Autonomous fire extinguishing robot dog cluster disaster relief material seamless supply method based on dynamic load triggering

Abstract

The invention discloses an autonomous fire extinguishing robot dog cluster disaster relief material seamless supply method based on dynamic load triggering, and the method comprises the steps: constructing a multi-resource health degree model, dividing robot dog resources into a core output resource, an auxiliary output resource and an action survival resource, and calculating the health degree of each type of resources through weighting; adjusting a resource weight coefficient in real time according to the fire behavior grade and the service life of a quick-wear part; a multi-level threshold triggering mechanism is defined, a threshold is dynamically calculated according to the hierarchical response logic of core output resource triggering, action survival resource triggering and auxiliary output resource triggering in combination with the fire behavior change rate and the environment temperature, and a supply demand is triggered in advance; a resource health degree gap and a replenishment station distance penalty term are fused to generate robot dog individual priorities, replenishment tasks are dynamically distributed according to the priorities, and replenishment station congestion is avoided through a two-stage load balancing strategy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of autonomous robot cluster control, and particularly relates to a seamless disaster relief material supply method for an autonomous fire extinguishing robot dog cluster based on dynamic load triggering. BACKGROUND

[0002] The existing disaster relief robot system has the following defects:

[0003] Resource management is static, and task output resources (such as fire extinguishing agents) and action resources (such as battery life) are not distinguished, resulting in resource depletion or supply congestion;

[0004] The environmental adaptability is poor, and real-time states such as fire changes and robot dog damage risks are not dynamically responded, which easily leads to task interruption;

[0005] The cluster cooperation efficiency is low, and there is a lack of multi-priority triggering mechanism and supply station load balancing strategy, which easily causes multi-robot dog congestion or key task delay. SUMMARY

[0006] The application is to solve the problems existing in the prior art, and provides a seamless disaster relief material supply method for an autonomous fire extinguishing robot dog cluster based on dynamic load triggering. Through double resource type (action resource and task output resource) coupling management, a multi-level threshold triggering mechanism and a dynamic load balancing strategy, the application realizes the continuous optimization of the action ability and task output ability of the robot dog cluster in a complex disaster scene.

[0007] The technical scheme adopted by the application has the following advantages:

[0008] A seamless disaster relief material supply method for an autonomous fire extinguishing robot dog cluster based on dynamic load triggering, comprising the following steps:

[0009] S1) Constructing a multi-resource health degree model, dividing robot dog resources into core output resources, auxiliary output resources and action survival resources, and calculating the health degree of each type of resource through weighting;

[0010] S2) Adjusting the resource weight coefficient in real time according to the fire level and the service life of the vulnerable parts;

[0011] S3) Defining a multi-level threshold triggering mechanism, according to the hierarchical response logic of core output resource triggering, action survival resource triggering and auxiliary output resource triggering, dynamically calculating the threshold value according to the fire change rate and the environmental temperature, and triggering the supply demand in advance;

[0012] S4) Generating the individual priority of the robot dog by fusing the resource health degree gap and the supply station distance penalty term, dynamically allocating the supply task according to the priority, and avoiding the supply station congestion through a two-level load balancing strategy.

[0013] Further, in S1), the calculation formula of the multi-resource health degree model is:

[0014]

[0015] H CTR is the core output resource health degree, with a value range of 0 to 1, representing the overall health condition of the core output resource;

[0016] p is the number of types of core output resources;

[0017] is the weight coefficient of the i-th core output resource, used to represent the relative importance of each resource in the overall health degree;

[0018] M i is the current remaining amount of the i-th core output resource; M imax is the maximum load of the i-th core output resource;

[0019] H ATR is the auxiliary output resource health degree, with a value range of 0 to 1, representing the overall health condition of the auxiliary output resource;

[0020] q is the number of types of auxiliary output resources;

[0021] is the weight coefficient of the j-th auxiliary output resource, used to represent the relative importance of each resource in the overall health degree;

[0022] M j is the current remaining amount of the j-th auxiliary output resource; M jmax is the maximum load of the j-th auxiliary output resource;

[0023] H SR is the action survival resource health degree, with a value range of 0 to 1, representing the overall health condition of the robot dog action survival resource;

[0024] is the battery life weight coefficient, with a default value of 0.7, but can be dynamically adjusted in specific environments; is the wear and tear life weight coefficient, with a default value of 0.3, but can be dynamically adjusted in specific environments;

[0025] E current is the current power; E max is the full power of the battery; L remaining is the remaining life of the wear and tear; L total is the total life of the wear and tear.

