An emergency resource scheduling optimization system and method based on artificial intelligence

Abstract

The application discloses an emergency resource scheduling optimization system and method based on artificial intelligence, and relates to the technical field of resource scheduling optimization.The dynamic concurrent capacity is obtained according to the regional layout parameters of an emergency resource receiving area and candidate vehicle size parameters, and the node flow rate is obtained in combination with the buffer area parameter and the handling flow parameter.The real-time availability index is obtained according to the entrance queuing state and the exit traffic state of the emergency resource receiving area.The dynamic departure quota is obtained according to the node flow rate and the real-time availability index.The material demand is truncated and the vehicle type is matched according to the dynamic departure quota, and the constrained departure task set is obtained.The space nesting sorting is performed on the vehicles in the vehicle waiting area, and the vehicle sequence is obtained.The target vehicle is extracted from the vehicle sequence according to the dynamic departure quota, and the time sequence release instruction is obtained.The target vehicle is controlled to drive into the emergency resource receiving area in time sharing mode to perform the unloading operation according to the real-time availability index.

Description

Technical Field

[0001] This invention relates to the field of resource scheduling optimization technology, specifically an emergency resource scheduling optimization system and method based on artificial intelligence. Background Technology

[0002] In emergency material dispatch, the emergency resource receiving area, as the end node of the transportation chain, has a direct impact on the overall dispatch efficiency in terms of its capacity and circulation efficiency.

[0003] In existing technologies, the capacity of emergency resource receiving areas is usually set as a fixed value or based on area estimation, which makes it difficult to dynamically adjust according to the structural differences of the actual arriving vehicles, leading to a deviation between the scheduling plan and the actual capacity limit.

[0004] In assessing the availability of emergency resource receiving areas, existing technologies typically monitor the degree of congestion within the facility and the unobstructed flow at exits separately, then combine them into a single indicator. This parallel synthesis method implicitly assumes that congestion within the facility and unobstructed flow at exits are independent, and rarely considers the physical transmission relationship where queuing vehicles extend into the exit lanes, crowding out space and preventing unloaded vehicles from leaving. When congestion spreads from the inside out, there may be a situation where there is still room at the entrance but the exit is blocked. Existing evaluation indicators may still classify this as available, thus allowing vehicles to continue to depart.

[0005] In terms of departure planning, existing solutions are usually generated once based on the initial state. When subsequent vehicles drive to the emergency resource receiving area according to the predetermined plan, if the state of the emergency resource receiving area changes due to the delay of the preceding vehicles, the system will find it difficult to dynamically adjust the departure rhythm according to the real-time availability. Vehicles may not be able to enter after arriving at the entrance and can only queue at the entrance, resulting in secondary congestion. Summary of the Invention

[0006] The purpose of this invention is to provide an emergency resource scheduling optimization system and method based on artificial intelligence to solve the problems raised in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides an emergency resource scheduling optimization method based on artificial intelligence, comprising: Obtain the regional layout parameters and candidate vehicle size parameters of the emergency resource receiving area, calculate the spatial adaptability, and generate the dynamic concurrent capacity of the emergency resource receiving area; obtain the buffer area parameters and transportation flow parameters of the emergency resource receiving area, and combine them with the dynamic concurrent capacity to obtain the node flow rate of the emergency resource receiving area; Real-time monitoring of the real-time accumulation and departure congestion status of the emergency resource receiving area, generating a real-time availability index for the emergency resource receiving area; Based on the node throughput rate and the real-time availability index, the dynamic dispatch quota for the current scheduling period is obtained; Obtain the material demand sequence, and based on the dynamic dispatch quota and the candidate vehicle size parameters, truncate and perform shape matching on the material demand sequence to obtain a constrained dispatch task set; distribute the constrained dispatch task set to the vehicle waiting area, and perform spatial nesting sorting on the vehicles to be dispatched in the vehicle waiting area based on the candidate vehicle size parameters to generate a vehicle waiting sequence. Based on the dynamic vehicle dispatch quota, a corresponding number of target vehicles are extracted from the vehicle queue to be dispatched, and a timed release instruction is generated for the target vehicles. The timed release instruction is sent to the target vehicles, and the target vehicles are controlled to leave the vehicle waiting area in time slots and enter the emergency resource receiving area to perform unloading operations.

[0008] In conjunction with the first aspect, in the first implementation of the first aspect of this application, the step of obtaining the regional layout parameters and candidate vehicle size parameters of the emergency resource receiving area, calculating the spatial adaptability, and generating the dynamic concurrent capacity of the emergency resource receiving area includes: Obtain the regional layout parameters of the emergency resource receiving area, including the effective operating area A. e and loading / unloading aisle width W c Obtain the length, width, minimum turning radius, and loading volume of each candidate vehicle type as candidate vehicle size parameters, where the length of the i-th candidate vehicle is L. i Width is B i The minimum turning radius is R i The loading volume is V i Where i is the type number of the candidate vehicle; According to the length L i and width B i The single-vehicle operating area S of the i-th candidate vehicle is obtained. i According to the width W of the loading and unloading channel c and the minimum turning radius R i The channel adaptation coefficient C of the i-th type of candidate vehicle is obtained. i Remove the channel adaptation coefficient C. i Candidate vehicle types with a value less than 1; Obtain the proportion of each type of candidate vehicle, where the proportion of the i-th type of candidate vehicle is P. i According to the effective working area A e The aforementioned proportion P i The single-vehicle operating area S iand the channel adaptation coefficient C i The dynamic concurrent capacity C of the emergency resource receiving area is obtained. p The formula is: .

[0009] In conjunction with the first aspect, in the second implementation of the first aspect of this application, obtaining the buffer area parameter and transport flow parameter of the emergency resource receiving area, and combining them with the dynamic concurrency capacity, to obtain the node flow rate of the emergency resource receiving area includes: Get the duration T of a single unload job u and the time T for the vehicle to leave after unloading d Calculate the average operating turnaround time T for a single vehicle. t According to the dynamic concurrency capacity C p and the average operating turnaround time T of a single vehicle t The node throughput R of the emergency resource receiving area is obtained. n .

