Resource allocation strategy to ensure grid observability based on NOMA

Abstract

The present invention discloses a strategy for allocating communication resources for PMU in an uplink non-orthogonal multiple access mode under the premise of ensuring the observability of the smart grid system and satisfying the QoS communication delay. The communication resources in the present invention include bandwidth resources and power resources. The present invention adopts a non-orthogonal multiple access mode, and proposes an allocation algorithm for bandwidth resources and power resources in this mode based on a simulated annealing algorithm and a dichotomy method. The algorithm can ensure the observability performance of the power grid while improving the spectrum efficiency by about 20% compared to the OMA mode. At the same time, the effective capacity theory is used to ensure that the communication delay is less than the delay threshold from a statistical perspective, and the real-time performance of the data is required. Finally, the optimal allocation scheme for communication resources based on the uplink NOMA mode when the power grid system meets the delay requirements is given, and the spectrum efficiency is greatly improved while ensuring the observability performance and communication performance of the power grid.

Description

Technical Field

[0001] The present invention relates to the fields of mobile communication system technology and smart grid system, and in particular to a resource allocation strategy for ensuring grid observability in a NOMA manner. Background Art

[0002] As the concept of smart grid is proposed and regarded as the development direction of future power grid system, the network topology of the power grid has become more complex than before. Therefore, an effective energy management system is needed to guide the flow of energy. In addition, the volatility and variability of renewable energy, the uncertainty related to electric vehicles, the load changes over time caused by changes in electricity prices, etc. have brought new challenges to the power system. In order to achieve stable operation of the power system and realize real-time monitoring of the operating status of the power grid system, it is necessary to maintain the observability of the power grid. Such a goal can only be achieved with the help of PMU. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a resource allocation strategy based on the NOMA mode to ensure the observability of the power grid, so as to ensure the observability performance and communication performance of the power grid while greatly improving the spectrum efficiency.

[0004] To solve the above technical problems, the present invention combines the observability of power grid, effective capacity theory, non-orthogonal multiple access technology, binary search algorithm and simulated annealing algorithm. It includes the following steps:

[0005] Step 1: Combining NOMA technology and effective capacity theory, a system model based on uplink NOMA and observability of smart grid under communication constraints is obtained;

[0006] Step 2: Based on the simulated annealing algorithm and binary search, a resource allocation algorithm is proposed to solve the system model, and the user grouping results, power allocation scheme within the group, and communication bandwidth allocation scheme between groups of the NOMA system are obtained.

[0007] Preferably, step 1 comprises the following specific steps:

[0008] Step 1.1: Verify the grid observability using the network topology of the grid system and the PMU installation locations; use the effective capacity theory to incorporate communication delay into the grid observability system model and calculate the probability that the data packet generated by the PMU is transmitted to the control center within the set delay threshold;

[0009] Step 1.2: Obtain a system model for smart grid observability based on uplink NOMA and satisfying communication constraints.

[0010] Preferably, step 1.1 specifically includes:

[0011] Step 1.1.1: In a power grid containing N buses, the elements of the power grid connection matrix H are represented as follows: when i = j or bus i is connected to bus j, h i,j =1, otherwise, h i,j =0;

[0012] Step 1.1.2: Use a binary column vector x = {x1, x2, …, x N} T To indicate whether a node in the network is equipped with a PMU; the elements of the vector x are given by: when the bus node is equipped with a PMU, x i =1, otherwise, x i =0;

[0013] Step 1.1.3: If the voltage and current phasors of a bus are available simultaneously, the bus is called an observable bus; thus, the bus observability vector b is derived as:

[0014] b=Hx

[0015] where each element b in the bus observability vector b i represents the number of PMUs that can make bus i fully observable;

[0016] Step 1.1.4: Assume that there are K PMUs installed in the power grid, denoted as PMU k , and k=1,2,…,K and K≤N; ensure that PMU k With probability 0≤p k ≤1 satisfies D max The communication delay requirement is

[0017] Pr{d k ≤D max =p k

[0018] Step 1.1.5: Considering the communication delay constraint, the system observability model is as follows: Define a vector to represent D max The success probability of each PMU transmission is P = {P1, P2, ..., P N}, the elements of this vector are given by: If PMU k Installed on bus i, then P i =p k , otherwise, P i = 0; thus, the grid observability vector is obtained when the communication delay constraint is satisfied It is expressed as:

[0019]

[0020] where Λ Qrepresents the diagonal communication constraint matrix, defined as:

[0021] Λ Q =diag{Q1,Q2,…,Q N}

[0022] Q i ,i∈{1,2,…,N}, is a binary random variable, and its probability mass function is given by:

[0023] Q i =1}=P i

[0024] Q i =0}=1-P i

[0025] At the same time, the expected grid observability vector is obtained when the communication delay constraint is satisfied It is expressed as:

[0026]

[0027] where Λ P represents the diagonal probability matrix, defined as:

[0028] Λ P =diag{P1,P2,…,P N}

[0029] Step 1.1.6: The effective capacity formula is written as:

[0030]

[0031] Where C represents the throughput of the system in a single time block, T represents the duration of a time block, θ represents the QoS index, and when θ→0, the effective capacity is approximately the Shannon capacity;

[0032] Step 1.1.7: Assume PMU k Assigned B k The bandwidth, the corresponding signal-to-noise ratio is recorded as SNR k , by PMU k and the fading channel between the base station; therefore, for PMU k , its effective capacity is expressed as:

[0033]

[0034] in is the normalized QoS index;

[0035] Step 1.1.8: Based on the effective capacity theory and queuing theory, we get:

[0036]

[0037] Preferably, the system model in step 1.2 is expressed as:

[0038]

[0039] The specific form of the optimization objective function is one of the following three forms:

[0040] (1) When the focus is on improving the observability redundancy of the smart grid system while satisfying the communication constraints, the objective function is:

[0041]

[0042] (2) When the focus is on improving the observability sensitivity of the smart grid system under the condition of satisfying the communication constraints, the objective function is:

[0043]

[0044] (3) When the focus is on improving system performance by optimizing the probability that the observable random variable of each bus is higher than the expected threshold constant vector, the objective function is:

[0045]

[0046] Where λ is the expected observability, which is a constant vector of 1×N dimensions.

[0047] Preferably, step 2 specifically includes:

[0048] Step 2.1: Allocate communication resources to the PMU device based on the NOMA method and obtain the pairing result;

[0049] Step 2.2: Given the bandwidth allocation scheme and the power allocation scheme, solve the QoS index;

[0050] Step 2.3: Given the bandwidth allocation scheme, solve the power allocation scheme;

[0051] Step 2.4: Use the simulated annealing algorithm to solve the bandwidth resource allocation solution.

[0052] Preferably, step 2.1 specifically includes:

[0053] Step 2.1.1: Assume that a user group is composed of PMU a and PMU b The total transmit power in a group is represented by P, and a fixed power allocation scheme is adopted, namely PMU i The transmission power is α iP,i=a,b, where α is defined as the power allocation factor within the same user group, satisfying:

[0054] α a +α b =1

[0055] The superimposed signals of two PMU users received by the receiving end are expressed as:

[0056]

[0057] Where S a and S b is the transmission symbol, h a and h b are the channel coefficients of the two devices, N is Gaussian white noise with a mean of 0 and a variance of σ 2 ;

[0058] Step 2.1.2: The channel gain of the two devices is |h a | 2 and |h b | 2 , assuming that the order of the average channel gain is:

[0059]

[0060] After receiving the superimposed signal, the control end decodes the superimposed signal according to the order of the channel gain of the signal based on the serial interference cancellation theory: first decode the PMU b The signal is a The signal is treated as interference noise; then the superimposed signal is subtracted from the decoded PMU b After the signal is decoded, the PMU a According to the decoding theory, the signal-to-interference-to-noise ratio of the two PMU devices is obtained, which is expressed as follows:

[0061]

[0062]

[0063] in i=a, b is the instantaneous signal-to-noise ratio, which obeys exponential distribution and its mean can be expressed as:

[0064]

[0065] Step 2.1.3: Assume that a user group occupies bandwidth B ab , at this time, the effective capacity of the two PMU devices is expressed as:

[0066]

[0067]

[0068] Assuming that the channel environment is a Rayleigh fading channel, the closed-form expression of the effective capacity of each of the two PMU devices is expressed as:

[0069]

[0070]

[0071] where U(·,·,·) is the confluent hypergeometric function, that is:

[0072]

[0073] Where F(·,·;·;·) is a hypergeometric function, and its specific form is:

[0074]

[0075] Step 2.1.4: Use the unified channel difference pairing algorithm as the pairing algorithm for the PMU devices.