[0026] Further, in S2), the resource weight coefficient is adjusted in real time through a dynamic weight adjustment mechanism, including:

[0027] (1) When the remaining life L remaining <10 hours, the life weight coefficient of the vulnerable part is raised to a2=0.7;

[0028] (2) When the fire alarm level is upgraded to level three or above, the weight coefficient of the core output resource is adjusted, and the weight coefficients of the auxiliary output resource and the action survival resource are set to zero;

[0029] (3) When receiving manual intervention instructions, the central control terminal overrides local decision-making and locks resource allocation strategy.

[0030] Further, in S3), the threshold calculation formula of the core output resource triggering mechanism is:

[0031]

[0032] Where, T CTR，i is the triggering threshold of the i-th core output resource;

[0033] is the weight coefficient of the i-th core output resource, which represents the relative importance of each resource in the overall health degree;

[0034] M i,current is the current remaining amount of the i-th core output resource; M i,max is the maximum load of the i-th core output resource;

[0035] γ i is the fire change sensitivity coefficient, with a value range of [0, 0.2], which is obtained by historical fire data regression analysis. For example, in the forest fire scene, γ i = 0.15, and in the building fire scene, γ i = 0.08.

[0036] is the fire intensity change rate (unit: kW / s), which is obtained by edge computing through an infrared thermal imager in real time;

[0037] δ i is the environmental temperature compensation coefficient; unit: ℃ -1 , and the calculation formula is:

[0038] T ambient is the environmental temperature (unit: ℃), which is measured and obtained by the temperature and humidity sensor (SHT35, accuracy ±0.2℃) integrated by the robot dog.

[0039] represents the degree of resource use, and the less the resource, the larger this value.

[0040] represents the influence of fire intensity change on the threshold, the faster the fire intensity changes, the larger this value is.

[0041] δ i ·T ambient represents the influence of ambient temperature on the threshold, the higher the temperature, the larger this value is.

[0042] Further, the calculation formula of the fire intensity change rate is:

[0043]

[0044] wherein F(t) is the fire intensity at time t (unit: kW), and Δt is the sampling time interval (unit: seconds, default Δt=5 seconds).

[0045] Further, in S3), the condition for the action survival resource triggering mechanism is:

[0046]

[0047] When L remaining When L is 10 hours, the dynamic adjustment weight coefficient is adjusted to 0.7, and simultaneously α1 is dynamically adjusted to 1-α2, i.e. 0.3;

[0048] When the remaining life Lremaining of the vulnerable part is greater than 10 hours, the default weight coefficient is restored, i.e. α1=0.7 and α2=0.3;

[0049] When HSR<0.4, the action survival resource supply is triggered. The robot dog will preferentially return to charge or replace the vulnerable part.

[0050] Further, in S3), the condition for the auxiliary output resource triggering mechanism is:

[0051]

[0052] H ATR,j is the health degree of the jth auxiliary output resource, and the value range is 0 to 1. When the auxiliary output resource health degree H ATR,j <0.2 is triggered;

[0053] Only when the core output resource health degree H CTR >0.5 and the action survival resource health degree H SR >0.6 is triggered.

[0054] The triggering condition of the auxiliary output resource (such as a medical kit, a demolition tool, etc.) is designed as the lowest priority, so as to ensure that the core fire extinguishing task and the robot dog's own survival resource are guaranteed before considering triggering.

[0055] This design can avoid distraction to secondary tasks in resource-intensive or mission-critical stages, thereby improving overall task execution efficiency and safety.

