A natural computing method based on swarm intelligence

Through a natural calculation method based on population intelligence, using dynamic biological gradient field information and the 'optimum state retention' mechanism, the micro-nano robot motion path planning in a dynamic environment is realized, and the contrast agent delivery under the influence of biological gradient field changes is solved. The problem of low efficiency achieves precise targeting and efficient delivery of tumor lesions.

CN114021690BActive Publication Date: 2024-11-22YANGTZE DELTA REGION INST (QUZHOU) UNIV OF ELECTRONIC SCI & TECH OF CHINA +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111158377.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-09-30
Publication Date
2024-11-22
Estimated Expiration
2041-09-30

AI Technical Summary

Technical Problem

Under the influence of dynamic changes in the biological gradient field, it is difficult for the prior art to realize real-time planning of micro-nano robot motion paths, resulting in low contrast agent delivery efficiency.

Method used

Adopting natural computing methods based on group intelligence, by establishing natural computing models, dynamic biological gradient field characterization and 'optimal state retention' mechanism, combined with external controllable micro-nano robots, we overcome the influence of the dynamic environment in the body, and achieve accurate solution and efficient delivery. .

Benefits of technology

The efficiency of contrast agent delivery is improved, ensuring that micro-nano robots can effectively move along the biological gradient field information, achieving accurate targeting of tumor lesions, and significantly improving delivery efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114021690B_ABST
    Figure CN114021690B_ABST
Patent Text Reader

Abstract

The present invention discloses a natural computing method based on swarm intelligence. The method is based on the "weak first" evolutionary strategy in static living body computing and proposes an "optimal state retention" operation mechanism. The mechanism performs multiple samplings in the tracking cycle of a micro-nano robot, calculates the fitness of the intelligent agent at each sampling point, and then selects the maximum fitness value of the same intelligent agent in the same tracking cycle as the final fitness of the intelligent agent in this tracking cycle. The method can be used to evaluate the fitness of the micro-nano robot in a tracking cycle multiple times, and then extract the optimal evaluation state, overcome the time-varying characteristics of the intelligent agent fitness calculation caused by the dynamic biological gradient field, and improve the relative accuracy of the fitness evaluation. Therefore, this operation mechanism ensures the accuracy of the information shared in the multi-agent collaborative search process in the dynamic optimization environment in the body, thereby improving the efficiency of solving the global optimal solution in natural computing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of interdisciplinary studies and involves research directions such as computational intelligence, micro-nano robotics, and information medicine, and specifically relates to a natural computing method based on swarm intelligence. Background Art

[0002] Traditional medical imaging technology is limited by the resolution of the imaging system and cannot achieve live tissue imaging at the cellular level. In recent years, with the continuous penetration of nanotechnology in the medical field, medical imaging technology based on nanoparticles has also developed rapidly. For example, the existing nanoparticle system relies on the human body's blood circulation to provide power for the nanoparticles injected into the human body. The high permeability of blood vessels in the diseased tissue area causes the nanoparticles to continuously gather around the lesion to achieve the transmission of the contrast agent. From a computational perspective, this process can be regarded as a brute-force search. Because it relies on blood circulation as the power source of the nanoparticles and does not apply external force to control them, it will cause the nanoparticles to spread throughout the body and cannot reach the target location along the most effective path.

[0003] Inspired by the idea of ​​multi-agents cooperating with each other in group intelligent computing to find the optimal solution, the present invention proposes a natural computing method based on group magnetic micro-nano robots to improve the delivery efficiency of contrast agents. The method relies on the "biological gradient field (BGF)" information (such as blood flow velocity, pH value, enzyme activity, oxygen concentration, etc.) induced by the lesion to provide guidance for the movement of the micro-nano robot, and monitors it in real time through an imaging system. The robot group shares the BGF information of its position, and then continuously adjusts its motion state through an external control device, and finally realizes the efficient delivery of contrast agent molecules. The BGF of living tissue is affected by the fluidity of body fluids and is in a dynamic change over time, and it needs to be approximated by a dynamically changing optimizable function. Based on this, the present invention proposes a natural computing method for dynamic BGF, which can be used to control the group micro-nano robot in vitro to overcome the physical constraints brought by the tissue microenvironment in the body, and realize the efficient delivery of contrast agent molecules at the target position. Summary of the invention

