Dynamic multi-objective optimization method, device and equipment for population co-evolution and medium

Abstract

This invention specifically relates to a dynamic multi-objective optimization method, apparatus, device, and medium based on population co-evolution, comprising: acquiring the objective function and constraints of a target radio network; generating three populations based on the objective function and constraints, including a primary population, a first auxiliary population, and a second auxiliary population; detecting the environment of the target radio network and obtaining the detection results; when the detection results indicate that the environment is static, optimizing the three populations using multi-population co-evolution (MPCE); when the detection results indicate that the environment is changing, updating the three populations using an adaptive combined response mechanism (AIRS), and then optimizing the three populations using MPCE; and obtaining the optimization result after optimizing the three populations for a preset number of rounds. This invention improves the communication efficiency of radio networks and reduces communication overhead.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, specifically relating to a dynamic multi-objective optimization method, apparatus, device, and medium for group collaborative evolution. Background Technology

[0002] Dynamically constrained multi-objective optimization problems (DCMOPs) combine the characteristics of dynamic multi-objective optimization problems (DMOPs) and constrained multi-objective optimization problems (CMOPs), including time-varying objective functions, problem parameters, and constraints. Taking power system dispatch as an example, its objective is to reduce power generation costs and carbon emissions while meeting power demand and load constraints at different points in time.

[0003] Due to the influence of time-varying objective functions and constraints, DCMOPs have unique characteristics that differ from DMOPs and CMOPs:

[0004] 1) The changing nature of complex feasible regions: Unlike the static feasible regions in CMOPs, DCMOPs are characterized by time-varying constraints, resulting in dynamic shapes, sizes, and locations of feasible regions. This often leads to irregular and fragmented feasible regions. With limited computational resources, a comprehensive exploration of these feasible regions becomes extremely challenging, significantly increasing the difficulty of finding solutions in dynamic environments.

[0005] 2) In DCMOPs, in addition to the time-varying objective function in DMOPs, the changes in the feasible region caused by time-varying constraints further affect the feasibility and continuity of CPF. The dual influence of the time-varying objective function and constraints leads to complex changes in the distribution, shape, and size of CPF.

[0006] However, existing technologies have the following problems:

[0007] (1) Lack of adaptive integration of response strategies. Most existing DCMOEAs rely on a single type of response strategy. A few DCMOEAs use dynamic response strategies that integrate response strategies, but the integration of different strategies is fixed by static rules. These response strategies cannot adaptively adjust the integration mode and weight configuration of response strategies according to the severity of environmental changes, resulting in inefficiency when dealing with drastic environmental changes.

[0008] (2) Underutilization of potentially valuable but infeasible solutions. On the one hand, infeasible solutions located between the CPF and the unconstrained Pareto optimal front (UPF) help explore feasible regions close to the CPF. On the other hand, infeasible solutions close to the UPF can help other solutions cross infeasible regions, and these valuable infeasible solutions provide important information for the population's evolution toward the CPF in dynamic environments. Summary of the Invention

[0009] The main objective of this invention is to propose a dynamic multi-objective optimization method, apparatus, device, and medium for population co-evolution.

[0010] One aspect of the present invention provides a dynamic multi-objective optimization method for population co-evolution, comprising:

[0011] Obtain the objective function and constraints of the target radio network. The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of each communication link, with each communication link satisfying a minimum SINR threshold. The constraints include the dynamic influence of the sum of SINR on channel gain, interference level, and noise power.

[0012] Based on the objective function and the constraints, three populations are generated, including the primary population, the first auxiliary population, and the second auxiliary population.

[0013] The environment of the target radio network is detected, and the detection results are obtained, including both static and changing environmental conditions.

[0014] When the test results show that the environment is static, the three populations are optimized using a multi-population co-evolution approach.

[0015] When the detection result indicates environmental change, an adaptive combined response mechanism is used to update the three groups, and then the three groups are optimized through multi-group co-evolution. The environmental change is used to characterize the dynamic changes of channel gain, interference level and noise power over time.

[0016] The three groups are optimized through a preset number of rounds to obtain the optimization results, which are used to characterize the CPS and CPF at different time steps. The CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and the CPF is the mapping of the CPS in the objective function space.

[0017] According to the dynamic multi-objective optimization method of population co-evolution, the three types of populations include:

[0018] The main population searches for CPFs in the feasible region according to the dynamic constraint dominance principle and constraint conditions, where the feasible region is used to characterize solutions that meet the constraint conditions under dynamic constraints.

[0019] The first auxiliary population is used to find regions close to the UPF, where UPF represents the Pareto optimal solution set without constraints, and the solutions in the optimal solution set are used to characterize the satisfaction of the objective function.

[0020] The second auxiliary population is used to explore the solution space between UPF and CPF, where CPF represents the set of all communication link power levels that, under the condition that all communication links meet the SINR threshold, cannot improve one objective function by adjusting the communication link power level to minimize the total power and maximize the sum of SINR of all communication links without worsening the other objective function; CPS is used to characterize the set of total power and sum of SINR of communication links corresponding to the communication link levels in CPS.