[0076] Preferably, step 2.1.4 specifically includes:

[0077] Consider a smart grid system with N buses and K PMU devices installed. k , the bus it is installed on is i k , sort the K PMUs in ascending order according to their average signal-to-noise ratio, and the jth largest one belongs to Then we have:

[0078]

[0079] Based on the channel gain, the PMU is divided into two sets, and the weak channel set is:

[0080]

[0081] The strong channel set is:

[0082]

[0083] Then, the unified channel difference pairing algorithm is used for pairing, that is, the user with a higher signal-to-noise ratio in one set will be paired with the user with a higher signal-to-noise ratio in another set, that is:

[0084]

[0085] When K is an even number, NOMA user pairing is as follows:

[0086]

[0087] When K is an odd number, the NOMA user pairing is as follows:

[0088]

[0089] Preferably, step 2.2 specifically includes: given a bandwidth allocation scheme, namely The power allocation scheme, that is, {α1,…,α K}, the QoS index of each PMU can be optimized separately without affecting each other; the optimal QoS index should make the following equation true:

[0090]

[0091]

[0092] Given and {α1,…,α K}, the above two equations can be directly solved by the dichotomy method to obtain the optimal QoS index vector, that is,

[0093] Preferably, step 2.3 specifically includes:

[0094] Step 2.3.1: Assume that the optimization is currently being performed on user group g; optimize the power coefficient of user group g. At this time, except for user group g, the power coefficients of other groups are considered as known parameters; use the golden section search algorithm to find the optimal power coefficient so that and Achieve a balance between the two, while maximizing the objective function value; get the optimal and Then, use step 2.2 to solve the optimal QoS index; refer to step 2.2 and use the dichotomy method to find the corresponding and

[0095] Step 2.3.2: Loop through the user group g, and execute step 2.3.1 each time to optimize and find the optimal power allocation coefficient of the two PMU devices in the user group; then calculate the value of the objective function based on the obtained optimal power allocation coefficient; until the objective function value converges, exit the loop optimization process and obtain all the final optimal power allocation coefficients and in

[0096] Preferably, step 2.4 specifically includes:

[0097] Step 2.4.1: The objective function is recorded as:

[0098]

[0099] Planning algorithm initialization, assuming the total bandwidth is B total , the initial bandwidth of K PMUs in the system is Then use step 2.3 to solve the power allocation scheme, use step 2.2 to solve the QoS index, and calculate the initial objective function value, denoted as F;

[0100] Step 2.4.2: Randomly generate a neighborhood value B′ of the current bandwidth vector B;

[0101] Step 2.4.3: Solve the optimal power allocation vector α′ according to steps 2.2-2.3;

[0102] Step 2.4.4: Solve the QoS index vector θ′ according to step 2.2;

[0103] Step 2.4.5: Calculation

[0104] Step 2.4.6: Let ΔE = F′-F. If ΔE < 0, B = B′, F = F′. Otherwise, with probability Accept the new solution, B = B′;

[0105] Step 2.4.7: Repeat steps 2.4.2 to 2.4.6 until M iterations have been completed, where M is the average Markov chain length.

[0106] Step 2.4.8: Update the temperature, set T = βT, where β is the temperature attenuation coefficient;

[0107] Step 2.4.9: Repeat steps 2.4.2 to 2.4.8 until the termination temperature is reached.

[0108] The beneficial effects of the present invention are as follows: the present invention first uses the effective capacity theory to incorporate the channel uncertainty of wireless communication into the model of power grid observability, and combines the dichotomy method and simulated annealing algorithm to propose a resource allocation algorithm based on the NOMA method to ensure the observability of the power grid. The algorithm can ensure the observability of the power grid and meet the communication delay performance while improving the spectrum efficiency. Numerical results show that the proposed algorithm can output a PMU device resource allocation scheme for any fixed topology power grid structure based on the uplink non-orthogonal multiple access method, and the resources include power resource allocation and bandwidth resource allocation. And compared with the existing OMA method, the spectrum efficiency of our solution is significantly improved. DETAILED DESCRIPTION

[0109] The invention will be verified using the IEEE14 bus power system test case which has been widely used as a standard test case.

[0110] In the simulation, Rayleigh fading is selected as the fading characteristic of the wireless channel, and the effective capacity under the OMA mode is given by the following formula.

[0111]

[0112] Assuming that the data center is located at the center of the power grid, and assuming that all PMUs generate measurement data packets at a rate of 60kbps, and that the maximum delay allowed for these data packets to be transmitted from the PMU to the control center is set to 10ms. The SNRs of the individual PMU transmissions are shown in the following table.