[0004] The technical problem to be solved by the present invention is: under the influence of the dynamic changes of the biological gradient field, ensure that the group of micro-nano robots can utilize effective biological gradient field information, perform real-time planning of the motion path of the micro-nano robots through natural computing methods, and achieve improved contrast agent delivery efficiency.

[0005] The present invention proposes a natural computing method based on swarm intelligence, which is implemented based on an externally controllable micro-nano robot (hereinafter referred to as a micro-nano robot) to overcome the impact of the dynamic changes of the internal optimization environment on the fitness calculation of multiple agents and achieve accurate solutions in the internal dynamic optimization environment. The research focus of the present invention includes the model and solution strategy of natural computing.

[0006] The present invention proposes a natural computing method based on swarm intelligence, which includes the following three steps:

[0007] 1) Establishment of natural computing model

[0008] Inspired by the idea of ​​swarm intelligence computing, this paper proposes a basic theoretical framework of natural computing and builds a natural computing model based on it. In this natural computing model, it is assumed that there are K intelligent agents A1, A2, ..., A K , which are used to search the space Search for the best solution

[0009] In the present invention, it is assumed that the bio-gradient field has periodic and non-periodic changes over time. The in vivo objective function to be optimized can be expressed as follows:

[0010]

[0011] Among them, ms (x, y, z; A k ; t) represents the kth agent A k (k=1, 2, ..., K) an externally measurable target function at position (x, y, z) at time t; in (x, y, z; t) represents the in vivo target function at (x, y, z) at time t, i.e., the biological gradient field function; fit (x, y, z; A k ; t) is to overcome A k and The interaction of ex (x, y, z; A k ; t) and the fitness correction term introduced; Indicates that Ak's position coordinates (x, y, z) are always in the search space In this natural computational model, the global optimal solution The position of will not change during the entire calculation process, that is, the position of the tumor lesion remains unchanged during the detection process.

[0012] 2) Establishment of fitness calculation model

[0013] Since the calculation of the agent's fitness in the natural computing process plays a decisive role in the performance of each iteration of the algorithm, the establishment of the agent's fitness calculation model is an important part of the natural computing method. The construction of the fitness calculation model includes the following two sub-steps:

[0014] a) Characterization of dynamic bio-gradient fields

[0015] Limited by medical technology conditions, existing literature records only make qualitative experimental observations on the biological gradient field involved in the present invention, and there is no widely accepted quantitative model yet. Therefore, at this stage, some representative standard test functions are mainly used to represent the overall distribution of the biological gradient field. For example, the body fluid in the tumor tissue microenvironment is in a flowing state, and the distribution of biochemical information (such as blood flow velocity, pH value, enzyme activity, cytokines, etc.) in the fluid changes dynamically over time. Therefore, the static biological gradient field cannot accurately represent the real application scenario of natural computing. In order to conduct a deeper study on the theory of natural computing, it is necessary to consider the dynamic changes of the biological gradient field over time.

[0016] The present invention designs a regularized biological gradient field (BGF) function, whose function value gradually decreases as the coordinate position approaches the origin in the parameter space. The spherical area with a radius of 0.5 mm at the center of the parameter space represents the position to be solved. The specific mathematical expression of the function is shown in formula (2), which represents the static state of the BGF function. Based on this, in order to simulate the dynamic change of the biological gradient field, the present invention adds a dynamic change factor to the basic function (the biological gradient field function described by formula (2) is used as the basic function), thereby obtaining two corresponding time-varying models.