[0021] According to the dynamic multi-objective optimization method of population co-evolution, when the detection result indicates that the environment is static, the three populations are optimized using a multi-population co-evolution (MPCE) approach, including:

[0022] Genetic operators are used to generate offspring populations for the three populations;

[0023] The offspring population was integrated with each of the three populations to obtain an integrated population;

[0024] For each integrated population, an environmental selection method is used to search for individuals, and the integrated population is updated based on the search results.

[0025] According to the dynamic multi-objective optimization method for population co-evolution, an environmental selection method is used to search for individuals in each integrated population, and the integrated population is updated based on the search results, including:

[0026] When the integrated population is obtained by integrating the main population and its offspring population, environmental selection is performed using the dynamic constraint dominance principle, which is expressed as follows:

[0027]

[0028] in, and The step size is The solution at that time, if and If any condition of the dynamic constraint dominance principle is satisfied, then Dynamic constraint domination , recorded as , Let be the feasible region with step size t. To constrain the amount of violations;

[0029] When the integrated population is obtained by integrating the first auxiliary population and the offspring population of the first auxiliary population, environmental selection is performed using dynamic Pareto dominance, where dynamic Pareto dominance is expressed as:

[0030]

[0031] If and If all conditions for dynamic Pareto dominance are met, then Dynamic Pareto Domination , represented as , i and j The objective function is identified by its ordinal number. m The total number of objective functions;

[0032] When the integrated population is obtained by integrating the second auxiliary population and the offspring population of the second auxiliary population, an improved dynamic constraint dominance principle is adopted for environmental selection, wherein the improved dynamic constraint dominance principle is as follows:

[0033]

[0034] in, Indicates the current iteration's step length. The constraint violation threshold is calculated as follows:

[0035]

[0036]

[0037] in, It is the number of iterations for each environment. It is the iteration counter for the current environment. It is control The parameter of the rate of descent. It is time The initial constraint violation value is equal to the time. main population The maximum degree of constraint violation.

[0038] According to the dynamic multi-objective optimization method of population co-evolution, when the detection result indicates environmental change, the three populations are updated using an adaptive combined response mechanism (AIRS), and then optimized through a multi-population co-evolutionary approach (MPCE), including:

[0039] Multiple offspring were selected from the main population at the previous time step using memory-based response strategies, diversity-based response strategies, and prediction-based response strategies to obtain the first offspring.

[0040] After applying adaptive weights to the first descendant generated by the response strategy, a selection is made to obtain the first temporary descendant and the second temporary descendant.

[0041] The first temporary offspring and the second temporary offspring are combined to obtain the second offspring;

[0042] Environmental selection was carried out on the second offspring using the dynamic constraint dominance principle, dynamic Pareto dominance, and improved dynamic constraint dominance principle to obtain a new primary population, a new first auxiliary population, and a new second auxiliary population.

[0043] According to the dynamic multi-objective optimization method for population co-evolution, multiple offspring are selected from the main population at the previous time step using a memory-based response strategy, a diversity-based response strategy, and a prediction-based response strategy to obtain the first offspring, including:

[0044] A new principal population is generated using a memory-based response strategy. This includes initializing an empty population, selecting multiple individuals from the principal population based on the dynamic constraint dominance principle, and performing multiple Gaussian mutation operations on each individual. The resulting new solution is as follows:

[0045]

[0046] in, For the new solution, Represents a random number that follows a standard normal distribution. For the solution The j One variable, For local search parameters, For the first j The upper limit of a variable, For the first j The lower bound of each variable.

[0047] Furthermore, the variable range for the new solution is revised using the following formula:

[0048]

[0049] The revised new solution Add individuals to the empty population, perform an evaluation based on the dynamic constraint dominance principle, and select multiple individuals to obtain a new main population;

[0050] A diversity-based response strategy was employed to screen the main populations based on historical environment, yielding the first offspring. Differential perturbation was then applied to the first offspring to obtain a new solution. The difference perturbation is handled as follows:

[0051]

[0052] in, and These are two randomly selected different solutions. A random number in the range [0,1]. It is a scaling factor used to control the magnitude of the differential perturbation, correcting the variable range of the new solution, and based on the environment selection method of the dynamic constraint dominance principle, it is used for re-evaluation. Selected Individuals receive a new first auxiliary population. Indicates the first One solution;

[0053] A predictive response strategy is used to generate new second auxiliary population offspring, including generating the first offspring and calculating the population centroids at the first two time steps. and The calculation formula is:

[0054]

[0055] in, Indicates the time step Approximate CPS obtained at that time express The number of solutions;

[0056] Calculate the population centroid The movement vector:

[0057]

[0058] Among them, the center point The movement vector represents the evolution trend of CPS in the predicted response strategy. >1;

[0059] right Each individual application in the process is based on a centroid prediction operation to generate a new solution:

[0060]

[0061] in, This is a scaling factor used to control the magnitude of the prediction shift.