[0010] In conjunction with the first aspect, in the third implementation of the first aspect of this application, the real-time monitoring of the real-time accumulation and departure congestion status of the emergency resource receiving area, and the generation of a real-time availability index for the emergency resource receiving area, includes: Real-time monitoring of the queue length L of vehicles waiting to unload at the entrance of the emergency resource receiving area. q (t) and the number of vehicles N currently unloading. u (t), combined with the dynamic concurrency capacity C p The in-field backlog P is obtained. q (t), where t is the current time; based on the in-field accumulation degree P q From (t), we derive the remaining capacity rate R(t) in the field, which is given by the formula R(t) = 1 - P. q (t); Real-time monitoring of the average passing speed v of vehicles leaving the emergency resource receiving area exit. d (t), to obtain the free-flow velocity v of the vehicle leaving the passage. f The exit velocity hold rate s(t) is obtained, and the formula is s(t) = v d (t) / v f ; Based on the remaining capacity rate R(t) in the field and the exit velocity retention rate s(t), calculate the actual availability of the exit μ. d (t), the formula is μ d (t) = s(t) × R(t); Based on the remaining capacity rate R(t) in the field and the actual availability of the outlet μ d(t), generate the real-time availability index α(t) of the emergency resource receiving area, with the formula α(t) = R(t) × μ d (t).

[0011] In conjunction with the first aspect, in the fourth implementation of the first aspect of this application, the step of obtaining the dynamic dispatch quota for the current scheduling period based on the node flow rate and the real-time availability index includes: Get the duration T of the current scheduling period p Combined with the node flow rate R n The theoretical departure quota N0 is obtained; the theoretical departure quota N0 is corrected according to the real-time availability index α(t) to obtain the dynamic departure quota N(t) for the current scheduling period.

[0012] In conjunction with the first aspect, in the fifth implementation of the first aspect of this application, the step of obtaining the material demand sequence, based on the dynamic dispatch quota and the candidate vehicle size parameters, involves truncating and morphologically matching the material demand sequence to obtain a constrained dispatch task set, including: Obtain the material demand sequence M = {m1, m2, ..., m k}, where k is the total number of batches of goods to be transported, m j The required loading volume for the jth batch of supplies; Using the dynamic dispatch quota N(t) as the upper limit, the material demand sequence is truncated, and the first K batches of materials are truncated, with the formula K=min(k,N(t)), where K is the actual number of batches of materials truncated; after truncation, the material subsequence to be matched is obtained as M'={m1,m2,...,m K}; For the cut-off supplies m j Based on the loading volume V in the candidate vehicle size parameters i Calculate the loading morphology matching degree G for each candidate vehicle type. ij ; Select the loading form matching degree G ij Greater than or equal to 1 and G ij The smallest candidate vehicle type as material m j Matching vehicle type v j This forms a constrained departure task set T = {(m j ,v j )|j=1,2,...,K}.

[0013] In conjunction with the first aspect, in the sixth implementation of the first aspect of this application, the step of distributing the constrained departure task set to the vehicle waiting area, and spatially nesting and sorting the vehicles to be dispatched in the vehicle waiting area based on the candidate vehicle size parameters to generate a sequence of vehicles to be dispatched includes: Obtain all vehicles and their corresponding vehicle types within the vehicle waiting area, and determine the length L of each vehicle based on the candidate vehicle size parameters. i Width B i and loading volume V i ; All vehicles within the vehicle waiting area are subjected to spatial nesting sorting, wherein the first layer is based on the loading volume V. i Arranged in descending order, the second layer is based on the first layer according to the length L. i Arranged in ascending order, the third layer is based on the second layer according to the width B mentioned above. i The vehicles are arranged in alternating odd and even order; the sorted vehicles are matched with the constrained departure task set T to generate a sequence of vehicles to be dispatched, S.

[0014] In conjunction with the first aspect, in the seventh implementation of the first aspect of this application, the step of extracting a corresponding number of target vehicles from the queue of vehicles to be dispatched based on the dynamic dispatch quota, and generating a timing release instruction for the target vehicles, includes: Extract the aforementioned dynamic departure quota N(t) vehicles from the vehicle sequence S to be dispatched as target vehicles, denoted as s1, s2, ..., s N(t) The vehicle identifier of the r-th target vehicle is s. r r is the target vehicle number, r = 1, 2, ..., N(t); obtain the length L of each target vehicle. i Calculate the minimum safe time interval Δt between two adjacent target vehicles. s ; The real-time availability index α(t) and exit speed maintenance rate s(t) at the current moment are input into a pre-trained three-layer BP neural network model, which outputs a normalized offset β0; based on the normalized offset β0 and the minimum safe time interval Δt... s The departure fine-tuning offset β of the r-th target vehicle is obtained. r ; With the minimum safe time interval Δt s The departure fine-tuning offset β is added at intervals. r The release time t for the r-th target vehicle is obtained. r ; According to the vehicle identification s of each target vehicle r and release time t r Generate timing release instruction set Z = {(s r ,tr )|r=1,2,…,N(t)}.

[0015] In conjunction with the first aspect, in the eighth implementation of the first aspect of this application, the step of sending the timing release instruction to the target vehicle and controlling the target vehicle to leave the vehicle waiting area in stages and enter the emergency resource receiving area to perform unloading operations includes: Each timing release instruction in the timing release instruction set Z is sent to the corresponding target vehicle in a time-sharing manner; each target vehicle is released at its corresponding release time t. r The vehicle starts and travels along the designated path towards the emergency resource receiving area. When the target vehicle arrives at the entrance of the emergency resource receiving area, the real-time availability index α(t) is obtained. When α(t) is greater than or equal to the preset entry availability threshold, the target vehicle is allowed to enter the emergency resource receiving area to perform unloading operations. When α(t) is less than the preset entry availability threshold, the target vehicle is controlled to wait in the entrance waiting area until α(t) is greater than or equal to the preset entry availability threshold before it is allowed to enter.