[0017]

[0018] o in step 1) involved in the present invention in The (x, y, z; t) function may satisfy periodic sinusoidal variation or non-periodic Gaussian variation. The two corresponding time-varying models are the dynamic BGF (i.e., step 1) that satisfies periodic sinusoidal variation (H sin(ωt)) in (x, y, z; t)), where H represents the amplitude of the sinusoidal change of the function, ω represents the frequency of the sinusoidal change of the function; and the dynamic BGF that satisfies the non-periodic Gaussian change (i.e., o in step 1) in (x, y, z; t)).

[0019] b) “Optimal state retention” mechanism

[0020] Each motion cycle of natural computing includes two stages: position update and position estimation of the agent. Among them, the control process of the micro-nano robot corresponds to the directional update stage of the agent's position, which directly affects the optimization results of each step of the biological gradient field function. The tracking process of the micro-nano robot corresponds to the position estimation stage of the agent. This process estimates the wide-area biological gradient field information by learning the collected local biological gradient field information, and then guides the next control process. Therefore, the tracking accuracy of the micro-nano robot directly affects its next control effect.

[0021] Affected by the dynamic changes of the bio-gradient field, the local bio-gradient field information collected by the micro-nano robot is in dynamic change during the tracking cycle, which brings challenges to the evaluation of the fitness of the intelligent agent. In order to solve this problem, the present invention proposes an "optimal state retention" mechanism.

[0022] The specific steps are as follows:

[0023] i) By controlling the agents A1, A2, ..., A K During multiple samplings in the tracking cycle, we calculate the S1, S2, ..., S of each agent at different times in the tracking cycle. m The fitness of , where m represents the total number of sampling times in a tracking cycle.

[0024] ii) For each agent at time S1, S2, ..., S m The calculated fitnesses are compared and the maximum value is selected as the final fitness of the agent.

[0025] iii) Select the final fitness of all agents A1 / S3, A2 / S1, A3 / S respectively m , …, A K / S2 constitutes the final fitness sequence of the multi-agent system in this tracking cycle.

[0026] 3) Multi-agent control strategy

[0027] In swarm intelligence algorithms (such as PSO (particle swarm optimization algorithm), GSA (gravitational search algorithm), etc.), multiple agents implement iterative updates of their own positions and movement speeds according to certain operation rules. The agent update method is instantaneous, that is, each agent can complete the ideal position update according to its own needs at the same time. In living body computing, the position update of the agent is consistent under the influence of the external unified control magnetic field. Therefore, in each step of the iterative update, it is necessary to select an optimal position update method to achieve the overall optimal evolution of the multi-agent system.

[0028] The present invention uses the "weak first" evolutionary strategy to control the micro-nano robot group to achieve the optimal update of the intelligent body position in the living body calculation under the action of the external unified control magnetic field. m , …, A K The individual corresponding to the minimum fitness in / S2 is selected as the individual w1 in the group, and the position update method of w1 is used as the update method of the entire micro-nano robot group, and the state update of the micro-nano robot group is realized under the control of the external magnetic field. According to the above control strategy, the individual w2 with the minimum fitness is selected in the next tracking cycle of the iterative operation to realize the further update of the movement of the micro-nano robot group.

[0029] The present invention is based on the "weak first" evolutionary strategy in static in vivo computing and proposes an "optimal state retention" operation mechanism. This mechanism performs multiple samplings in the tracking cycle of the micro-nano robot, calculates the fitness of the agent at each sampling point, and then sorts the fitness of multiple sampling points of the same agent in the same tracking cycle by size, and selects the maximum value as the final fitness of the agent in this tracking cycle. This method can be used to evaluate the fitness of the micro-nano robot in a motion cycle multiple times, and then extract the optimal evaluation state, overcome the time-varying characteristics of the agent fitness calculation caused by the dynamic biological gradient field, and improve the relative accuracy of the fitness evaluation. Therefore, this operation mechanism ensures the accuracy of the information shared in the multi-agent collaborative search process in the dynamic optimization environment in the body, thereby improving the efficiency of solving the global optimal solution in natural computing. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Characterization function of BGF: (a) 3D cross-section of static basic BGF function; (b) cross-section of periodically changing BGF at two different times; (c) cross-section of non-periodically changing BGF at two different times;