[0062] According to the dynamic multi-objective optimization method of population co-evolution, the adaptive weights include:

[0063] Calculate the average minimum Euclidean distance between the generated first offspring and the ideal set, and determine the population quality using the average minimum Euclidean distance. The calculation method is as follows:

[0064]

[0065] in, This represents a set of memory-based response strategies, diversity-based response strategies, and prediction-based response strategies. Indicates population The non-dominated solution set, where Indicates two individuals in the first descendant. and Spatial distance;

[0066] According to the non-dominated solution set The softmax function is used to calculate the weights of each response strategy. for:

[0067]

[0068] Among them, weight With population quality Proportional.

[0069] Another aspect of the present invention discloses a dynamic multi-objective optimization device for population co-evolution, comprising:

[0070] The first module is used to obtain the objective function and constraints of the target radio network. The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of each communication link, and each communication link satisfies the minimum SINR threshold. The constraints include the dynamic influence of the sum of SINR on channel gain, interference level and noise power.

[0071] The second module is used to generate three populations based on the objective function and the constraints. The three populations include the main population, the first auxiliary population, and the second auxiliary population.

[0072] The third module is used to detect the environment of the target radio network and obtain the detection results, which include both static and dynamic environments.

[0073] The fourth module is used to optimize the three groups using a multi-group co-evolution method when the detection result is that the environment is static; or, when the detection result is that the environment changes, to update the three groups using an adaptive combined response mechanism, and then optimize the three groups through a multi-group co-evolution method. The environmental change is used to characterize the dynamic changes of the channel gain, interference level and noise power over time.

[0074] The fifth module is used to optimize the three groups through a preset number of rounds to obtain the optimization results. The optimization results are used to characterize the CPS and CPF at different time steps. The CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and the CPF is the mapping of the CPS in the objective function space.

[0075] Another aspect of the present invention provides an electronic device, including a processor and a memory;

[0076] The memory is used to store programs;

[0077] The processor executes the program to implement the method as described above.

[0078] This invention also discloses a computer-readable storage medium storing a program that is executed by a processor to implement the dynamic multi-objective optimization method for population co-evolution.

[0079] The beneficial effects of this invention are as follows: By using two auxiliary populations—one with no constraints and one with relaxed constraints—the main population is helped to explore the optimal solution space by utilizing potential infeasible solutions. An automatic adaptive response strategy is implemented in the dynamic response component. Based on the severity and pattern of environmental changes, the number of solutions selected from different response strategies is adaptively adjusted, and three initial populations are generated for the new environment. That is, a dynamic multi-objective optimization method of population co-evolution is adopted, which realizes the rapid search for the optimal solution of the power system, improves the communication efficiency of the radio network, and reduces communication consumption. Attached Figure Description

[0080] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0081] Figure 1 This is a schematic diagram of the dynamic multi-objective optimization method for group cooperative evolution according to an embodiment of the present invention.

[0082] Figure 2 This is a schematic diagram of multi-population co-evolution (MPCE) in an embodiment of the present invention.

[0083] Figure 3 This is a framework diagram of the Adaptive Combined Response (AIRS) mechanism according to an embodiment of the present invention.

[0084] Figure 4 This is a schematic diagram of the adaptive combined response mechanism (AIRS) processing flow according to an embodiment of the present invention.

[0085] Figure 5 This is a pseudocode diagram of the memory-based response strategy (RS-M) according to an embodiment of the present invention.

[0086] Figure 6 This is a pseudocode diagram of the diversity-based response strategy (RS-D) according to an embodiment of the present invention.

[0087] Figure 7 This is a pseudocode diagram of the prediction-based response strategy (RS-P) according to an embodiment of the present invention.

[0088] Figure 8This is a schematic diagram of the adaptive weight strategy allocation algorithm according to an embodiment of the present invention.

[0089] Figure 9 This is a schematic diagram of a dynamic multi-objective optimization device for population cooperative evolution according to an embodiment of the present invention. Detailed Implementation

[0090] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings. Throughout the description, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions. In the following description, suffixes such as "module," "part," or "unit" used to denote elements are used only for the purpose of illustrative purposes and have no specific meaning in themselves. Therefore, "module," "part," or "unit" can be used interchangeably. Terms such as "first," "second," etc., are used only to distinguish technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the sequential relationship of the indicated technical features. In the following description, the consecutive reference numerals for method steps are for ease of review and understanding. Adjusting the implementation order of steps, in conjunction with the overall technical solution of the present invention and the logical relationship between the various steps, will not affect the technical effect achieved by the technical solution of the present invention. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0091] refer to Figure 1 , Figure 1 This is a flowchart illustrating a dynamic multi-objective optimization method for population co-evolution, which includes, but is not limited to, steps S100~S500:

[0092] S100: Obtain the objective function and constraints of the target radio network.