[0016] Secondly, the present invention provides an emergency resource scheduling optimization system based on artificial intelligence, comprising: The basic capacity and flow calculation module includes a parameter acquisition unit and a capacity calculation unit. The parameter acquisition unit acquires the regional layout parameters of the emergency resource receiving area, the candidate vehicle size parameters, the duration of a single unloading operation, and the vehicle departure time after unloading. The capacity calculation unit calculates the dynamic concurrent capacity based on the regional layout parameters of the emergency resource receiving area and the candidate vehicle size parameters, and calculates the node flow rate based on the duration of a single unloading operation and the vehicle departure time after unloading. The real-time status monitoring and availability evaluation module includes a status monitoring unit and an availability calculation unit. The status monitoring unit monitors in real time the queue length of vehicles waiting to be unloaded at the entrance of the emergency resource receiving area and the number of vehicles currently unloading, as well as the average passing speed of vehicles leaving the exit. The availability calculation unit recursively generates a real-time availability index based on dynamic concurrent capacity, queue length of vehicles waiting to be unloaded, number of vehicles currently unloading, average passing speed of vehicles leaving the exit, and free flow speed. The scheduling quota and task matching module includes a quota generation unit and a task matching unit. The quota generation unit generates dynamic dispatch quotas based on node flow rate and real-time availability index. The task matching unit truncates the material demand sequence based on the dynamic dispatch quotas and matches candidate vehicle types for each batch of materials after truncation, generating a constrained dispatch task set. Vehicle sorting and instruction generation module: includes a sorting unit and an instruction generation unit; wherein, the sorting unit performs spatial nested sorting of vehicles to be dispatched in the vehicle waiting area to generate a sequence of vehicles to be dispatched; the instruction generation unit extracts target vehicles from the sequence of vehicles to be dispatched according to the dynamic dispatch quota and generates a time-sequenced release instruction. The execution control module includes an execution control unit and an entry verification unit. The execution control unit sends a timed release command to the target vehicle, controlling the target vehicle to enter the emergency resource receiving area in a timed manner to perform unloading operations. The entry verification unit controls whether to release the vehicle to enter the emergency resource receiving area based on the real-time availability index when the target vehicle arrives at the entrance.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. Based on the constraint relationship between the emergency resource receiving area space and the candidate vehicle size, this invention obtains the dynamic concurrent capacity and node flow rate that change dynamically with the vehicle model structure, thus solving the problem that the fixed capacity cannot adapt to changes in vehicle model structure, leading to the failure of scheduling decisions.

[0018] 2. This invention determines the remaining capacity within the facility based on the queuing status at the entrance, and the smoothness of the passageway based on the exit speed. It also uses the remaining capacity within the facility as a loss factor applied to the exit speed, thereby progressively generating a real-time availability index. This allows the crowding-out effect of the backlog within the facility on the exit to participate in the availability calculation, solving the problem in existing technologies that fail to identify the overall unavailability of the emergency resource receiving area due to the backlog of vehicles at the exit when combining the facility and the exit.

[0019] 3. This invention uses a real-time availability index to simultaneously constrain dynamic vehicle dispatch quotas and entry releases, so that both dispatch and entry are subject to reverse regulation by the same availability index. This solves the problem in existing technologies where dispatch plans deviate from the real-time status of the emergency resource receiving area, resulting in vehicles being unable to unload after arrival. Attached Figure Description

[0020] Figure 1 This is a schematic diagram illustrating the steps of an artificial intelligence-based emergency resource scheduling optimization method according to the present invention. Figure 2 This is a flowchart illustrating the steps involved in generating a real-time availability index for an artificial intelligence-based emergency resource scheduling optimization method according to the present invention. Figure 3 This is a system architecture diagram of an emergency resource scheduling optimization system based on artificial intelligence according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example: Figures 1-3 As shown, the present invention provides a technical solution. like Figure 1 A schematic diagram illustrating the steps of an artificial intelligence-based emergency resource scheduling optimization method is provided in this invention. The method includes: Step S100: Obtain the regional layout parameters and candidate vehicle size parameters of the emergency resource receiving area, calculate the spatial adaptability, and generate the dynamic concurrent capacity of the emergency resource receiving area; obtain the buffer area parameters and transportation flow parameters of the emergency resource receiving area, and combine them with the dynamic concurrent capacity to obtain the node flow rate of the emergency resource receiving area; Specifically, obtain the regional layout parameters of the emergency resource receiving area, including the effective operating area A. e and loading / unloading aisle width W c Obtain the length, width, minimum turning radius, and loading volume of each candidate vehicle type as candidate vehicle size parameters, where the length of the i-th candidate vehicle is L. i Width is B i The minimum turning radius is R i The loading volume is V i Where i is the type number of the candidate vehicle; According to the length L i and width B i The single-vehicle operating area S of the i-th candidate vehicle is obtained. i The formula is S i =L i ×B i The single-vehicle operating area S i This represents the actual area occupied by the i-th type of candidate vehicle when unloading in the emergency resource receiving area; According to the width W of the loading and unloading channel c and the minimum turning radius R i The channel adaptation coefficient C of the i-th type of candidate vehicle is obtained. i The formula is C i =W c / R i Remove the channel adaptation coefficient C. iCandidate vehicle types with a value less than 1 are removed, and the remaining proportions of each vehicle type are renormalized. The channel adaptation coefficient C is then used. i The channel adaptation coefficient C is used to determine whether the i-th type of candidate vehicle can turn and pass normally in the emergency resource receiving area. i A value less than 1 indicates that the i-th type of candidate vehicle cannot turn and pass normally in the emergency resource receiving area; Obtain the proportion of each type of candidate vehicle, where the proportion of the i-th type of candidate vehicle is P. i According to the effective working area A e The aforementioned proportion P i The single-vehicle operating area S i and the channel adaptation coefficient C i The dynamic concurrent capacity C of the emergency resource receiving area is obtained. p The formula is: ; The dynamic concurrency capacity C p This indicates the maximum number of vehicles that the emergency resource receiving area can accommodate at any given time.

[0023] Get the duration T of a single unload job u and the time T for the vehicle to leave after unloading d Calculate the average operating turnaround time T for a single vehicle. t The formula is T t =T u +T d The average operating turnaround time T for a single vehicle t This represents the average time required for a vehicle to begin unloading operations, complete unloading, and leave the emergency resource receiving area. According to the dynamic concurrency capacity C p and the average operating turnaround time T of a single vehicle t The node throughput R of the emergency resource receiving area is obtained. n The formula is R n =C p / T t The node flow rate R n This indicates the average number of vehicles that can complete unloading operations and release berths within a unit of time in the emergency resource receiving area, where a berth is an area in the emergency resource receiving area for vehicles to park and unload.

[0024] In one specific embodiment, in an emergency material dispatch scenario in a chemical industrial park, the effective working area A of the emergency resource receiving area is... e 480m 2 Loading and unloading aisle width W cThe length is 4.2m. The currently available candidate vehicle types for the rear assembly area include light vans and extended light vans. The light vans have a length L1 = 4.2m, width B1 = 1.8m, minimum turning radius R1 = 3.8m, and loading volume V1 = 16m³. 3 The proportion of P1=70%; the length of the extended minivan is L2=4.5m, the width is B2=1.9m, the minimum turning radius is R2=4.0m, and the loading volume is V2=20m³. 3 The percentage is P2=30%. Calculate the operating area per vehicle: S1=7.56m². 2 S2 = 8.55m 2 The channel adaptation coefficients are C1=1.10 and C2=1.05. Both vehicle types meet the passage condition C≥1, and their proportions remain unchanged. Calculate the dynamic concurrent capacity C. p =61 vehicles. Obtain the duration T of a single unloading operation. u =0.35h, the time T for the vehicle to leave after unloading d =0.15h, thus obtaining the average turnaround time T for a single vehicle. t =0.50h, node rotation rate R n =122 vehicles / hour.