[0031] Figure 2 This is a schematic diagram of the "optimal state retention" mechanism of the present invention;

[0032] Figure 3 Schematic diagram of the multi-agent control strategy of the present invention;

[0033] Figure 4 This is an example diagram of the application of natural computing in contrast agent delivery;

[0034] Figure 5 This is a comparison chart of contrast agent delivery simulation results;

[0035] Figure 6 The following are statistical diagrams of simulation results: (a) Statistical results using traditional brute force search method; (b) Statistical results using natural calculation. DETAILED DESCRIPTION

[0036] The present invention will be described in detail below based on the accompanying drawings and embodiments, and the technical solution of the present invention will be clearly described. The embodiments selected here are only used to explain the present invention and cannot limit the present invention.

[0037] The present invention proposes a natural computing method based on swarm intelligence. The method is implemented based on an externally controllable micro-nano robot to overcome the influence of the dynamic change of the internal optimization environment on the fitness calculation of multiple agents and realize accurate solution under the internal dynamic optimization environment. The agent refers to an externally controllable micro-nano robot, hereinafter referred to as a micro-nano robot. The natural computing method specifically includes the following steps:

[0038] S1) Establishment of natural computational models

[0039] Construct a natural computing model, in which it is assumed that there are K intelligent agents A1, A2, ..., A K Construct a multi-agent system to search the space Search for the global optimal solution

[0040] Assuming that the biological gradient field has periodic and non-periodic changes over time, the in vivo objective function to be optimized can be expressed as follows:

[0041]

[0042] Among them, ms (x, y, z; A k ; t) represents the kth agent A k An externally measurable target function at position (x, y, z) at time t; o in (x, y, z; t) represents the in vivo target function at (x, y, z) at time t, i.e., the dynamic bio-gradient field function; fit (x, y, z; A k ; t) is to overcome A k and The interaction of ex (x, y, z; A k ; t) and the fitness correction term introduced; Indicates A k The position coordinates (x, y, z) are always in the search space , where k = 1, 2, ..., K;

[0043] In this natural computational model, the global optimal solution The position of will not change during the entire calculation process;

[0044] S2) Establishment of fitness calculation model

[0045] Since the calculation of the agent's fitness plays a decisive role in the performance of each iteration in the natural computing process, the establishment of the agent's fitness calculation model is an important part of the natural computing method. The establishment of the fitness calculation model includes the following two steps:

[0046] S21) Dynamic biological gradient field characterization

[0047] like Figure 1 As shown in (a), a regularized biological gradient field function is designed, the function value of which gradually decreases as the coordinate position approaches the origin in the parameter space. The spherical area with a radius of 0.5 mm and the origin of the function parameter space as the center represents the position to be solved; the mathematical expression of the regularized biological gradient field function f(x, y, z) is shown in formula (2), and the three-dimensional image of the function is shown in Figure 1 (a) shows a static bio-gradient field function; a dynamic change factor is added to the regularized bio-gradient field function to obtain two corresponding time-varying models, respectively as follows: Figure 1 (b) and (c) show the dynamic biological gradient field function o that satisfies periodic sinusoidal changes. in The states of (x, y, z; t) at two different times, and the dynamic biological gradient field function o that satisfies the non-periodic Gaussian change in The state of (x, y, z; t) at two different times, the formula (2) is expressed as:

[0048]

[0049] S22) “Optimal State Retention” Mechanism

[0050] Affected by the dynamic changes of the biological gradient field, the local biological gradient field information collected by the agent is in dynamic change during the tracking cycle, which brings challenges to the evaluation of the agent's fitness. In order to solve this problem, an "optimal state retention" mechanism is proposed, such as Figure 2 As shown, the "optimal state retention" mechanism specifically includes the following operation steps:

[0051] i) By controlling the agents A1, A2, ..., A K Perform multiple sampling in the current tracking cycle and calculate S1, S2, ..., S for each agent at different times in the tracking cycle. m The fitness of, where m represents the total number of sampling times in a tracking cycle;

[0052] ii) For each agent at time S1, S2, ..., S mThe calculated fitness values ​​are compared and the maximum value is selected as the final fitness value of the agent;

[0053] iii) If Figure 2 As shown, the final fitness of all agents is selected to form the final fitness sequence of the multi-agent system in the current tracking cycle.

[0054] S3) Constructing multi-agent control strategy

[0055] In living body computing, under the influence of an external unified control magnetic field, the position update of the agent is consistent. In each step of iterative update, an optimal position update method needs to be selected to achieve the overall optimal evolution of the multi-agent system. Therefore, the "weak first" evolutionary strategy is used to control the multi-agent to achieve the optimal update of the agent position in living body computing under the influence of an external unified control magnetic field: Figure 3 As shown in the figure, the agent corresponding to the minimum fitness in the final fitness sequence of the current tracking cycle is selected as the selected individual w1 in the multi-agent system, and the position update method of w1 is used as the update method of the entire multi-agent system, and the state update of the multi-agent system is realized under the action of the external unified control magnetic field; according to the above control strategy, the individual w2 with the minimum fitness in the final fitness sequence is selected in the next tracking cycle of the iterative operation to realize further updating of the motion status of the multi-agent system.

[0056] The present invention is mainly used in contrast agent or drug targeted delivery based on multi-agent. The main goal is to achieve accurate targeting of the target position by the multi-agent system in a dynamic bio-gradient field environment. The method first samples the agent in the tracking cycle multiple times, calculates the fitness of different sampling points respectively, and then uses the "optimal state retention" mechanism to achieve the best fitness evaluation of the agent in the tracking cycle, overcomes the influence of the dynamic change characteristics of the bio-gradient field, and finally, uses the "weak first" evolutionary strategy to achieve the optimal realization of multi-agent motion control. The following content is the implementation steps of the method of the present invention applied to specific scenarios. At this time, the swarm intelligence algorithm selects the PSO algorithm.

[0057] The natural computing method based on swarm intelligence proposed in the present invention is applied to detecting the location of tumor lesions in contrast agent or drug targeted delivery based on multi-agents, and specifically comprises the following steps:

[0058] a. Set the specific number of agents in the multi-agent system and the initial position of each agent in the search space;

[0059] b. Calculate the initial fitness of the agent based on the position distribution of the agent in the search space;

[0060] c. Using the PSO algorithm as the basic algorithm, calculate the "historically optimal" agent position and the “globally optimal” agent position

[0061] d. Perform multiple sampling of randomly moving agents during the tracking period, and use the “optimal state retention” mechanism to calculate the final fitness of each agent in each iterative calculation.

[0062] e. Use the "weak first" evolutionary strategy to achieve directional motion control of the multi-agent system in the control cycle (the control cycle and the tracking cycle constitute the motion cycle of the agent).

[0063] f. Evaluate the fitness and lifespan of each agent, estimate the relative distance between each agent and the tumor lesion, and remove the agent from the multi-agent system when the lifespan of an agent ends; when an agent reaches the location of the tumor lesion, the agent stops moving;

[0064] g. When the maximum running time of the algorithm is reached, the final position distribution of the multi-agent system in the search space is calculated to obtain the spatial position of the tumor lesion and the detection efficiency of the agent for the tumor lesion position.