[0093] The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of each communication link, with each communication link satisfying a minimum SINR threshold; the constraints include the dynamic influence of the sum of SINR on channel gain, interference level and noise power.

[0094] SINR represents the signal-to-interference-plus-noise ratio.

[0095] S200 generates three populations based on the objective function and constraints: the primary population, the first auxiliary population, and the second auxiliary population.

[0096] In some practical examples, the primary population searches for the CPF in the feasible region according to the dynamic constraint dominance principle and constraint conditions, where the feasible region is used to characterize the solution that meets the objective function under dynamic constraints; the first auxiliary population is used to search for regions close to the UPF, where the UPF represents the Pareto optimal solution set without constraints; and the second auxiliary population is used to explore the solution space between the UPF and the CPF.

[0097] In some embodiments, power management issues in radio networks are prevalent in densely populated wireless network environments where efficient energy use is crucial due to the high density of communication devices and their varying power requirements.

[0098] The embodiments of this invention are presented as a dynamically constrained multi-objective optimization problem. The objective is to minimize the total power and maximize the sum of the signal-to-interference-plus-noise ratio (SINR) for each communication link under dynamic conditions, while ensuring that each communication link meets the minimum SINR threshold. The solution includes the network... The power level of a communication link. SINR is dynamically affected by channel gain, interference level, and noise power. Channel gain represents the quality of the transmission link, which changes over time due to environmental changes or node movement within the network. Interference level represents the impact of other transmitters and electronic noise in the environment on signal quality and also changes dynamically over time. Noise power represents external noise sources affecting communication, such as thermal noise, and also changes dynamically over time. The dynamic environment in this problem refers to the dynamic changes in channel gain, interference level, and noise power over time, thus dynamically affecting the SINR of the communication link. In this problem, CPS means the set of all communication link power levels that, under the condition that all communication links meet the SINR threshold, cannot improve one of the two objective functions of total power and total SINR without worsening the other objective function by adjusting the communication link power level. CPF means the set of total power and total SINR values ​​corresponding to these communication link levels in CPS.

[0099] S300 detects the environment of the target radio network and obtains the detection results, which include both static and dynamic environmental conditions.

[0100] In some practical examples, the environment of the target radio network is its objective function and constraints.

[0101] S400: When the detection result is that the environment is static, the three groups are optimized using a multi-group co-evolution method; or, when the detection result is that the environment changes, the three groups are updated using an adaptive combined response mechanism, and then optimized through a multi-group co-evolution method. The environmental change is used to characterize the dynamic changes of the channel gain, interference level and noise power over time.

[0102] In some embodiments, environmental changes are used to characterize the dynamic effects of channel gain, interference level, and noise power on the dynamic changes over time.

[0103] In some embodiments, the three populations are updated using an adaptive combinatorial response mechanism (AIRS), and then optimized through a multi-population co-evolution (MPCE) approach, including: generating offspring populations from the three populations using genetic operators; integrating the offspring populations with the three populations respectively to obtain integrated populations; performing individual search using an environmental selection method for each integrated population; and updating the integrated populations based on the search results.

[0104] For each integrated population, an environmental selection method is used to search for individuals, and the integrated population is updated based on the search results, including:

[0105] When the integrated population is obtained by integrating the dominant population and its offspring, environmental selection is carried out using the dynamic constraint dominance principle.

[0106] (1)

[0107] in, and The step size is The solution at that time, if and If any condition of the dynamic constraint dominance principle is satisfied, then Dynamic constraint domination , recorded as , Let be the feasible region with step size t. To constrain the amount of violations;

[0108] When the integrated population is obtained by integrating the first auxiliary population and its offspring, environmental selection is performed using dynamic Pareto dominance, which is expressed as:

[0109] (2)

[0110] Among them, if and If all conditions for dynamic Pareto dominance are met, then Dynamic Pareto Domination , represented as , i and j The objective function is identified by its ordinal number. m The total number of objective functions;

[0111] When the integrated population is obtained by integrating the second auxiliary population and the offspring population of the second auxiliary population, an improved dynamic constraint dominance principle is adopted for environmental selection. The improved dynamic constraint dominance principle is as follows:

[0112] (3)

[0113] in, Indicates the current iteration's step length. The constraint violation threshold is calculated as follows:

[0114] (4)

[0115] (5)

[0116] in, It is the number of iterations for each environment. It is the iteration counter for the current environment. It is control The parameter for the rate of descent. It is time The initial constraint violation value is equal to the time. main population The maximum degree of constraint violation.

[0117] In some practical examples, taking MPCE as an example, such as Figure 2 The multi-population co-evolution (MPCE) algorithm 2 randomly initializes three populations, each with a size of [missing value]. : , and . The main population, and and As auxiliary populations, the algorithm enters a static optimization phase when the environment remains constant, as shown in Algorithm 2. The three populations adopt different search directions through different environment selection methods, and their co-evolution enables a more comprehensive search of the CPF in the current environment. When the environment changes, the algorithm enters a dynamic response phase, first updating the time step. Then, AIRS (Adaptive Combined Response) is applied to update the three initial populations in the new environment to effectively respond to environmental changes and facilitate rapid tracking of CPF in the new environment. If the termination condition is met, the CPF and CPS at different time steps are output.