[0025] Step S200: Monitor the real-time accumulation and departure congestion status of the emergency resource receiving area in real time, and generate the real-time availability index of the emergency resource receiving area; Specifically, the queue length L of vehicles waiting to unload at the entrance of the emergency resource receiving area is monitored in real time. q (t) and the number of vehicles N currently unloading. u (t), combined with the dynamic concurrency capacity C p The in-field backlog P is obtained. q (t), the formula is P q (t)=(L q (t)+N u (t)) / C p Where t is the current time; and the field backlog P q (t) represents the proportion of the total number of vehicles waiting to unload and vehicles currently unloading within the emergency resource receiving area to the dynamic concurrent capacity, P q The larger the value of (t), the closer the emergency resource receiving area is to saturation. According to the in-field backlog P q From (t), we derive the remaining capacity rate R(t) in the field, which is given by the formula R(t) = 1 - P. q (t); The remaining capacity rate R(t) in the field represents the proportion of the current field capacity that has not been occupied. The larger R(t) is, the more idle the emergency resource receiving area is. Real-time monitoring of the average passing speed v of vehicles leaving the emergency resource receiving area exit. d (t), to obtain the free-flow velocity v of the vehicle leaving the passage. f The exit velocity hold rate s(t) is obtained, and the formula is s(t) = v d (t) / v f The exit speed retention rate s(t) represents the degree to which the current passage speed of the exit channel is maintained relative to the free flow speed; the larger the s(t) is, the smoother the exit. Based on the remaining capacity rate R(t) in the field and the exit velocity retention rate s(t), calculate the actual availability of the exit μ. d (t), the formula is μ d (t)=s(t)×R(t); The actual availability of the export is μ d (t) represents the actual air discharge capacity of the outlet after being constrained by the backlog of the field; Based on the remaining capacity rate R(t) in the field and the actual availability of the outlet μ d (t), generate the real-time availability index α(t) of the emergency resource receiving area, with the formula α(t) = R(t) × μ d (t); The real-time availability index α(t) represents the current availability of the emergency resource receiving area.

[0026] In one specific embodiment, at the fifth scheduling cycle t=10:30, real-time status data of the emergency resource receiving area is collected, including the queue length L of vehicles waiting to unload at the entrance. q (t) = 18 vehicles, and the number of vehicles N currently unloading is... u (t) = 12 vehicles, the average speed v of vehicles leaving the exit. d (t) = 2.8 m / s, free-flow velocity v leaving the passage f =6.0m / s. Combined with dynamic concurrent capacity C p =61 vehicles, the calculated backlog P in the yard q R(t) = 0.492, remaining capacity rate within the site R(t) = 0.508, exit velocity maintenance rate s(t) = 0.467, actual availability at the exit μ d With t = 0.237, the real-time availability index α(t) = 0.120 is obtained.

[0027] Step S300: Based on the node throughput rate and the real-time availability index, obtain the dynamic dispatch quota for the current scheduling period; Specifically, obtain the duration T of the current scheduling period. p Combined with the node flow rate R n The theoretical departure quota N0 is obtained, and the formula is N0=R n ×Tp The theoretical dispatch quota N0 represents the number of vehicles that the emergency resource receiving area can normally receive within the current dispatch cycle. The theoretical dispatch quota N0 is corrected based on the real-time availability index α(t) to obtain the dynamic dispatch quota N(t) for the current scheduling period, with the formula N(t) = ⌈N0 × α(t)⌉. The dynamic dispatch quota N(t) for the current scheduling period represents the actual number of dispatches after dynamic adjustment based on the current availability of the emergency resource receiving area. When N(t) is less than the preset lower limit constraint N... min When, let N(t) = N min This is the minimum number of vehicles dispatched to ensure the continuity of rescue efforts; when N(t) is greater than the preset upper limit constraint N... max When, let N(t) = N max The maximum number of vehicles that can be dispatched to prevent congestion in the emergency resource receiving area.

[0028] In one specific embodiment, the current scheduling period duration T is obtained. p =0.5h, combined with node flow rate R n =122 vehicles / hour, calculate the theoretical dispatch quota N0 = 61 vehicles. The current real-time availability index α(10:30) = 0.120, calculate the dynamic dispatch quota N(t) = 8 vehicles. Set a lower limit constraint N. min =3 vehicles, upper limit constraint N max =20 vehicles, N(t)=8 satisfies the constraint range, and the final dynamic dispatch quota is 8 vehicles.

[0029] Step S400: Obtain the material demand sequence; based on the dynamic dispatch quota and the candidate vehicle size parameters, truncate and perform shape matching on the material demand sequence to obtain a constrained dispatch task set; send the constrained dispatch task set to the vehicle waiting area; based on the candidate vehicle size parameters, perform spatial nesting sorting on the vehicles to be dispatched in the vehicle waiting area to generate a vehicle waiting sequence. Specifically, recorded event handling records are extracted from the historical emergency response database. Each record includes the event type, event level, response level, incident area type, handling task list, and corresponding responsible department information. The event type is mapped to an integer code, the event level to an integer code of 1 to 5, the response level to an integer code of 1 to 4, and the incident area type to an integer code. These four codes are concatenated to form an input feature vector of length 4. The task types in the handling task list are encoded as binary vectors of length Q, where Q is the total number of task types in the task database. If the p-th task exists in the task list, the corresponding position is set to 1; otherwise, it is set to 0. The responsible departments corresponding to each task in the handling task list are encoded as a Q-row, D-column one-hot matrix, where D is the total number of responsible departments in the department database. The p-th row represents the responsible department corresponding to the p-th task; if the task does not exist, the corresponding row is all set to 0. At least 500 historical event records are collected as a training dataset, divided into training and validation sets in an 8:2 ratio.