[0065] The agent's directional motion formula:

[0066]

[0067] Where c1 and c2 are two different learning factors, usually any number in [0, 4]; r1 and r2 represent random numbers between (0, 1) respectively; represents the blood flow rate at time (t+1); Represents the agent A selected by the “weak first” evolutionary strategy j (j=1, 2, ..., K) velocity at time t; and Respectively represent any agent A i (i=1, 2, ..., K) position coordinates at time t and time t+1; and Represents agent A i (i=1, 2, ..., K) is the movement speed at time t and time t+1; d0 represents the movement speed of the agent under the sole control of the external driving field; A is obtained by applying the "optimal state retention" mechanism and the "weak first" evolutionary strategy. i The speed of movement at time t+1; ||*||2 represents the 2-norm.

[0068] The algorithm flow of the above method (abbreviated as Nat.Comp.) is as follows:

[0069]

[0070] like Figure 4 As shown, the natural computing method proposed in the present invention is applied to the contrast agent delivery problem in tumor medical imaging to improve the delivery efficiency. The number of intelligent agents (i.e., micro-nano robots) used in this example is 20, and the above algorithm flow can be used to complete the iterative operation process of natural computing through micro-nano robot control and tracking equipment. Figure 5 It shows the simulation results of targeted delivery to the target area (tumor) using the traditional brute force search (BF) method and the natural computing (Nat. Comp.) method proposed by the present invention. It can be seen that the motion trajectory of the agent using the natural computing method can pass through the target area, and most of the agents eventually stay in the target area. However, the motion trajectory of the agent using the brute force search method failed to pass through the target area, and most of the agents eventually failed to stay in the target area. It can be seen that the natural computing method proposed by the present invention performs better than the traditional method in this example. Figure 6 The simulation results after running the simulation example 1000 times are shown in the figure. The horizontal axis represents the number of agents, the vertical axis represents the number of simulations, and η represents the efficiency of contrast agent delivery (the proportion of agents staying at the target position to the total number of agents). It can be seen that the delivery efficiency of the natural computing method is 31.1% (e.g. Figure 6 (b)), which is much higher than the traditional brute force search method with an efficiency of 11.9% ( Figure 6 This proves the effectiveness of the method proposed in the present invention.

[0071] The embodiments described above are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Claims