[0118] MPCE comprises three populations: , and . Since it is the main population, considering the constraints, we search for CPFs within the feasible area. The environment selection is based on the Dynamic Constraint Dominance Principle (DCDP). It is a support population that explores regions close to the UPF without considering constraints. Environment selection is achieved through dynamic programming. Using infeasible solutions close to UPF to assist Looking for CPF. It is another auxiliary population with relaxed constraints, designed to explore the solution space between UPF and CPF.

[0119] In the evolution of each environment, the threshold for constraint relaxation gradually decreases, prompting... From UPF to CPF, even in infeasible regions. Therefore, By employing an infeasible solution between UPF and CPF, help A more comprehensive search for optimal regions, with three populations generating offspring independently, helps facilitate knowledge transfer by assimilating offspring from other populations.

[0120] The specific process of MPCE is as follows: First, update the current iteration time. The constraint violation threshold is used as a basis. The environmental selection is the basis. Subsequently, each population is generated through genetic operators. One descendant.

[0121] These descendants then... , and Integration, forming new populations. , and Finally, through DCDP, DPD and Environmental selection methods, from , and Selected from each Individuals constitute a population that evolves co-evolved. , and .

[0122] like Figure 3 The adaptive combined response mechanism (AIRS) framework diagram shown below and Figure 4The diagram shown illustrates the Adaptive Combined Response (AIRS) processing flow. When an environmental change is detected, RS-M and RS-D will adjust their responses based on the dominant population from the previous time step. ,generate and Each contains One descendant (lines 1-2). When time... When the value is greater than 1, RS-P will adjust according to the values ​​of the previous two time steps. and population, for generate One descendant. Otherwise, due to a lack of sufficient historical information for prediction, Through RS-M and Initialization is performed (lines 3-7). The proposed adaptive weighting strategy is used to allocate the number of solutions selected from the three response strategies based on the population quality generated by these strategies. , and (Line 8). To balance exploration and exploitation, two different environment selection methods were employed to select a corresponding number of individuals from three response strategies. Regarding exploitation, the DCDP-based environment selection method was used to select individuals from... , and Select the appropriate number of individuals, and then merge them to form... (Lines 9-10). Similarly, in the exploration aspect, the DPD-based environment selection method was used to select from... , and Select the appropriate number of individuals, and then merge them to form... (Lines 11-12). Finally, and Formed by merger (Line 13). Finally, apply the three types of environment selection. Thus producing , and As a new initial population (lines 14-17).

[0123] refer to Figure 5 , Figure 6 , Figure 7 The pseudocode diagrams for the RS-M, RS-D, and RS-P methods are shown. In RS-M, high-quality solutions, after selecting an initial environment, explore the surrounding solution space using a Gaussian mutation algorithm, thereby helping to discover an initial population suitable for the current environment. Algorithm 4 provides the pseudocode for the RS-M algorithm. The RS-M algorithm first initializes an empty population. (Line 1). DCPD-based environment selection is used to select from... Select from Individual, formation (Line 2). Each individual in Central cities will experience The Cauchy mutation operation, in which It is a scaling factor that determines the number of operations the response strategy performs on the selected individual.

[0124] In some practical examples, a memory-based response strategy is used to generate a new primary population. This includes initializing an empty population, selecting multiple individuals from the primary population based on the dynamic constraint dominance principle, and performing multiple Gaussian mutation operations on each individual to obtain a new solution.

[0125] (6)

[0126] in, For the new solution, Represents a random number that follows a standard normal distribution. For the solution The j One variable, For local search parameters, For the first j The upper limit of a variable, For the first j The lower bound of each variable.

[0127] The variable range for the new solution is revised using the following formula:

[0128] (7)

[0129] The revised new solution Add individuals to the empty population, perform an evaluation based on the dynamic constraint dominance principle, and select multiple individuals to obtain a new main population;

[0130] For the RS-D algorithm, a diversity-based response strategy is used to screen the main population based on historical environment to obtain the first offspring. The first offspring are then subjected to differential perturbation to obtain a new solution. The difference perturbation is handled as follows:

[0131] (8)

[0132] in, and These are two randomly selected different solutions. A random number in the range [0,1]. It is a scaling factor used to control the magnitude of the differential perturbation, correcting the variable range of the new solution, and based on the environment selection method of the dynamic constraint dominance principle, it is used for re-evaluation. Selected Individuals receive a new first auxiliary population. Indicates the first One solution;

[0133] A predictive response strategy is used to generate new second auxiliary population offspring, including generating the first offspring and calculating the population centroids at the first two time steps. and The calculation formula is:

[0134] (9)

[0135] in, Indicates the time step Approximate CPS obtained at that time express The number of solutions;

[0136] Calculate the population centroid The movement vector:

[0137] (10)

[0138] Among them, the center point The movement vector represents the evolution trend of CPS in the predicted response strategy. >1;

[0139] right Each individual application in the process is based on a centroid prediction operation to generate a new solution:

[0140] (11)

[0141] Here, γ is a scaling factor used to control the magnitude of the prediction shift.