[0113] PMU 1 2 3 4 5 6 7 8 9 SNR(W) 13 32 31 12 10 25 29 30 11

[0114] The following planning method is used, including the following steps:

[0115] (1) First, consider the use of PMU to ensure the observability of the power grid: PMU can simultaneously measure voltage, current and corresponding phase angle data to achieve real-time synchronous measurement of the operating status of the wide-area power grid. PMU can not only measure the status information of the currently installed bus, but also obtain relevant data of the adjacent bus through measurement. Therefore, combined with the network topology of the power grid system and the installation location of the PMU in the system, it is possible to verify whether the power grid is observable. Note that communication constraints are not considered at this time.

[0116] (2) Secondly, consider using the effective capacity theory to describe the time-varying delay of communication between PMU and data center from a statistical perspective. The related applications based on PMU devices have strict requirements on communication delay. Combining the effective capacity theory can incorporate communication delay into the power grid observability system model. Through the effective capacity theory, the probability of the data packet generated by the PMU being transmitted to the control center within the set delay threshold can be calculated.

[0117] (3) Again, consider allocating communication resources to PMU devices based on the NOMA method: Although OMA technology has many advantages such as no interference between users and low complexity, its spectrum efficiency and system capacity are low. In contrast, non-orthogonal multiple access (NOMA) can greatly improve spectrum efficiency by allowing users to share resources (in the frequency domain or time domain). In order to ensure that PMU devices meet the quality of service requirements and save communication bandwidth, a system model of smart grid observability based on uplink NOMA and satisfying communication constraints is obtained by combining non-orthogonal multiple access (NOMA) technology and effective capacity theory. This model is designed to improve the observability of the power system by searching for the optimal NOMA grouping, optimizing power resource allocation, and optimizing bandwidth resource allocation under a given power grid topology.

[0118] (4) Based on the interior point method, simulated annealing algorithm and binary search, a resource allocation algorithm is proposed to generate the bandwidth and power allocation scheme of the PMU. Through steps (1), (2) and (3), an observable system model is established under the NOMA mode when the communication delay constraint is met. The system model is solved by the pairing algorithm proposed in this step to obtain the user grouping results, power allocation scheme within the group and communication bandwidth allocation scheme between groups of the NOMA system. Compared with the OMA mode, the spectrum efficiency can be improved by about 20% when the observable probability performance of 99.9% is met.

[0119] Verifying the observability of the power grid by using the network topology structure of the power grid system and the PMU installation location in step (1) includes the following steps:

[0120] (11) In a power grid containing N buses, the elements of the power grid connection matrix H can be expressed as follows. When i = j or bus i is connected to bus j, h i,j = 1. In other cases, h i,j =0.

[0121] (12) Using a binary column vector x = {x1, x2, …, x N} T To indicate whether a node in the network is equipped with a PMU. The elements of vector x are given as follows. When a bus node is equipped with a PMU, x i = 1. Otherwise, x i =0.

[0122] (13) If the voltage and current phasors of a bus can be obtained simultaneously, the bus is called an observable bus. Therefore, the bus observability vector b can be derived as:

[0123] b=Hx

[0124] where each element b in the bus observability vector b i represents the number of PMUs that can make bus i fully observable.

[0125] In step (2), using the effective capacity theory to incorporate communication delay into the power grid observability system model and calculating the probability that the data packet generated by the PMU is transmitted to the control center within the set delay threshold specifically includes the following steps:

[0126] (21) First, we introduce the communication delay. Assume that there are K PMUs installed in the power grid, denoted as PMU k , and k=1,2,…,K and K≤N. Ideally, if cost and other factors are not considered, the communication system can theoretically ensure that PMU k The delay requirement is denoted as d k , less than Dmax However, in practice, it has been shown that it is not feasible to provide a deterministic delay bound for communication systems in most fading channel environments. Therefore, it is considered to ensure that the PMU k With probability 0≤p k ≤1 satisfies D max The communication delay requirement is

[0127] Pr{d k ≤D max =p k

[0128] (22) Then, the observability modeling of the system considering the communication delay constraint is as follows: Define a vector, representing D max The success probability of each PMU transmission is P = {P1, P2, ..., P N}, the elements of this vector are given by: k Installed on bus i, then P i =p k Otherwise, P i = 0. Thus, the grid observability vector can be obtained when the communication delay constraint is satisfied. It is expressed as:

[0129]

[0130] where Λ Q represents the diagonal communication constraint matrix, defined as:

[0131] Λ Q =diag{Q1,Q2,…,Q N}

[0132] Here is the Q i ,i∈{1,2,…,N}, is a binary random variable, and its probability mass function is given by the following formula.

