coef > rand, select the weight factor value corresponding to the current best solution as a reference position for exploring the optimal solution. When coef＜rand, the mathematical model of whirlwind foraging behavior is as follows: Among them, x _n,m (t) is the position of the n-th manta ray in the t-th iteration of the m-th cognitive user, is a random position randomly generated in the search space,/> r is a random vector in the range of (0,1); Ub ^m and Lb ^m are the upper and lower limits of the mth cognitive user respectively. Under the cooperative spectrum sensing model, Ub ^m =1, Lb ^m =0; β is Weight coefficient,/> where r ₁ is a random number in (0, 1); When coef>rand, the mathematical model of whirlwind foraging behavior is as follows: in, is the position of the high concentration of plankton in the t-th iteration, and the weight factor value corresponding to the current best solution is selected as the reference position for exploring the optimal solution; The chain foraging behavior mathematical model is used to update the position using the following formula to update the model: Among them, α is the weight coefficient, Calculate the optimal value of the fitness value of the updated position and its corresponding weight factor value: calculate the fitness value of each manta ray corresponding to the updated position, compare the fitness values of each two manta rays, choose the smaller one, and then add the smaller Compare it with the fitness value of the next manta ray until you find the optimal value at this time and the corresponding weight factor value; The local escape operator in the GBO algorithm is introduced to improve the manta ray foraging optimization algorithm to determine the mathematical model update position, which specifically includes setting the probability parameter pr=0.5. If rand<pr, the local escape operator in the GBO algorithm is used to update the position. ; Otherwise, update the position according to the somersault foraging mathematical model; The use of the local escape operator in the GBO algorithm to update the location includes: Each agent position is updated according to the direction specified by the gradient of the optimal value and the corresponding weight factor value; in order to ensure the balance between exploration of important search space areas and close to the optimal point and the global point, the important parameter ρ ₁ is expressed as follows: Among them, γ _min and γ _max are 0.2 and 1.2 respectively, t is the number of iterations, MaxIt is the maximum number of iterations, and the calculation formula of the gradient search rule is as follows: Among them, Δx is the current optimal position x _best and the random position The step size between them, parameter θ is used to ensure that Δx changes with the number of iterations, and the calculation expression is as follows: Among them, rand(1:N) is an N-dimensional random number, r1, r2, r3 and r4 (r1≠r2≠r3≠r4≠n) are different integers randomly selected from (1,N), and step is step size, with x _best and Related,/> is following/> Update the new vector generated, position/> Random points are created in the search space specified by the gradient search rule and the moving direction. The moving direction refers to the current vector x _n moving in the x _best -x _n direction. The expression of the moving direction DM is as follows: DM＝rand·ρ ₁ (x _best -x _n ) (20) So we get: Where, yp _n = _yn + Δx, yq _n = yn - _Δx ; _yn is the average of zn ₊₁ and _xn , and the expression of zn ₊₁ is as follows: Among them, x _n is the current solution vector, randn is a random solution vector with dimension n, x _worst and x _best are the current worst and best solutions respectively. By Corresponds to the current vector in the formula/> Replace with the position of the best vector x _best , the new vector /> Expressed as follows: The new position update formula in the next iteration is as follows: Among them, _ra and _rb are random numbers between (0,1), The calculation formula is as follows: The mathematical expression of LEO is: Among them, f ₁ is a random number obeying a uniform distribution within the range of (-1,1); f ₂ is a random number obeying a normal distribution, with a mean of 0 and a standard deviation of 1; u ₁ , u ₂ and u ₃ are three random number. In order to balance global exploration and local exploration, its parameters are set as follows: The expression is:/> And x _rand =Lb ^m +rand(0,1)·(Ub ^m -Lb ^m ),/> is the randomly selected overall solution (p∈[1,2,...,N]), μ ₂ is a random number in the range of (0,1); The updating position according to the somersault foraging mathematical model is performed by using the following formula to update the model: Among them, S is the somersault coefficient that determines the manta ray's somersault range, S=2; r ₂ and r ₃ are two random numbers in (0,1); Calculate the updated optimal value and the corresponding position after the updated position: compare the updated position of the mathematical model with the optimal value, whichever is smaller is the updated optimal value, and obtain the corresponding position; Whether the optimization result is achieved is determined based on the number of iterations t and the maximum number of iterations MaxIt. 2. The collaborative spectrum sensing optimization method of cognitive radio in a virtual power plant as described in claim 1 is characterized in that: the number of iterations t is judged whether it reaches the maximum number of iterations MaxIt, if it reaches it, the detection probability _Pd is calculated and output; otherwise, the number of iterations t = t+1, and step 2 is repeated until the maximum number of iterations MaxIt is reached.