[0056] Further, in S3), the hierarchical response logic is:

[0057] Rule 1: Core output resource trigger > action survival resource trigger > auxiliary output resource trigger;

[0058] Rule 2: If multiple core output resources are triggered at the same time, select the one with the largest trigger value for priority response;

[0059] Rule 3: Manual intervention instructions can override automatic decisions and force the locking of a certain resource supply. If necessary, manual intervention can override the automatic decisions of the system and force the locking and priority supply of a specific resource.

[0060] Further, in S4), the priority score calculation formula for each robot dog is:

[0061]

[0062] D i Euclidean distance from robot i to the supply station;

[0063] D max = the maximum effective working radius of the system, determined by the signal strength of the communication module;

[0064] λ, μ, ν: all are weight coefficients, determined by Monte Carlo simulation.

[0065] This formula unifies the multi-dimensional parameters of the robot dog (core resource health, survival resource health, distance to the supply station) into a priority score PS i , which is used to dynamically adjust the action strategy of the robot dog cluster to avoid congestion and improve efficiency.

[0066] In this way, the system can consider the core resource health, survival resource health, and distance to the supply station of each robot dog to calculate the priority score PS i of each robot dog, thereby effectively managing the action of the robot dog cluster, avoiding congestion, and improving overall efficiency.

[0067] Further, in S4), the two-level load balancing strategy is:

[0068] Primary allocation: all robot dogs default to the nearest supply station;

[0069] Dynamic allocation: for each robot dog, select the station with the lowest load rate and meet the distance tolerance threshold from the available supply stations:

[0070]

[0071] argmin denotes finding the parameter that makes the objective function take the minimum value, s.t. denotes being restricted by the condition;

[0072] D ik is the distance from the robot dog to the kth supply station;

[0073] D max is the maximum effective working radius of the robot dog;

[0074] θ is the distance tolerance threshold, and the calculation formula is: θ = θ base + η·PS i

[0075] wherein:

[0076] θ base is the basic distance tolerance;

[0077] η is the coefficient of the system;

[0078] PS i is the priority score of the robot dog, and the higher the priority, the farther but lower load site is allowed to be selected.

[0079] The beneficial effects of the present application are:

[0080] 1) The present application improves the response speed of resource supply by more than 30% through dynamic weight adjustment and threshold triggering, and reduces the waste rate of fire extinguishing agent by 25%; the environmental strong adaptability: the fire change rate (dF / dt), the life of the vulnerable part and other parameters participate in the decision in real time, and the task interruption rate is reduced by 40%; the two-level load balancing strategy makes the service efficiency of the supply station improved by 50%, and the key task delay rate tends to be zero.

[0081] 2) The multi-resource health degree model and multi-level threshold triggering mechanism enable the robot dog to make autonomous decisions based on its own resource state and environmental factors, improving the intelligent level of task execution.

[0082] 3) In a complex disaster scene, by dynamically adjusting resource allocation and supply strategy, the robot dog cluster can continue to operate stably and reduce the task failure caused by insufficient or overloaded resources. Artificial intervention instructions can override automatic decisions, forcibly lock a certain resource supply, and enhance the flexibility and controllability of the system in emergency situations.

[0083] 4) In a disaster scene, the robot dog cluster can autonomously execute tasks, reducing the time and risk of rescue personnel directly exposed to dangerous environments. By optimizing resource management and task scheduling, the robot dog cluster can more efficiently execute tasks in the disaster site, improving the overall rescue efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0084] Figure 1 Flowchart of the present application. DETAILED DESCRIPTION

[0085] The application will be further described below with reference to the accompanying drawings. Figure 1 The application will be further described below with reference to the accompanying drawings.

[0086] A seamless disaster relief material supply method based on dynamic load triggering autonomous fire extinguishing robot dog cluster, comprising the following steps:

[0087] S1) Construct a multi-resource health degree model, divide the robot dog resources into core output resources (CTR, such as fire extinguishing agent), auxiliary output resources (ATR, such as medical kit) and action survival resources (SR, such as battery life, life of vulnerable parts), and calculate the health degree (value range 0-1) of each type of resource through weighted calculation;

[0088] S2) According to the fire level and the life of vulnerable parts, the resource weight coefficient is adjusted in real time;

[0089] Dynamic weight adjustment: according to the fire alarm level, the life of vulnerable parts and other environmental parameters, the resource weight coefficient is adjusted in real time (such as increasing the weight of the life of vulnerable parts to 0.7 in high temperature environment), to ensure that the resource allocation is strongly coupled with the disaster scene.