[0142] like Figure 8 The diagram shown illustrates the adaptive weight allocation strategy algorithm, which calculates the average minimum Euclidean distance between the first descendant and the ideal set. Population quality is determined by the average minimum Euclidean distance, calculated as follows:

[0143] (12)

[0144] in, This represents a set of memory-based response strategies, diversity-based response strategies, and prediction-based response strategies. Indicates population The non-dominated solution set, where Indicates two individuals in the first descendant and Spatial distance;

[0145] According to the non-dominated solution set The softmax function is used to calculate the weights of each response strategy. for:

[0146] (13)

[0147] Among them, weight With population quality Proportional.

[0148] S500 optimizes the three groups through a preset number of rounds to obtain the optimization results. The optimization results are used to characterize the CPS and CPF at different time steps. CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and CPF is the mapping of CPS in the objective function space.

[0149] It should be noted that UPF represents the Pareto optimal solution set without constraints, that is, the Pareto optimal solution set without considering the dynamic effects of channel gain, interference level, and noise power; CPF represents the set of all communication link power levels that, under the condition that all communication links meet the SINR threshold, cannot improve one objective function by adjusting the communication link power level to minimize the total power and maximize the sum of SINR of all communication links without worsening the other objective function; CPS is used to characterize the set of total power and the sum of SINR of communication links corresponding to the communication link levels in CPS.

[0150] In some practical examples, in each evolutionary iteration, MPCE-AIRS consists of two parts: AIRS for dynamic response and MPCE for static optimization. , , and These represent the number of objective functions, population size, number of decision variables, and scaling factors for the number of times the control response strategy is applied to an individual, respectively.

[0151] Figure 9 This is a diagram of a dynamic multi-objective optimization analysis device for population co-evolution according to an embodiment of the present invention. The device includes a first module 910, a second module 920, a third module 930, a fourth module 940, and a fifth module 950.

[0152] in,

[0153] The first module is used to obtain the objective function and constraints of the target radio network. The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of each communication link, with each communication link satisfying a minimum SINR threshold. The constraints include the dynamic influence of channel gain, interference level, and noise power on the sum of SINR. The second module is used to generate three populations based on the objective function and constraints: a primary population, a first auxiliary population, and a second auxiliary population. The third module is used to detect the environment of the target radio network and obtain the detection results, including both static and changing environments. The fourth module is used for... When the detection result indicates a static environment, the three groups are optimized using a multi-group co-evolutionary approach; or, when the detection result indicates an environmental change, the three groups are updated using an adaptive combined response mechanism, and then optimized using a multi-group co-evolutionary approach. The environmental change is used to characterize the dynamic changes in channel gain, interference level, and noise power over time. The fifth module is used to optimize the three groups through a preset number of rounds to obtain the optimization results. The optimization results are used to characterize the CPS and CPF at different time steps, where CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and CPF is the mapping of CPS in the objective function space.

[0154] For example, with the cooperation of the first to fifth modules in the device, the embodiment device can implement any of the aforementioned dynamic multi-objective optimization methods of population co-evolution, namely, obtaining the objective function and constraints of the target radio network, wherein the objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of each communication link, and each communication link satisfies a minimum SINR threshold; wherein the constraints include the dynamic influence of the sum of SINR on channel gain, interference level, and noise power; generating three groups according to the objective function and constraints, including a primary group, a first auxiliary group, and a second auxiliary group; detecting the environment of the target radio network to obtain the detection results. The results are as follows: the detection results include static and changing environments; when the detection result is static, the three groups are optimized using a multi-group co-evolution method; when the detection result is changing, the three groups are updated using an adaptive combined response mechanism, and then optimized using a multi-group co-evolution method. The environmental change is used to characterize the dynamic changes of channel gain, interference level, and noise power over time. After a preset number of optimization rounds, the optimization results are obtained, which are used to characterize the CPS and CPF at different time steps. The CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and the CPF is the mapping of the CPS in the objective function space. The beneficial effects of this invention are as follows: by using two auxiliary populations, one with no constraints and one with relaxed constraints, the main population is helped to explore the optimal solution space by utilizing potential infeasible solutions. An automatic adaptive response strategy is implemented in the dynamic response component. The number of solutions selected from different response strategies is adaptively adjusted according to the severity and pattern of environmental changes, and three initial populations are generated for the new environment. That is, a dynamic multi-objective optimization method of population co-evolution is adopted, which improves the communication efficiency of the radio network and reduces communication consumption.