[0133] Q i =1}=P i

[0134] Q i =0}=1-P i

[0135] Assume that the random variable Q i ,i∈{1,2,…,N} are independent of each other. Note that the observability vector under the statistical delay guarantee is a random vector.

[0136] At the same time, the expected grid observability vector can be obtained when the communication delay constraint is satisfied: It is expressed as:

[0137]

[0138] where Λ P represents the diagonal probability matrix, defined as:

[0139] Λ P =diag{P1,P2,…,P N}

[0140] (23) Then the effective capacity theory is introduced, which describes the communication delay from a statistical perspective. The effective capacity formula can be written as:

[0141]

[0142] Where C represents the throughput of the system in a single time block, T represents the duration of a time block, θ represents the QoS index, and when θ→0, the effective capacity is approximately the Shannon capacity.

[0143] (24) Assume that PMU k Assigned B k The corresponding signal-to-noise ratio is recorded as SNR k , by PMU k The fading channel between the PMU and the base station determines the k , its effective capacity can be expressed as:

[0144]

[0145] in is the normalized QoS index.

[0146] (25) Based on the effective capacity theory and queuing theory, we can see that:

[0147]

[0148] (26) Assume that PMU k by Obviously, in order to transmit data stably, the effective capacity cannot be less than the sending rate. Right now

[0149]

[0150] In step (3), allocating communication resources to the PMU device based on the NOMA method includes the following steps:

[0151] (31) First, NOMA is used to allocate resources to PMU. When NOMA is used as the multiple access method, in order to reduce the complexity of analysis, it is assumed that each user group contains only two PMU devices, that is, the two PMU devices in the same user group will share the same time-frequency resources, and the two will be distinguished and decoded at the receiving end based on the difference in transmission power. Assume that a user group is composed of PMU a and PMU b The total transmit power in a group is represented by P, and a fixed power allocation scheme is adopted, namely PMU i The transmission power is α i P,i=a,b. Where α is defined as the power allocation factor within the same user group, satisfying:

[0152] α a +α b =1

[0153] The superimposed signals of two PMU users received by the receiving end can be expressed as:

[0154]

[0155] Where S a and S b is the transmission symbol, h a and h b are the channel coefficients of the two devices, N is Gaussian white noise with a mean of 0 and a variance of σ 2 .

[0156] (32) The channel gain of the two devices is |h a | 2 and |h b | 2 , assuming that the order of the average channel gain is:

[0157]

[0158] After the control end receives the superimposed signal, it decodes the superimposed signal according to the order of the signal's channel gain based on the serial interference cancellation theory. b The signal is a The signal is treated as interference noise. Then the superimposed signal is subtracted from the decoded PMU b After the signal is decoded, the PMU a According to the decoding theory, the signal-to-interference-to-noise ratio of the two PMU devices can be obtained, which is expressed as follows:

[0159]

[0160]

[0161] in i=a, b is the instantaneous signal-to-noise ratio, which obeys exponential distribution and its mean can be expressed as:

[0162]

[0163] (33) Then, the effective capacity theory is combined with the uplink NOMA system. Assume that a user group occupies a bandwidth of B ab , at this time, the effective capacity of the two PMU devices can be expressed as:

[0164]

[0165]

[0166] Assuming that the channel environment is a Rayleigh fading channel, the closed-form expression of the effective capacity of each of the two PMU devices can be expressed as:

[0167]

[0168]

[0169] where U(·,·,·) is the confluent hypergeometric function, that is:

[0170]

[0171] Where F(·,·;·;·) is a hypergeometric function, and its specific form is:

[0172]

[0173] (34) Considering the user pairing algorithm under NOMA mode, the present invention uses the unified channel difference pairing algorithm as the pairing algorithm of PMU devices. Consider a smart grid system with N buses and K PMU devices installed. For PMU k , the bus it is installed on is i k Sort the K PMUs in ascending order according to their average signal-to-noise ratio, and the jth largest one belongs to Then we have:

[0174]

[0175] Based on the channel gain, the PMU is divided into two sets, and the weak channel set is:

[0176]

[0177] The strong channel set is:

[0178]

[0179] Then the unified channel difference pairing algorithm can be used for pairing, that is, users with higher signal-to-noise ratio in one set will be paired with users with higher signal-to-noise ratio in another set, that is:

[0180]

[0181] Specifically, when K is an even number, the NOMA user pairing is as follows:

[0182]

[0183] When K is an odd number, the NOMA user pairing is as follows:

[0184]

[0185] In step (4), a resource allocation algorithm is proposed based on the simulated annealing algorithm and the bisection method, and the bandwidth and power allocation scheme of the PMU is generated, which includes the following steps:

[0186] (41) The observability system model of the power grid system when the communication constraints are met in the uplink NOMA system is:

[0187]

[0188] The specific form of the optimization objective function can be expressed in the following three forms according to different focuses:

[0189] (42) When the focus is on improving the observability redundancy of the smart grid system while satisfying the communication constraints, the objective function is:

[0190]

[0191] (43) When the focus is on improving the observability sensitivity of the smart grid system while satisfying the communication constraints, the objective function is:

[0192]

[0193] (44) When the focus is on improving system performance by optimizing the probability that the observable random variable of each bus is higher than the expected threshold constant vector, the objective function is:

[0194]

[0195] Where λ is the expected observability, which is a constant vector of 1×N dimensions.

[0196] (45) To solve the objective function, the pairing result is first obtained. According to step (34), the user pairing result of K PMUs in the power grid system under the NOMA system can be obtained, which is expressed as

[0197] (46) Given the bandwidth allocation scheme and power allocation scheme, solve the QoS index. Given the bandwidth allocation scheme, that is, The power allocation scheme, that is, {α1,…,α K}, the QoS index of each PMU can be optimized separately without affecting each other. In order to maximize observability, we hope that the probability p k The larger the value, the greater the probability p. k Only with θ k Monotonically increasing, and the effective capacity increases with θ k Monotonically decreasing, according to step (26), we know that the effective capacity must be at least greater than the data transmission rate. Then the optimal QoS index should make the following equation hold:

[0198]

[0199]

[0200] Given and {α1,…,α K}, the above two equations can be directly solved by the dichotomy method to obtain the optimal QoS index vector, that is,

[0201] (47) Given a bandwidth allocation scheme, solve the power allocation scheme. Fixed bandwidth allocation scheme, that is, Then, for each user group g, the objective function is the power allocation coefficient The present invention adopts an iterative optimization method to obtain the optimal value until convergence. At this time, the objective function is recorded as:

[0202]

[0203] (48) Assume that the optimization is currently being performed on user group g. Optimize the power coefficient of user group g. At this time, except for user group g, the power coefficients of other groups are considered as known parameters. Use the golden section search algorithm to find the optimal power coefficient so that and A balance is achieved between the two, while maximizing the objective function value. and After that, the optimal QoS index can be solved using step (46). Refer to step (46) and use the binary search method to find the corresponding and

[0204] (49) Loop through the user group g, and execute step (48) each time to optimize and find the optimal power allocation coefficient of the two PMU devices in the user group. Then calculate the value of the objective function based on the obtained optimal power allocation coefficient. After the objective function value converges, exit the loop optimization process and obtain all the optimal power allocation coefficients. and in

[0205] (410) Use simulated annealing algorithm to solve the bandwidth resource allocation scheme. At this time, the objective function is recorded as:

[0206]

[0207] Planning algorithm initialization, assuming the total bandwidth is B total , the initial bandwidth of K PMUs in the system is Then use steps (47)-(49) to solve the power allocation scheme, use step (46) to solve the QoS index, and calculate the initial objective function value, denoted as F.

[0208] (411) Randomly generate a neighborhood value B′ of the current bandwidth vector B.

[0209] (412) Solve the optimal power allocation vector α′ according to steps (47)-(49).

[0210] (413) Solve the QoS index vector θ′ according to step (46).

[0211] (414) Calculation

[0212] (415) Let ΔE = F′-F. If ΔE < 0, B = B′, F = F′. Otherwise, with probability Accept the new solution, B=B′.

[0213] (416) Repeat steps (411) to (415) until M iterations are reached, where M is the average Markov chain length.

[0214] (417) Update the temperature, setting T = βT, where β is the temperature attenuation coefficient.

[0215] (418) Repeat (411) to (417) until the termination temperature is reached.

[0216] Although the present invention has been illustrated and described with respect to the preferred embodiments, it will be appreciated by those skilled in the art that various changes and modifications may be made to the present invention without departing from the scope of the present invention as defined by the claims.