[0030] A three-layer backpropagation (BP) neural network model is constructed. The input layer has 4 neurons, which receive event type encoding, event level encoding, response level encoding, and event location type encoding, respectively. The hidden layer is a single layer with 12 neurons and the activation function is sigmoid. The output layer has two parallel branches. The task classification branch has Q neurons and the activation function is sigmoid. The output value of each neuron indicates whether the task is a sub-item that should be executed in the current event. When the output value is greater than 0.5, the event is considered to include the task. The department matching branch has an output layer of a Q-row D-column matrix and the activation function is softmax. For each task sub-item, it outputs the probability distribution of the responsible department corresponding to the task. The department with the highest probability is selected as the matching responsible department for the task. Training employs the backpropagation algorithm. The loss function for the task classification branch is the binary cross-entropy loss, and the loss function for the department matching branch is the multi-class cross-entropy loss. The total loss function is the weighted sum of the losses of the two branches, with weights of 0.6 and 0.4, respectively. The initial learning rate is set to 0.01, the batch size is set to 32, and the training epochs are set to 200. Training is terminated early when the validation set loss no longer decreases for 20 consecutive epochs, thus completing the learning of the task decomposition rules in the historical emergency response records.

[0031] When a new emergency occurs, the event type, event level, response level, and incident area type are extracted and encoded as input feature vectors, which are then input into the trained model. The model invokes the learned task decomposition rules. The task classification branch outputs a set of related task sub-items that should be executed for the current event, and the department matching branch outputs the responsible department corresponding to each task sub-item. The model automatically outputs the related task sub-items that should be executed for the event and the responsible department for each task sub-item. Each responsible department determines the required material type and quantity based on the tasks it undertakes. After arranging the materials in descending order of priority, a material demand sequence M = {m1, m2, ..., m} is generated. k}

[0032] Obtain the material demand sequence M = {m1, m2, ..., m k}, where k is the total number of batches of goods to be transported, m j The loading volume requirement for the j-th batch of materials is given by the material requirement sequence, which is arranged from high to low priority. Using the dynamic dispatch quota N(t) as the upper limit, the material demand sequence is truncated, and the first K batches of materials are truncated, with the formula K=min(k,N(t)), where K is the actual number of batches of materials truncated; after truncation, the material subsequence to be matched is obtained as M'={m1,m2,...,m K}; For the cut-off supplies m j Based on the loading volume V in the candidate vehicle size parameters i Calculate the loading morphology matching degree G for each candidate vehicle type. ij The formula is G ij =V i / m j The loading form matching degree G ij Let be the matching degree of the loading form of the i-th type of vehicle with the j-th batch of materials, and represent the ratio of the vehicle loading volume to the required loading volume of the materials. Select the loading form matching degree G ij Greater than or equal to 1 and G ij The smallest candidate vehicle type as material m j Matching vehicle type v j , where v j The vehicle type number is used to match the j-th batch of supplies; the selection indicates that the loading capacity of the selected vehicle type is closest to the volume requirement of the supplies among all vehicle types that can carry the supplies. At this time, the loading capacity utilization rate of a single vehicle is the highest and the transportation capacity waste is the least, forming a constrained departure task set T = {(m j ,v j )|j=1,2,...,K}, indicating that there are K scheduling tasks, each scheduling task specifying a batch of materials and their matching vehicle type.

[0033] Obtain all vehicles and their corresponding vehicle types within the vehicle waiting area, and determine the length L of each vehicle based on the candidate vehicle size parameters. i Width B i and loading volume V i The vehicle waiting area is a vehicle queuing buffer area independent of the emergency resource receiving area. It is located outside the entrance of the emergency resource receiving area and is used to temporarily store vehicles waiting to be dispatched to enter the emergency resource receiving area. All vehicles within the vehicle waiting area are subjected to spatial nesting sorting, wherein the first layer is based on the loading volume V. i Arranged in descending order, the second layer is based on the first layer according to the length L. i Arranged in ascending order, the third layer is based on the second layer according to the width B mentioned above. i The vehicles are arranged in alternating odd and even order; the sorted vehicles are matched with the constrained departure task set T to generate a sequence of vehicles to be dispatched, S.

[0034] In one specific embodiment, the dynamic dispatch quota N(t) = 8 vehicles is used as the upper limit, starting from the material demand sequence M = {120m}. 3 Chemical protective suit, 80m 3 Water pump, 60m 3 Sandbag, 45m 3 Tent, 30m 3 Extracting the first K=5 batches of materials from the food category, we obtain the subsequence M'={120m} of materials to be matched. 3 Chemical protective suit, 80m 3 Water pump, 60m 3 Sandbag, 45m 3 Tent, 30m 3 Food. Based on the candidate vehicle's loading volume V1 = 16m³. 3 and V2=20m 3 Calculate the matching degree G between the loading configuration of each batch of materials and each vehicle type. ij Among them, the chemical protective suit is 120m 3 Matching V1 requires 8 vehicles, G 11 =1.067, requires 6 vehicles to match V2, G 21 =1.0, select V2; water pump 80m 3 Five vehicles are needed to match V1, G 12 =1.0, requires 4 vehicles to match V2, G 22 =1.0, both vehicle types have a utilization rate of 100%, but the V2 vehicle is used less frequently, so the V2 is selected; Sandbag 60m 3 Four vehicles are needed to match V1, G 13 =1.067, requires 3 vehicles to match V2, G 23 =1.0, select V2; tent 45m3 Three vehicles are needed to match V1, G 14 =1.067, requires 3 vehicles to match V2, G 24 =1.333, select V1; food 30m 3 Two vehicles are needed to match V1, G 15 =1.067, requires 2 vehicles to match V2, G 25 =1.333, select V1. Form a constrained dispatch task set T={(chemical protective suits, 6 extended minivans), (water pumps, 4 extended minivans), (sandbags, 3 extended minivans), (tents, 3 minivans), (food, 2 minivans)}. This task set contains the transportation needs of 5 batches of supplies, requiring a total of 18 vehicle trips, including 13 extended minivans and 5 minivans. Obtain all 18 vehicles waiting to be dispatched in the vehicle waiting area, and determine the length L of each vehicle based on the candidate vehicle size parameters. i Width B i and loading volume V i Of these, 13 were extended minivans and 5 were regular minivans. All vehicles were sorted using a nested spatial arrangement, with the first layer based on loading volume V. i Arrange in descending order, with the second layer based on the length L of the first layer. i Arranged in ascending order, the third layer is based on the second layer and ordered by width B. i The odd and even numbers are alternately arranged to generate the sequence S of vehicles to be dispatched.