[0090] The above fire alarm level is usually divided into different levels according to the severity of the fire, the influence range and the rescue difficulty, specifically including:

[0091] First-level fire alarm: characteristics: no casualties or trapped, ordinary building fire with small burning area, or fire involving live equipment / lines.

[0092] Second-level fire alarm: characteristics: a small number of casualties or trapped, ordinary building fire with large burning area, or small area fire in high-rise buildings, underground spaces, and densely populated places.

[0093] Third-level fire alarm: characteristics: a small number of casualties or trapped, and the fire occurs in a high-risk place (such as flammable and explosive goods warehouse, important facilities, etc.).

[0094] Fourth-level fire alarm: characteristics: more casualties or trapped, or large area fire in high-risk places.

[0095] Fifth-level fire alarm: characteristics: a large number of casualties or trapped, or the fire in high-risk places spreads rapidly and is difficult to control.

[0096] The fire alarm level can also be set according to user's own standards.

[0097] S3) Define a multi-level threshold triggering mechanism, according to the hierarchical response logic of core output resource triggering, action survival resource triggering, and auxiliary output resource triggering, dynamically calculate the threshold combining the fire rate of change and the environmental temperature, and trigger the supply demand in advance.

[0098] Define the hierarchical response logic of CTR trigger>SR trigger>ATR trigger, dynamically calculate the threshold value (such as CTR threshold value containing fire sensitive item γ i ·dF / dt) combined with parameters such as fire rate of change (dF / dt), ambient temperature (T ambient ) and other parameters to trigger the supply demand in advance. Support manual intervention channel, forcibly lock resource allocation strategy (such as forcibly allocate all robot dogs to priority supply of extinguishing agent when fire explosion) through control end.

[0099] S4) Generate robot dog individual priority by fusing resource health gap and supply station distance penalty term, and dynamically allocate supply task according to priority, and avoid supply station congestion through two-level load balancing strategy.

[0100] Priority score (PS i ): Generate robot dog individual priority by fusing resource health gap and supply station distance penalty term (D i / D max ), and dynamically allocate supply task.

[0101] Two-level load balancing: default selection of nearest supply station; when the load rate of a supply station is more than 80%, redirect to low-load station according to priority score and distance tolerance threshold (θ) to avoid cluster congestion.

[0102] The application adopts a control architecture combining distribution and centralization. Each robot dog has local decision-making ability and can make autonomous decisions according to its own sensor data and preset algorithm. At the same time, the system is equipped with a central control end for unified scheduling and control of the robot dog cluster in specific situations (such as emergency tasks, fire escalation, etc.). The central control end can override the local decision of the robot dog to ensure efficient cooperation of the entire cluster in complex disaster scenarios.

[0103] (I) Multi-resource health model

[0104] 1. Health calculation formula and parameter definition

[0105] According to the task scene, dynamically divide the resource level:

[0106] Core output resource (H CTR ) health: carry the total health of main resources (such as extinguishing agent, fire blanket, etc. in fire extinguishing task).

[0107] Auxiliary output resource (H ATR ) health: carry the total health of secondary resources (such as medical kit, demolition tool).

[0108] Action survival resource (H SRHealth: resources to maintain the ability of the robot dog to move (battery level, life of consumables)

[0109] The health changes as the task progresses, the state of the robot dog itself and the state of the resources it carries.