1. A natural computing method based on swarm intelligence, characterized in that: The natural computing method is used to overcome the influence of dynamic changes in the in vivo optimization environment on the fitness calculation of multiple intelligent agents, and realize accurate solution in the in vivo dynamic optimization environment, wherein the intelligent agent refers to an externally controllable micro-nano robot; the natural computing method specifically includes the following steps: S1) Establishment of natural computational models Construct a natural computing model, in which it is assumed that there are K intelligent agents A1, A2, …, A K Construct a multi-agent system to search the space Search for the global optimal solution Assuming that the biological gradient field has periodic and non-periodic changes over time, the in vivo objective function to be optimized can be expressed as follows: in, represents the kth agent A k An externally measurable target function at position (x, y, z) at time t; o in (x, y, z; t) represents the in vivo target function at (x, y, z) at time t, i.e., the dynamic bio-gradient field function; To overcome A k and Interaction The fitness correction term introduced; Indicates A k The position coordinates (x, y, z) are always in the search space , where k = 1, 2, ..., K; In this natural computational model, the global optimal solution The position of will not change during the entire calculation process; S2) Establishment of fitness calculation model Since the calculation of the agent's fitness plays a decisive role in the performance of each iteration in the natural computing process, the establishment of the agent's fitness calculation model is an important part of the natural computing method. The establishment of the fitness calculation model includes the following two steps: S21) Dynamic biological gradient field characterization A regularized biological gradient field function is designed, the function value of which gradually decreases as the coordinate position approaches the origin in the parameter space. A spherical area with a radius of 0.5 mm and the origin of the function parameter space as the center represents the position to be solved. The mathematical expression of the regularized biological gradient field function f(x, y, z) is shown in formula (2), which represents the static biological gradient field function. A dynamic change factor is added to the regularized biological gradient field function to obtain two corresponding time-varying models, namely, a dynamic biological gradient field function o that satisfies periodic sinusoidal changes and a dynamic biological gradient field function o that satisfies periodic sinusoidal changes. in (x, y, z; t), and the dynamic biological gradient field function o that satisfies the non-periodic Gaussian change in (x, y, z; t), the formula (2) is expressed as: S22) "Optimal state retention" mechanism Affected by the dynamic changes of the bio-gradient field, the local bio-gradient field information collected by the agent during the tracking cycle is in dynamic change, which brings challenges to the evaluation of the agent's fitness. In order to solve this problem, an "optimal state retention" mechanism is proposed, which specifically includes the following operating steps: i) By controlling the agents A1, A2, ..., A K Perform multiple sampling in the current tracking cycle and calculate S1, S2, ..., S for each agent at different times in the tracking cycle. m The fitness of, where m represents the total number of sampling times in a tracking cycle; ii) For each agent at time S1, S2, ..., S m The calculated fitness values ​​are compared and the maximum value is selected as the final fitness value of the agent; iii) Selecting the final fitness of all agents respectively to form the final fitness sequence of the multi-agent system in the current tracking cycle; S3) Constructing multi-agent control strategy In living body computing, under the influence of an external unified control magnetic field, the position update of the agent is consistent. In each step of iterative update, an optimal position update method needs to be selected to achieve the overall optimal evolution of the multi-agent system. Therefore, the "weak first" evolutionary strategy is adopted to control the multi-agent to achieve the optimal update of the agent position in living body computing under the influence of an external unified control magnetic field: the agent corresponding to the minimum fitness in the final fitness sequence of the current tracking cycle is selected as the selected individual w1 in the multi-agent system, and the position update method of w1 is used as the update method of the entire multi-agent system, and the state update of the multi-agent system is achieved under the influence of an external unified control magnetic field; according to the above control strategy, the individual w2 with the minimum fitness in the final fitness sequence is selected in the next tracking cycle of the iterative operation to achieve further update of the motion of the multi-agent system; When the natural computing method is applied to detecting the location of tumor lesions in contrast agent or drug targeted delivery, it specifically includes the following steps: a. Set the number of agents in the multi-agent system and the initial position of each agent in the search space; b. Calculate the initial fitness of each agent based on its initial position in the search space; c. Use the particle swarm optimization algorithm as a swarm intelligence algorithm to calculate the "historically optimal" position of the agent and the "globally optimal" agent position d. Sample the randomly moving agents multiple times during the tracking period, and use the "optimal state retention" mechanism to calculate the final fitness of each agent in each iterative calculation; e. Use the "weak first" evolutionary strategy to achieve directional motion control of the multi-agent system in the control cycle. The control cycle and tracking cycle constitute the motion cycle of the agent; f. Evaluate the fitness and lifespan of each agent, and estimate the relative distance between each agent and the tumor lesion. When the lifespan of an agent ends, remove the agent from the multi-agent system; when an agent reaches the location of the tumor lesion, the agent stops moving; g. When the maximum running time of the algorithm is reached, the final position distribution of the multi-agent system in the search space is calculated to obtain the spatial position of the tumor lesion and the detection efficiency of the multi-agent system for the tumor lesion position.

2. The natural computing method based on swarm intelligence according to claim 1, characterized in that: The directional motion formula of the agent is expressed as follows: Among them, c1 and c2 are two different learning factors, and their values ​​are any numbers in [0,4]. r1 and r2 represent random numbers between (0,1). represents the blood flow rate at time (t+1); Represents the agent A selected by the "weak first" evolutionary strategy j The speed at time t; and Respectively represent any agent A i Position coordinates at time t and time t+1; and Represents agent A i The movement speed at time t and time t+1; d0 represents the movement speed of the agent under the sole control of the external driving field; A is obtained by applying the "optimal state retention" mechanism and the "weak first" evolutionary strategy. i The movement speed at time t+1, i=1,2,…,K, j=1,2,…,K; ||*||2 represents the 2-norm.

3. The natural computing method based on swarm intelligence according to claim 2, characterized in that: Said K=20.