[0035] Step S500: Based on the dynamic vehicle dispatch quota, extract the corresponding number of target vehicles from the vehicle queue to be dispatched, generate a timed release instruction for the target vehicles; send the timed release instruction to the target vehicles, and control the target vehicles to leave the vehicle waiting area in time slots and enter the emergency resource receiving area to perform unloading operations.

[0036] Specifically, N(t) vehicles are extracted from the sequence of vehicles to be dispatched S as target vehicles, denoted as s1, s2, ..., s N(t) The vehicle identifier of the r-th target vehicle is s. r r is the target vehicle number, r = 1, 2, ..., N(t); obtain the length L of each target vehicle. i Calculate the minimum safe time interval Δt between two adjacent target vehicles. s The formula is Δt s =max(L i / v min ,τ0); where v min τ0 is the minimum safe passage speed within the preset emergency resource receiving area, and τ0 is the minimum delay time for preset command response. The real-time availability index α(t) and exit velocity retention rate s(t) at the current moment are input into a pre-trained three-layer BP neural network model. The BP neural network model outputs a normalized offset β0, where β0∈[0,1]. The input layer of the BP neural network model has 2 neurons, which receive α(t) and s(t) respectively. The hidden layer is a single layer with 5 neurons and uses the Sigmoid activation function. The output layer has 1 neuron and uses the Sigmoid activation function. The normalized offset β0 is multiplied by the minimum safe time interval Δt. s The departure fine-tuning offset β of the r-th target vehicle is obtained. r The BP neural network model is pre-trained using historical scheduling data. The training samples are α(t) and s(t) at each moment in the historical scheduling records, along with their corresponding optimal offset labels. The optimal offset labels are obtained through statistical analysis of the actual departure offsets in the historical scheduling records and the subsequent congestion levels in the emergency resource receiving area. Specifically, this involves extracting α(t), s(t), and the actual departure fine-tuning offset β at each moment from the historical scheduling records. r Clustering is performed based on the values ​​of α(t) and s(t). Within the same cluster, the actual β value that minimizes subsequent congestion in the emergency resource receiving area is selected. r The value is used as the optimal offset label for this cluster, and the optimal offset label is divided by the Δt corresponding to the time. s After normalization, the labels are used as training labels; training uses the backpropagation algorithm and the loss function is the mean squared error. With the minimum safe time interval Δt s The departure fine-tuning offset β is added at intervals. r The release time t for the r-th target vehicle is obtained. r The formula is t r =t0+(r-1)×Δt s +β r Where t0 is the current time; According to the vehicle identification s of each target vehicle r and release time t r Generate timing release instruction set Z = {(s r ,t r )|r=1,2,…,N(t)}.

[0037] Each timing release instruction in the timing release instruction set Z is sent to the corresponding target vehicle in a time-sharing manner; each target vehicle is released at its corresponding release time t. rThe system initiates and travels along a designated path towards the emergency resource receiving area. When the target vehicle arrives at the entrance of the emergency resource receiving area, the real-time availability index α(t) is obtained. If α(t) is greater than or equal to a preset entry availability threshold, the target vehicle is allowed to enter the emergency resource receiving area to perform unloading operations. If α(t) is less than the preset entry availability threshold, the target vehicle is controlled to wait in the entrance waiting area until α(t) is greater than or equal to the preset entry availability threshold before being allowed to enter. After each target vehicle's unloading operation is completed, the real-time accumulation volume and departure congestion status of the emergency resource receiving area are updated. The entrance waiting area is located between the vehicle waiting area and the entrance of the emergency resource receiving area and is used to temporarily store vehicles that have left the vehicle waiting area but have not yet been allowed to enter the emergency resource receiving area.

[0038] In one specific embodiment, the first N(t) = 8 vehicles are extracted from the sequence of vehicles to be dispatched S as target vehicles, denoted as s1, s2, ..., s8. The current time t0 = 11:00:00 is obtained, the length of each target vehicle L2 = 4.5m, and the minimum safe passage speed v within the preset emergency resource receiving area is determined. min =1.5m / s, preset minimum command response delay time τ0=2.0s, calculate the minimum safe distance Δt between two adjacent target vehicles. s =3.0s. The current real-time availability index α(11:00) = 0.85, the exit speed maintenance rate s(t) = 0.78, the input is a pre-trained three-layer BP neural network model, and the output is a normalized offset β0 = 0.4, multiplied by Δt. s The departure fine-tuning offset β is obtained. r=1.2s. The actual departure interval is calculated to be 4.2s, and the release times for each target vehicle are obtained: t1=11:00:00, t2=11:00:04.2, t3=11:00:08.4, t4=11:00:12.6, t5=11:00:16.8, t6=11:00:21.0, t7=11:00:25.2, t8=11:00:29.4. A time-series release instruction set Z={(s1, 11:00:00), (s2, 11:00:04.2), ...,(s8, 11:00:29.4)} is generated. The time-series release instructions are sent to the corresponding target vehicles in a time-division manner, and each target vehicle starts to leave the vehicle waiting area at the corresponding release time. When the first target vehicle s1 arrives at the entrance of the emergency resource receiving area, the real-time availability index α(11:10) is 0.82, which is greater than the preset entry availability threshold of 0.4, allowing s1 to enter the emergency resource receiving area to perform unloading operations. When the second target vehicle s2 arrives, α(11:12) is 0.79, which is greater than 0.4, allowing it to enter. When the third target vehicle s3 arrives, α(11:14) is 0.35, which is less than 0.4, controlling s3 to wait in the entrance waiting area, continuously monitoring α(t). By 11:17, α(t) recovers to 0.52, which is greater than or equal to 0.4, allowing s3 to enter. Subsequent vehicles are released sequentially according to the same rules. After each target vehicle's unloading operation is completed, the system updates the real-time accumulation and departure congestion status of the emergency resource receiving area, serving as input data for calculating the real-time availability index α(t) in the next scheduling cycle.