[0155] This invention also provides an electronic device, which includes a processor and a memory;

[0156] The memory stores the program;

[0157] The processor executes a program to perform the aforementioned dynamic multi-objective optimization method of population co-evolution; the electronic device has the function of carrying and running the software system of dynamic multi-objective optimization of population co-evolution provided in the embodiments of the present invention, such as a personal computer, minicomputer, mainframe, workstation, network or distributed computing environment, standalone or integrated computer platform, or communicating with charged particle tools or other imaging devices, etc.

[0158] This invention also provides a computer-readable storage medium storing a program that is executed by a processor to implement the dynamic multi-objective optimization method for population co-evolution as described above.

[0159] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0160] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned dynamic multi-objective optimization method for population co-evolution.

[0161] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0162] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0163] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0164] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0165] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0166] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0167] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0168] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A dynamic multi-objective optimization method for co-evolution of a population, characterized in that, include: Obtain the objective function and constraints of the target radio network. The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of all communication links, with each communication link satisfying a minimum SINR threshold. The constraints include the dynamic influence of channel gain, interference level, and noise power on the sum of SINR. Based on the objective function and the constraints, three groups are generated: a primary group, a first auxiliary group, and a second auxiliary group. The primary group searches for the Common Process Factor (CPF) in the feasible region according to the dynamic constraint dominance principle and the constraints, where the feasible region represents a solution that meets the constraints under dynamic constraints. The first auxiliary group searches for regions close to the Universal Process Factor (UPF), where the UPF represents the Pareto optimal solution set without constraints, and the solutions in the optimal solution set represent solutions that satisfy the objective function. The second auxiliary group explores the solution space between the UPF and the CPF, where the CPF represents the set of all communication link power levels that, under the condition that all communication links meet the SINR threshold, cannot improve one objective function (minimizing total power and maximizing the sum of SINR of all communication links) without worsening the other objective function by adjusting the communication link power levels. The Common Process Factor (CPS) represents the set of total power and the sum of SINR of communication links corresponding to the communication link levels in the CPS. The environment of the target radio network is detected, and the detection results are obtained, including both static and changing environmental conditions. When the test results show that the environment is static, the three populations are optimized using a multi-population co-evolution approach. When the detection result indicates environmental change, an adaptive combined response mechanism is used to update the three groups, and then the three groups are optimized through multi-group co-evolution. The environmental change is used to characterize the dynamic changes of channel gain, interference level and noise power over time. The three groups are optimized through a preset number of rounds to obtain the optimization results, which are used to characterize the CPS and CPF at different time steps. The CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and the CPF is the mapping of the CPS in the objective function space.

2. The dynamic multi-objective optimization method for swarm co-evolution according to claim 1, wherein, The detection results indicate that, when the environment is static, the three populations are optimized using a multi-population co-evolution approach, including: Genetic operators are used to generate offspring populations for the three populations; The offspring population was integrated with each of the three populations to obtain an integrated population; For each integrated population, an environmental selection method is used to search for individuals, and the integrated population is updated based on the search results.

3. The dynamic multi-objective optimization method for population co-evolution according to claim 2, characterized in that, The process of using an environmental selection method to search for individuals in each integrated population, and updating the integrated population based on the search results, includes: When the integrated population is obtained by integrating the main population and its offspring population, environmental selection is performed using the dynamic constraint dominance principle, which is expressed as follows: in, and The step size is The solution at that time, if and If any condition of the dynamic constraint dominance principle is satisfied, then Dynamic constraint domination , recorded as , Let be the feasible region with step size t. To constrain the amount of violations; When the integrated population is obtained by integrating the first auxiliary population and the offspring population of the first auxiliary population, environmental selection is performed using dynamic Pareto dominance, where dynamic Pareto dominance is expressed as: If and If all conditions for dynamic Pareto dominance are met, then Dynamic Pareto Domination , represented as i and j are the indexes of the objective functions, and m is the total number of objective functions; When the integrated population is obtained by integrating the second auxiliary population and the offspring population of the second auxiliary population, an improved dynamic constraint dominance principle is adopted for environmental selection, wherein the improved dynamic constraint dominance principle is as follows: in, Indicates the current iteration's step length. The constraint violation threshold is calculated as follows: in, It is the number of iterations for each environment. It is the iteration counter for the current environment. It is control The parameter of the rate of descent. It is time The initial constraint violation value is equal to the time. main population The maximum degree of constraint violation.

4. The dynamic multi-objective optimization method for population co-evolution according to claim 1, characterized in that, When the detection results indicate environmental changes, an adaptive combined response mechanism is used to update the three populations, and then the three populations are optimized through multi-population co-evolution, including: Multiple offspring were selected from the main population at the previous time step using memory-based response strategies, diversity-based response strategies, and prediction-based response strategies to obtain the first offspring. After applying adaptive weights to the first descendant generated by the response strategy, a selection is made to obtain the first temporary descendant and the second temporary descendant. The first temporary offspring and the second temporary offspring are combined to obtain the second offspring; Environmental selection was carried out on the second offspring using the dynamic constraint dominance principle, dynamic Pareto dominance, and improved dynamic constraint dominance principle to obtain a new primary population, a new first auxiliary population, and a new second auxiliary population.