[0110] The calculation formula of the multi-resource health model is:

[0111]

[0112] H = H + H + H CTR H is the health of the core output resource, with a value range of 0 to 1, indicating the overall health of the core output resource;

[0113] H ATR is the health of the auxiliary output resource, with a value range of 0 to 1, indicating the overall health of the auxiliary output resource;

[0114] H SR is the health of the action survival resource, with a value range of 0 to 1, indicating the overall health of the action survival resource of the robot dog;

[0115] p is the number of types of core output resources; q is the number of types of auxiliary output resources;

[0116] is the weight coefficient of the i-th core output resource, used to represent the relative importance of each resource in the overall health; is the weight coefficient of the j-th auxiliary output resource, used to represent the relative importance of each resource in the overall health;

[0117] Example: fire extinguishing agent in a fire scene Medical kit

[0118] M i is the current remaining amount of the i-th core output resource; M j is the current remaining amount of the j-th auxiliary output resource; the multi-modal sensor array carried by the robot dog is used to collect (with an accuracy of ± 2%) in real time.

[0119] M imax is the maximum load of the i-th core output resource; M jmax is the maximum load of the j-th auxiliary output resource; the robot dog can simultaneously load multiple types of resources, or it can only load one type of resource. Therefore, H CTR , H ATR may exist at the same time, or only H CTR may exist.

[0120] E currentEcurrent is the current power (unit: kWh) provided by the BMS battery management system, with an accuracy of ±1%.

[0121] E max Efull is the full power of the battery, and the full power (e.g., 20 kWh) is determined by the battery specification.

[0122] E remaining Ltotal is the total life of the consumable, which is calculated based on laboratory data.

[0123] E total Ltotal is the total life of the consumable. In complex environments, the robot dog may have consumables, such as non-flame-retardant materials on the outside of the robot dog that may be disabled due to high temperatures in fire scenarios. The life of the consumable in real tasks needs to be considered, rather than the life in normal conditions. The life data of the consumable in complex environments comes from laboratory data, and then the control system is preset. In actual situations, even if the maximum life is reached, it can often be run for a period of time. Therefore, Ltotal is not directly used as a threshold here.

[0124] α1 is the battery life weight coefficient, with a default value of 0.7, but it can be dynamically adjusted in specific environments; α2 is the consumable life weight coefficient, with a default value of 0.3, but it can be dynamically adjusted in specific environments.

[0125] The task phase is dynamically adjusted, such as when the robot dog is in a harsh environment in a fire scene, the consumable life is <10h, and the system defaults α2 to 0.7. A sudden explosion causes the fire to intensify, leading to more serious consequences, and the control end artificially increases a2, forcing the robot dog cluster to go to the scene at all costs to output the carried resources (such as fire extinguishing agent).

[0126] If the robot dog has a self-generating function to spray fire extinguishing agent, the consumption of the fire extinguishing agent spray is included in Ecurrent.

[0127] The above parameters are calculated by the robot dog's own sensors at a second level and the data is returned to the control end.

[0128] II. Dynamic weight adjustment mechanism

[0129] The dynamic weight adjustment mechanism adjusts the resource weight coefficient in real time, including:

[0130] (1) When the remaining life of the consumable L remaining <10 hours, increase the consumable life weight coefficient to α2 = 0.7;

[0131] (2) When the fire alarm level is upgraded to level three or above, adjust the weight coefficient of the core output resource, and set the weight coefficients of the auxiliary output resource and the action survival resource to zero;

[0132] (3) When receiving the manual intervention instruction, the central control terminal overrides the local decision and locks the resource allocation strategy.

[0133] III. Multi-threshold triggering mechanism

[0134] This mechanism divides the threshold system into three levels: task output layer (output resource health), action survival layer (action resource health), and cluster coordination layer (supply station load rate), forming a three-level linkage triggering logic. Each level contains an independent threshold calculation model, and cross-level decision-making is achieved through dynamic weights and coupling constraints.

[0135] 1. Core output resource (CTR) triggering mechanism

[0136] For each type of core resource, set a threshold T CTR，i :

[0137]

[0138] where T CTR，i is the triggering threshold of the i-th core output resource;

[0139] is the weight coefficient of the i-th core output resource, representing the relative importance of each resource in the overall health;

[0140] M i,current is the current remaining amount of the i-th core output resource; M i,max is the maximum load of the i-th core output resource;

[0141] γ i is the fire change sensitivity coefficient, with a value range of [0, 0.2], obtained through historical fire data regression analysis. For example, γ i = 0.15 in the forest fire scenario, and γ i = 0.08 in the building fire scenario.