Claims

1. A resource allocation strategy based on NOMA to ensure grid observability, characterized by: The steps include: Step 1: Combining NOMA technology and effective capacity theory, a system model based on uplink NOMA and observability of smart grid under communication constraints is obtained; Step 2: Based on the simulated annealing algorithm and binary search, a resource allocation algorithm is proposed to solve the system model, and the user grouping results, power allocation scheme within the group, and communication bandwidth allocation scheme between groups of the NOMA system are obtained; Step 1 Specific steps: Step 1.1: Verify the grid observability using the network topology of the grid system and the PMU installation locations; use the effective capacity theory to incorporate communication delay into the grid observability system model and calculate the probability that the data packet generated by the PMU is transmitted to the control center within the set delay threshold; Step 1.2: Obtain a system model for smart grid observability based on uplink NOMA and satisfying communication constraints; Step 1.1 specifically includes: Step 1.1.1: In a power grid containing N buses, the elements of the power grid connection matrix H are represented as follows: when i = j or bus i is connected to bus j, h i，j =1, otherwise, h i，j =0; Step 1.1.2: Use a two-element column vector x = {x1, x2, ..., x N } T To indicate whether a node in the network is equipped with a PMU; the elements of the vector x are given by: when the bus node is equipped with a PMU, x i =1, otherwise, x i =0; Step 1.1.3: If the voltage and current phasors of a bus are available simultaneously, the bus is called an observable bus; thus, the bus observability vector b is derived as: b=Hx where each element b in the bus observability vector b i represents the number of PMUs that can make bus i fully observable; Step 1.1.4: Assume that there are K PMUs installed in the power grid, denoted as PMU k , and k=1,2,...,K and K≤N; ensure that PMU k With probability 0≤p k ≤1 satisfies D max The communication delay requirement is Pr{d k ≤D max }=p k Step 1.1.5: Considering the communication delay constraint, the system observability model is as follows: Define a vector to represent D max The success probability of each PMU transmission within, P = {P1, P2, ..., P N }, the elements of this vector are given by: If PMU k Installed on bus i, then P i =p k , otherwise, P i = 0; thus, the grid observability vector is obtained when the communication delay constraint is satisfied It is expressed as: where Λ Q represents the diagonal communication constraint matrix, defined as: Λ Q diag{Q1,Q2,...,Q N } Q i , i∈{1, 2, ..., N}, is a binary random variable whose probability mass function is given by: Pr{Q i =1}=P i Pr{Q i =0}=1-P i At the same time, the expected grid observability vector is obtained when the communication delay constraint is satisfied It is expressed as: where Λ P represents the diagonal probability matrix, defined as: Λ P ,diag{P1,P2,...,P N }} Step 1.1.6: The effective capacity formula is written as: Where C represents the throughput of the system in a single time block, T represents the duration of a time block, θ represents the QoS index, and when θ→0, the effective capacity is approximately the Shannon capacity; Step 1.1.7: Assume PMU k Assigned B k The bandwidth, the corresponding signal-to-noise ratio is recorded as SNRk, by PMU k and the fading channel between the base station; therefore, for PMU k , its effective capacity is expressed as: in is the normalized QoS index; Step 1.1.8: Based on the effective capacity theory and queuing theory, we get: The system model described in step 1.2 is expressed as: The specific form of the optimization objective function is one of the following three forms: (1) When the focus is on improving the observability redundancy of the smart grid system while satisfying the communication constraints, the objective function is: (2) When the focus is on improving the observability sensitivity of the smart grid system under the condition of satisfying the communication constraints, the objective function is: (3) When the focus is on improving system performance by optimizing the probability that the observable random variable of each bus is higher than the expected threshold constant vector, the objective function is: Where λ is the expected observability, which is a constant vector of 1×N dimensions.

2. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 1 is characterized in that: Step 2 specifically includes: Step 2.1: Allocate communication resources to the PMU device based on the NOMA method and obtain the pairing result; Step 2.2: Given the bandwidth allocation scheme and the power allocation scheme, solve the QoS index; Step 2.3: Given the bandwidth allocation scheme, solve the power allocation scheme; Step 2.4: Use the simulated annealing algorithm to solve the bandwidth resource allocation solution.

3. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 2 is characterized in that: Step 2.1 specifically includes: Step 2.1.1: Assume that a user group is composed of PMU a and PMU b The total transmit power in a group is represented by P, and a fixed power allocation scheme is adopted, namely PMU i The transmission power is α i P,i = a, b, where α is defined as the power allocation factor within the same user group, satisfying: α a +α b ＝1 The superimposed signals of two PMU users received by the receiving end are expressed as: Where S a and S b is the transmission symbol, h a and h b are the channel coefficients of the two devices, N is Gaussian white noise with a mean of 0 and a variance of σ 2 ; Step 2.1.2: The channel gain of the two devices is |h a | 2 and |h b | 2 , assuming that the order of the average channel gain is: After receiving the superimposed signal, the control end decodes the superimposed signal according to the order of the channel gain of the signal based on the serial interference cancellation theory: first decode the PMU b The signal is a The signal is treated as interference noise; then the superimposed signal is subtracted from the decoded PMU b After the signal is decoded, the PMU a According to the decoding theory, the signal-to-interference-to-noise ratio of the two PMU devices is obtained, which is expressed as follows: in is the instantaneous signal-to-noise ratio, which obeys exponential distribution and its mean can be expressed as: Step 2.1.3: Assume that a user group occupies bandwidth B ab , at this time, the effective capacity of the two PMU devices is expressed as: Assuming that the channel environment is a Rayleigh fading channel, the closed-form expression of the effective capacity of each of the two PMU devices is expressed as: where U(·,·,·) is the confluent hypergeometric function, that is: Where F(·,·;·;·) is a hypergeometric function, and its specific form is: Step 2.1.4: Use the unified channel difference pairing algorithm as the pairing algorithm for the PMU devices.

4. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 2 is characterized in that: Step 2.1.4 specifically includes: Consider a smart grid system with N buses and K PMU devices installed. k , the bus it is installed on is i k , sort the K PMUs in ascending order according to their average signal-to-noise ratio, and the jth largest one belongs to Then we have: Based on the channel gain, the PMU is divided into two sets, and the weak channel set is: The strong channel set is: Then, the unified channel difference pairing algorithm is used for pairing, that is, the user with a higher signal-to-noise ratio in one set will be paired with the user with a higher signal-to-noise ratio in another set, that is: When K is an even number, NOMA user pairing is as follows: When K is an odd number, the NOMA user pairing is as follows:

5. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 2 is characterized in that: Step 2.2 specifically includes: given a bandwidth allocation scheme, namely The power allocation scheme, that is, {α1,…,α K }, the QoS index of each PMU can be optimized separately without affecting each other; the optimal QoS index should make the following equation true: Given and {α1,…,α K }, the above two equations can be directly solved by the dichotomy method to obtain the optimal QoS index vector, that is, 6. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 2 is characterized in that: Step 2.3 specifically includes: Step 2.3.1: Assume that the optimization is currently being performed on user group g; optimize the power coefficient of user group g. At this time, except for user group g, the power coefficients of other groups are considered as known parameters; use the golden section search algorithm to find the optimal power coefficient so that and Achieve a balance between the two, while maximizing the objective function value; get the optimal and Then, use step 2.2 to solve the optimal QoS index; refer to step 2.2 and use the dichotomy method to find the corresponding and Step 2.3.2: Loop through the user group g, and execute step 2.3.1 each time to optimize and find the optimal power allocation coefficient of the two PMU devices in the user group; then calculate the value of the objective function based on the obtained optimal power allocation coefficient; until the objective function value converges, exit the loop optimization process and obtain all the final optimal power allocation coefficients and in 7. The resource allocation strategy for ensuring grid observability based on NOMA mode according to claim 2 is characterized in that: Step 2.4 specifically includes: Step 2.4.1: The objective function is recorded as: Planning algorithm initialization, assuming the total bandwidth is B total , the initial bandwidth of K PMUs in the system is Then use step 2.3 to solve the power allocation scheme, use step 2.2 to solve the QoS index, and calculate the initial objective function value, denoted as F; Step 2.4.2: Randomly generate a neighborhood value B′ of the current bandwidth vector B; Step 2.4.3: Solve the optimal power allocation vector α′ according to steps 2.2-2.3; Step 2.4.4: Solve the QoS index vector θ′ according to step 2.2; Step 2.4.5: Calculation Step 2.4.6: Let △E=F′-F. If △E<0, B=B′, F=F′; otherwise, with probability Accept the new solution, B = B′; Step 2.4.7: Repeat steps 2.4.2 to 2.4.6 until M iterations have been completed, where M is the average Markov chain length. Step 2.4.8: Update the temperature, set T = βT, where β is the temperature attenuation coefficient; Step 2.4.9: Repeat steps 2.4.2 to 2.4.8 until the termination temperature is reached.

Images