5. The dynamic multi-objective optimization method for population co-evolution according to claim 4, characterized in that, The first offspring is obtained by selecting multiple offspring from the main population at the previous time step using memory-based, diversity-based, and prediction-based response strategies, including: A new principal population is generated using a memory-based response strategy. This includes initializing an empty population, selecting multiple individuals from the principal population based on the dynamic constraint dominance principle, and performing multiple Gaussian mutation operations on each individual. The resulting new solution is as follows: in, For the new solution, Represents a random number that follows a standard normal distribution. For the solution The j-th variable, For local search parameters, Let j be the upper limit of the j-th variable. Let be the lower bound of the j-th variable. Furthermore, the variable range for the new solution is revised using the following formula: The revised new solution Add individuals to the empty population, perform an evaluation based on the dynamic constraint dominance principle, and select multiple individuals to obtain a new main population; A diversity-based response strategy was employed to screen the main populations based on historical environment, yielding the first offspring. Differential perturbation was then applied to the first offspring to obtain a new solution. The difference perturbation is handled as follows: in, and These are two randomly selected different solutions. A random number in the range [0,1]. It is a scaling factor used to control the magnitude of the differential perturbation, correcting the variable range of the new solution, and based on the environment selection method of the dynamic constraint dominance principle, it is used for re-evaluation. Selected Individuals receive a new first auxiliary population. Indicates the first One solution; A predictive response strategy is used to generate new second auxiliary population offspring, including generating the first offspring and calculating the population centroids at the first two time steps. and The calculation formula is: in, Indicates the time step Approximate CPS obtained at that time express The number of solutions; Calculate the population centroid The movement vector: Among them, the center point The movement vector represents the evolution trend of CPS in the predicted response strategy. >1; right Each individual application in the process is based on a centroid prediction operation to generate a new solution: in, This is a scaling factor used to control the magnitude of the prediction shift.

6. The dynamic multi-objective optimization method for population co-evolution according to claim 4, characterized in that, The adaptive weights include: Calculate the average minimum Euclidean distance between the generated first offspring and the ideal set, and determine the population quality using the average minimum Euclidean distance. The calculation method is as follows: in, This represents a set of memory-based response strategies, diversity-based response strategies, and prediction-based response strategies. Indicates population The non-dominated solution set, where Indicates two individuals in the first descendant. and Spatial distance; According to the non-dominated solution set The softmax function is used to calculate the weights of each response strategy. for: Among them, weight With population quality Proportional.

7. A dynamic multi-objective optimization device for population co-evolution, characterized in that, include: The first module is used to obtain the objective function and constraints of the target radio network. The objective function includes minimizing the total power of the target radio network and maximizing the sum of SINR of all communication links, and each communication link satisfies the minimum SINR threshold. The constraints include the dynamic influence of channel gain, interference level and noise power on the sum of SINR. The second module is used to generate three groups based on the objective function and the constraints. These three groups include a primary population, a first auxiliary population, and a second auxiliary population. The primary population searches for a Common Process Factor (CPF) in the feasible region according to the dynamic constraint dominance principle and the constraints. The feasible region represents a solution that meets the constraints under dynamic constraints. The first auxiliary population searches for regions close to the Universal Process Factor (UPF), where the UPF represents the Pareto optimal solution set without constraints, and the solutions in the optimal solution set represent solutions that satisfy the objective function. The second auxiliary population explores the solution space between the UPF and the CPF. The CPF represents the set of all communication link power levels that, under the condition that all communication links meet the SINR threshold, cannot improve one objective function (minimizing total power and maximizing the sum of SINRs of all communication links) without worsening the other objective function by adjusting the communication link power levels. The Common Process Factor (CPS) represents the set of total power and the sum of SINRs of communication links corresponding to the communication link levels in the CPS. The third module is used to detect the environment of the target radio network and obtain the detection results, which include both static and dynamic environments. The fourth module is used to optimize the three groups using a multi-group co-evolution method when the detection result is that the environment is static; or, when the detection result is that the environment changes, to update the three groups using an adaptive combined response mechanism, and then optimize the three groups through a multi-group co-evolution method. The environmental change is used to characterize the dynamic changes of the channel gain, interference level and noise power over time. The fifth module is used to optimize the three groups through a preset number of rounds to obtain the optimization results. The optimization results are used to characterize the CPS and CPF at different time steps. The CPS is the set of Pareto optimal solutions that satisfy the objective function and constraints, and the CPF is the mapping of the CPS in the objective function space.

8. An electronic device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the dynamic multi-objective optimization method for population co-evolution as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a program that is executed by a processor to implement the dynamic multi-objective optimization method for population co-evolution as described in any one of claims 1-6.

Images