[0142] is the fire intensity change rate (unit: kW / s), obtained by the infrared thermal imager through edge computing in real time;

[0143] The calculation formula of T

[0144]

[0145] where F(t) is the fire intensity at time t (unit: kW), and Δt is the sampling time interval (unit: seconds, default Δt = 5 seconds).

[0146] δ i is the environmental temperature compensation coefficient; the unit is ℃ -1, the calculation formula is:

[0147] T ambient The ambient temperature (unit: ℃) is measured by the temperature and humidity sensor (SHT35, accuracy ±0.2℃) integrated by the robot dog.

[0148] Indicates the degree of resource usage, the less the resource, the greater the value.

[0149] Indicates the influence of fire change on threshold, the faster the fire change, the greater the value.

[0150] δ i ·T ambient Indicates the influence of ambient temperature on threshold, the higher the temperature, the greater the value.

[0151] In actual scenarios, it is not necessary to wait for the core resource to be output before the robot dog returns to the supply station, because the use of the plug-in supply station also needs to be considered, and whether the robot dog needs to be persisted in the field fire situation and the like. Therefore, a dynamic threshold is set here, and it is linked with the strategy adjustment mechanism in the subsequent chapters.

[0152] 2. Action survival resources (SR)

[0153] The condition for the action survival resource trigger mechanism is:

[0154]

[0155] When L remaining >10 hours, the dynamic adjustment weight coefficient is adjusted to 0.7, and α1 is dynamically adjusted to 1-α2, that is, 0.3;

[0156] When the remaining life L remaining >10 hours, the default weight coefficient is restored, that is, α1=0.7 and α2=0.3;

[0157] When HSR<0.4, the action survival resource supply is triggered. When triggered, the robot dog returns to charge or replace the vulnerable parts first, and the priority is lower than that of the CTR fire extinguishing agent trigger. L remaining Calculated by the life prediction module carried by the robot dog.

[0158] 3. Auxiliary output resources (ATR)

[0159] The condition for the auxiliary output resource trigger mechanism is:

[0160]

[0161] H ATR,jThe health degree of the jth auxiliary output resource, with a value range of 0 to 1, is triggered when the auxiliary output resource health degree H ATR,j <0.2;

[0162] The auxiliary output resource health degree H CTR >0.5 and the action survival resource health degree H SR >0.6.

[0163] The trigger condition of the auxiliary output resource (such as a medical kit, a demolition tool, etc.) is designed as the lowest priority, so as to ensure that the core fire extinguishing task and the survival resource of the robot dog are guaranteed before triggering.

[0164] This design can avoid dispersing attention to secondary tasks in a resource-intensive or task-critical stage, thereby improving the overall task execution efficiency and safety.

[0165] The trigger condition of the auxiliary output resource (such as a medical kit, a demolition tool, etc.) is designed as the lowest priority, so as to ensure that the core fire extinguishing task and the survival resource of the robot dog are guaranteed before triggering.

[0166] The present application does not further design the trigger of the ATR. It is also possible that when H CTR >0.5 but H SR ≤0.4, the ATR trigger condition is covered by the SR trigger. Similarly, this is not included in the design of the present application. Only if it is covered, it is covered.

[0167] 4. Hierarchical trigger condition

[0168] The hierarchical response logic is as follows:

[0169] Rule 1: Core output resource trigger > action survival resource trigger > auxiliary output resource trigger;

[0170] Rule 2: If multiple core output resources are triggered at the same time, the one with the largest trigger value is selected for priority response;

[0171] Rule 3: Manual intervention instructions can override automatic decision-making and forcibly lock a certain resource supply. In necessary cases, manual intervention can override the automatic decision-making of the system and forcibly lock and prioritize a certain specific resource.

[0172] Four, strategy adjustment mechanism

[0173] 4.1 Robot dog cluster strategy mechanism

[0174] The control background needs to adjust the robot dog cluster action to ensure that all robot dogs do not concentrate at the fire scene or the supply station at the same time. The cluster state matrix Q(t) and the supply station service capacity constraint are defined, and multiple dimensions are unified into a single priority score:

[0175]

[0176] wherein: D i is the Euclidean distance from the robot i to the supply station;

[0177] D max is the maximum effective working radius of the system, determined by the signal strength of the communication module;

[0178] λ, μ, ν: are weight coefficients, determined by Monte Carlo simulation.

[0179] This formula unifies the multi-dimensional parameters of the robot dog (core resource health, survival resource health, distance to the supply station) into a priority score PS i , which is used to dynamically adjust the action strategy of the robot dog cluster to avoid congestion and improve efficiency.

[0180] In this way, the system can consider the core resource health, survival resource health, and distance to the supply station of each robot dog, calculate the priority score PS i of each robot dog, and effectively manage the action of the robot dog cluster to avoid congestion and improve overall efficiency.

[0181] 4.2 Supply station load balancing strategy:

[0182] Avoid congestion through a two-level shunting mechanism, replacing complex optimization models. Primary distribution, all robot dogs default to the nearest supply station. Real-time overload detection begins after the action starts, and the maximum service capacity Ck of each supply station k is determined by hardware performance, for example, a certain supply station can serve a maximum of 9 robot dogs per hour, so Ck = 9 units / hour. When the queue number Nk of a certain supply station reaches 80% of the capacity limit Ck, dynamic distribution is started. This is the trigger mechanism of the supply station, not the trigger mechanism of the robot dog, so it is single.

[0183] For each robot dog, select the station with the lowest load rate and that satisfies (θ is the distance tolerance threshold) from the available supply stations. D ik is the distance from the robot dog to the kth supply station.

[0184]

[0185] argmin represents finding the parameter that minimizes the objective function, that is, among all possible parameters k, finding the k value that minimizes the ratio, which is the optimal solution k*, s.t. represents the condition under the constraint.

[0186] D ik is the distance from the robot dog to the kth supply station;

[0187] D max The maximum effective working radius of the robot dog.

[0188] θ is the distance tolerance threshold, and the calculation formula is: θ = θ base + η·PS i

[0189] Wherein:

[0190] θ base is the basic distance tolerance (such as 0.3, allowing to bypass 30% of the maximum radius);

[0191] η is the coefficient of the joint, used to control the elasticity of the distance tolerance threshold.

[0192] PS i is the priority score of the robot dog, the higher the priority, the more distant but low-load sites are allowed to be selected.

[0193] Adjustment of the distance tolerance threshold:

[0194] η = 0: pure distance priority strategy, the robot dog only selects the nearest supply station.

[0195] η > 0.5: significantly biased towards load balancing, the robot dog is more inclined to select a low-load supply station, even if the distance is slightly far.

[0196] The above is only the preferred embodiment of the present application, it should be noted that for those skilled in the art, without departing from the principles of the present application can also make a number of improvements, these improvements should also be considered within the scope of the present application.

Claims

1. A method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering, characterized by: The steps include: S1) Construct a multi-resource health model, divide the robot dog resources into core output resources, auxiliary output resources, and action survival resources, and calculate the health of each resource type through weighted calculation; S2) Adjust resource weight coefficients in real time based on fire severity and life of wearing parts; S3) Define a multi-level threshold trigger mechanism, with a hierarchical response logic based on core output resource triggering, action survival resource triggering, and auxiliary output resource triggering. Combined with the fire change rate and ambient temperature, dynamically calculate thresholds to trigger supply needs in advance. S4) Integrate the resource health gap and the supply station distance penalty to generate the individual priorities of the robot dogs, dynamically allocate supply tasks according to the priority, and avoid congestion at the supply station through a two-level load balancing strategy.

2. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 1, characterized in that: In S1), the calculation formula of the multi-resource health model is: Among them: H CTR Outputs resource health for the core, with a value range of 0 to 1; p is the number of types of core output resources; is the weight coefficient of the i-th core output resource; M i is the current remaining amount of the i-th core output resource; M imax The maximum load of the i-th core output resource; H ATR To assist in outputting resource health, the value range is 0 to 1; q is the number of types of auxiliary output resources; is the weight coefficient of the j-th auxiliary output resource; M j is the current remaining amount of the j-th auxiliary output resource; M jmax The maximum load capacity of the jth auxiliary output resource; H SR The health of action survival resources ranges from 0 to 1; α1 is the weight coefficient of battery life; α2 is the weight coefficient of wearing parts life; E current is the current power; E max The battery is fully charged; L remaining L is the remaining life of wearing parts; total is the total life of wearing parts.

3. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 2, characterized in that: In S2), the resource weight coefficient is adjusted in real time through a dynamic weight adjustment mechanism, including: (1) When the remaining life of the wearing parts is L remaining When the service life is less than 10 hours, the weight coefficient of the wearing parts is increased; (2) When the fire alarm level is upgraded to level 3 or above, the weight coefficient of the core output resource is adjusted, and the weight coefficients of the auxiliary output resource and action survival resource are reset to zero; (3) When receiving manual intervention instructions, the central control end overrides the local decision and locks the resource allocation strategy.

4. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 2, characterized in that: In S3), the threshold calculation formula for the core output resource trigger mechanism is: Among them, T CTR，i is the trigger threshold of the core output resource of type i; M i,current is the current remaining amount of core output resources of type i; M i,max is the maximum load of the core output resource of type i; γ i is the fire intensity change sensitivity coefficient, with a value range of [0,0.2]; is the rate of change of fire intensity; δ i is the ambient temperature compensation coefficient; T ambient is the ambient temperature.

5. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 4, characterized in that: The calculation formula for the fire intensity change rate is: Where F(t) is the fire intensity at time t, and Δt is the sampling time interval.

6. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 4, characterized in that: In S3), the conditions for triggering the action survival resource mechanism are: When L remaining When the time is less than 10 hours, the weight coefficient is dynamically adjusted, α2 is adjusted to 0.7, and α1 is dynamically adjusted to 1-α2, that is, 0.3; When the remaining life of the wearing parts Lremaining>10 hours, the default weight coefficient is restored, that is, α1=0.7, α2=0.3; When HSR<0.4, the action of survival resource replenishment is triggered.

7. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 4, characterized in that: In S3), the conditions for the auxiliary output resource triggering mechanism are: H ATR,j is the health of the j-th auxiliary output resource, ranging from 0 to 1. ATR,j Triggered when <0.2; Output resource health H only in the core CTR >0.5 and the health of action survival resources H SR Triggered when >0.

6.

8. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 4, characterized in that: In S3), the hierarchical response logic is: Rule 1: Core output resource trigger > Action survival resource trigger > Auxiliary output resource trigger; Rule 2: If multiple core output resources are triggered simultaneously, the one with the largest trigger value is selected for priority response; Rule 3: Manual intervention instructions can override automatic decisions and forcibly lock a certain resource supply.

9. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 1, characterized in that: In S4), the priority score calculation formula for each robot dog is: D i is the Euclidean distance from robot dog i to the supply station; D max =The maximum effective operating radius of the system, determined by the signal strength of the communication module; λ, μ, ν: are weight coefficients, determined by Monte Carlo simulation.

10. The method for seamlessly replenishing disaster relief supplies using a swarm of autonomous fire-fighting robot dogs based on dynamic load triggering as claimed in claim 1, characterized in that: In S4), the two-level load balancing strategy is: Primary allocation: All robot dogs choose the nearest supply station by default; Dynamic allocation: For each robot dog, select the station with the lowest load rate and meeting the distance tolerance threshold from the available supply stations: argmin means finding the parameters that minimize the objective function, and st means being restricted by conditions; D ik is the distance from the robot dog to the kth supply station; D max is the maximum effective operating radius of the robot dog; θ is the distance tolerance threshold, and the calculation formula is: θ=θ base +η·PS i in: θ base Tolerance for basic distance; η is the node coefficient; PS i is the priority score of the robot dog. The higher the priority, the farther away but less loaded stations can be selected.

Images