[0039] like Figure 2 The flowchart illustrates the real-time availability index generation steps of an artificial intelligence-based emergency resource scheduling optimization method. This invention provides an artificial intelligence-based emergency resource scheduling optimization method, comprising: The real-time availability index generation process begins with data collection and consists of two parallel paths: inbound status monitoring and outbound status monitoring. The inbound status monitoring path monitors the queue length L of vehicles waiting to unload at the entrance of the emergency resource receiving area in real time. q (t) and the number of vehicles N currently unloading. u (t), L q (t) reflects the number of vehicles waiting to enter at the entrance, N u (t) reflects how many vehicles are occupying the unloading berth; simultaneously, the dynamic concurrent capacity C is obtained. p This refers to the maximum number of vehicles that the emergency resource receiving area can accommodate at the same time; L q (t) and N u Add (t) together to get the total number of vehicles in the yard, then divide by C. p The in-field backlog P is obtained. q(t) represents the proportion of the current total number of vehicles in the emergency resource receiving area to its maximum capacity. Based on this, P... q (t) Derive the remaining capacity rate R(t) within the field. Real-time monitoring of the exit status path shows the average passing speed v of vehicles departing from the exit of the emergency resource receiving area. d (t), and obtain the free-flow velocity v of the vehicle leaving the passage. f This refers to the speed at which vehicles pass through the exit normally under uncongested conditions. (v) d (t) divided by v f The export speed maintenance rate s(t) is obtained. The closer s(t) is to 1, the smoother the export is.

[0040] The remaining capacity rate R(t) within the field is used as a loss factor applied to the exit velocity maintenance rate s(t) to calculate the actual availability of the exit μ. d (t). The remaining capacity rate within the facility, R(t), is compared with the actual availability at the outlet, μ. d Multiplying (t) yields the real-time availability index α(t), which reflects the overall availability of berths in the emergency resource receiving area, indicating that there is currently sufficient capacity to accommodate new vehicles and that exits are unobstructed and can release berths.

[0041] like Figure 3 The system architecture diagram of an artificial intelligence-based emergency resource scheduling optimization system is shown. This invention provides an artificial intelligence-based emergency resource scheduling optimization system, comprising: The basic capacity and flow calculation module includes a parameter acquisition unit and a capacity calculation unit. The parameter acquisition unit acquires the regional layout parameters of the emergency resource receiving area, the candidate vehicle size parameters, the duration of a single unloading operation, and the vehicle departure time after unloading. The capacity calculation unit calculates the dynamic concurrent capacity based on the regional layout parameters of the emergency resource receiving area and the candidate vehicle size parameters, and calculates the node flow rate based on the duration of a single unloading operation and the vehicle departure time after unloading. The real-time status monitoring and availability evaluation module includes a status monitoring unit and an availability calculation unit. The status monitoring unit monitors in real time the queue length of vehicles waiting to be unloaded at the entrance of the emergency resource receiving area and the number of vehicles currently unloading, as well as the average passing speed of vehicles leaving the exit. The availability calculation unit recursively generates a real-time availability index based on dynamic concurrent capacity, queue length of vehicles waiting to be unloaded, number of vehicles currently unloading, average passing speed of vehicles leaving the exit, and free flow speed. The scheduling quota and task matching module includes a quota generation unit and a task matching unit. The quota generation unit generates dynamic dispatch quotas based on node flow rate and real-time availability index. The task matching unit truncates the material demand sequence based on the dynamic dispatch quotas and matches candidate vehicle types for each batch of materials after truncation, generating a constrained dispatch task set. Vehicle sorting and instruction generation module: includes a sorting unit and an instruction generation unit; wherein, the sorting unit performs spatial nested sorting of vehicles to be dispatched in the vehicle waiting area to generate a sequence of vehicles to be dispatched; the instruction generation unit extracts target vehicles from the sequence of vehicles to be dispatched according to the dynamic dispatch quota and generates a time-sequenced release instruction. The execution control module includes an execution control unit and an entry verification unit. The execution control unit sends a timed release command to the target vehicle, controlling the target vehicle to enter the emergency resource receiving area in a timed manner to perform unloading operations. The entry verification unit controls whether to release the vehicle to enter the emergency resource receiving area based on the real-time availability index when the target vehicle arrives at the entrance.

[0042] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An emergency resource scheduling optimization method based on artificial intelligence, characterized in that, include: Obtain the regional layout parameters and candidate vehicle size parameters of the emergency resource receiving area, calculate the spatial adaptability, and generate the dynamic concurrent capacity of the emergency resource receiving area; Obtain the buffer area parameters and transport flow parameters of the emergency resource receiving area, and combine them with the dynamic concurrency capacity to obtain the node flow rate of the emergency resource receiving area; Real-time monitoring of the real-time accumulation and departure congestion status of the emergency resource receiving area, generating a real-time availability index for the emergency resource receiving area; Based on the node throughput rate and the real-time availability index, the dynamic dispatch quota for the current scheduling period is obtained; Obtain the material demand sequence, and based on the dynamic dispatch quota and the candidate vehicle size parameters, truncate and perform shape matching on the material demand sequence to obtain a constrained dispatch task set; distribute the constrained dispatch task set to the vehicle waiting area, and perform spatial nesting sorting on the vehicles to be dispatched in the vehicle waiting area based on the candidate vehicle size parameters to generate a vehicle waiting sequence. Based on the dynamic vehicle dispatch quota, a corresponding number of target vehicles are extracted from the vehicle queue to be dispatched, and a timed release instruction is generated for the target vehicles. The timed release instruction is sent to the target vehicles, and the target vehicles are controlled to leave the vehicle waiting area in time slots and enter the emergency resource receiving area to perform unloading operations.

2. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The process of acquiring the regional layout parameters and candidate vehicle size parameters of the emergency resource receiving area, calculating the spatial adaptability, and generating the dynamic concurrent capacity of the emergency resource receiving area includes: Obtain the regional layout parameters of the emergency resource receiving area, including the effective operating area A. e and loading / unloading aisle width W c Obtain the length, width, minimum turning radius, and loading volume of each candidate vehicle type as candidate vehicle size parameters, where the length of the i-th candidate vehicle is L. i Width is B i The minimum turning radius is R i The loading volume is V i Where i is the type number of the candidate vehicle; According to the length L i and width B i The single-vehicle operating area S of the i-th candidate vehicle is obtained. i According to the width W of the loading and unloading channel c and the minimum turning radius R i The channel adaptation coefficient C of the i-th type of candidate vehicle is obtained. i Remove the channel adaptation coefficient C. i Candidate vehicle types with a value less than 1; Obtain the proportion of each type of candidate vehicle, where the proportion of the i-th type of candidate vehicle is P. i According to the effective working area A e The aforementioned proportion P i The single-vehicle operating area S i and the channel adaptation coefficient C i The dynamic concurrent capacity C of the emergency resource receiving area is obtained. p The formula is: 。 3. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The step of obtaining the buffer area parameters and transport flow parameters of the emergency resource receiving area, and combining them with the dynamic concurrency capacity, to obtain the node flow rate of the emergency resource receiving area includes: Get the duration T of a single unload job u and the time T for the vehicle to leave after unloading d Calculate the average operating turnaround time T for a single vehicle. t According to the dynamic concurrency capacity C p and the average operating turnaround time T of a single vehicle t The node throughput R of the emergency resource receiving area is obtained. n .

4. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The real-time monitoring of the accumulation and departure congestion status of the emergency resource receiving area, and the generation of a real-time availability index for the emergency resource receiving area, includes: Real-time monitoring of the queue length L of vehicles waiting to unload at the entrance of the emergency resource receiving area. q (t) and the number of vehicles N currently unloading. u (t), combined with the dynamic concurrency capacity C p The in-field backlog P is obtained. q (t), where t is the current time; based on the in-field accumulation degree P q From (t), we derive the remaining capacity rate R(t) in the field, which is given by the formula R(t) = 1 - P. q (t); Real-time monitoring of the average passing speed v of vehicles leaving the emergency resource receiving area exit. d (t), to obtain the free-flow velocity v of the vehicle leaving the passage. f The exit velocity hold rate s(t) is obtained, and the formula is s(t) = v d (t) / v f ; Based on the remaining capacity rate R(t) in the field and the exit velocity retention rate s(t), calculate the actual availability of the exit μ. d (t), the formula is μ d (t) = s(t) × R(t); Based on the remaining capacity rate R(t) in the field and the actual availability of the outlet μ d (t), generate the real-time availability index α(t) of the emergency resource receiving area, with the formula α(t) = R(t) × μ d (t).

5. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The step of obtaining the dynamic dispatch quota for the current scheduling period based on the node throughput rate and the real-time availability index includes: Get the duration T of the current scheduling period p Combined with the node flow rate R n The theoretical departure quota N0 is obtained; the theoretical departure quota N0 is corrected according to the real-time availability index α(t) to obtain the dynamic departure quota N(t) for the current scheduling period.

6. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The process of obtaining the material demand sequence involves truncating and performing shape matching on the material demand sequence based on the dynamic dispatch quota and the candidate vehicle size parameters to obtain a constrained dispatch task set, including: Obtain the material demand sequence M = {m1, m2, ..., m k }, where k is the total number of batches of goods to be transported, m j The required loading volume for the jth batch of supplies; Using the dynamic dispatch quota N(t) as the upper limit, the material demand sequence is truncated, and the first K batches of materials are truncated, with the formula K=min(k,N(t)), where K is the actual number of batches of materials truncated; after truncation, the material subsequence to be matched is obtained as M'={m1,m2,...,m K }; For the cut-off supplies m j Based on the loading volume V in the candidate vehicle size parameters i Calculate the loading morphology matching degree G for each candidate vehicle type. ij ; Select the loading form matching degree G ij Greater than or equal to 1 and G ij The smallest candidate vehicle type as material m j Matching vehicle type v j This forms a constrained departure task set T = {(m j ,v j )|j=1,2,...,K}.

7. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The step of distributing the constrained departure task set to the vehicle waiting area, and spatially nesting and sorting the vehicles to be dispatched in the vehicle waiting area based on the candidate vehicle size parameters to generate a sequence of vehicles to be dispatched includes: Obtain all vehicles and their corresponding vehicle types within the vehicle waiting area, and determine the length L of each vehicle based on the candidate vehicle size parameters. i Width B i and loading volume V i ; All vehicles within the vehicle waiting area are subjected to spatial nesting sorting, wherein the first layer is based on the loading volume V. i Arranged in descending order, the second layer is based on the first layer according to the length L. i Arranged in ascending order, the third layer is based on the second layer according to the width B mentioned above. i The vehicles are arranged in alternating odd and even order; the sorted vehicles are matched with the constrained departure task set T to generate a sequence of vehicles to be dispatched, S.

8. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The step of extracting a corresponding number of target vehicles from the queue of vehicles to be dispatched based on the dynamic dispatch quota, and generating a timed release instruction for the target vehicles, includes: Extract the aforementioned dynamic departure quota N(t) vehicles from the vehicle sequence S to be dispatched as target vehicles, denoted as s1, s2, ..., s N(t) The vehicle identifier of the r-th target vehicle is s. r r is the target vehicle number, r = 1, 2, ..., N(t); obtain the length L of each target vehicle. i Calculate the minimum safe time interval Δt between two adjacent target vehicles. s ; The real-time availability index α(t) and exit speed maintenance rate s(t) at the current moment are input into a pre-trained three-layer BP neural network model, which outputs a normalized offset β0; based on the normalized offset β0 and the minimum safe time interval Δt... s The departure fine-tuning offset β of the r-th target vehicle is obtained. r ; With the minimum safe time interval Δt s The departure fine-tuning offset β is added at intervals. r The release time t for the r-th target vehicle is obtained. r ; According to the vehicle identification s of each target vehicle r and release time t r Generate timing release instruction set Z = {(s r ,t r )|r=1,2,…,N(t)}.

9. The emergency resource scheduling optimization method based on artificial intelligence according to claim 1, characterized in that, The step of sending the timing release command to the target vehicle and controlling the target vehicle to leave the vehicle waiting area in shifts and enter the emergency resource receiving area to perform unloading operations includes: Each timing release instruction in the timing release instruction set Z is sent to the corresponding target vehicle in a time-sharing manner; each target vehicle is released at its corresponding release time t. r The vehicle starts and travels along the designated path towards the emergency resource receiving area. When the target vehicle arrives at the entrance of the emergency resource receiving area, the real-time availability index α(t) is obtained. When α(t) is greater than or equal to the preset entry availability threshold, the target vehicle is allowed to enter the emergency resource receiving area to perform unloading operations. When α(t) is less than the preset entry availability threshold, the target vehicle is controlled to wait in the entrance waiting area until α(t) is greater than or equal to the preset entry availability threshold before it is allowed to enter.

10. An artificial intelligence-based emergency resource scheduling optimization system, using the artificial intelligence-based emergency resource scheduling optimization method according to any one of claims 1-9, characterized in that, include: The basic capacity and flow calculation module includes a parameter acquisition unit and a capacity calculation unit. The parameter acquisition unit acquires the regional layout parameters of the emergency resource receiving area, the candidate vehicle size parameters, the duration of a single unloading operation, and the vehicle departure time after unloading. The capacity calculation unit calculates the dynamic concurrent capacity based on the regional layout parameters of the emergency resource receiving area and the candidate vehicle size parameters, and calculates the node flow rate based on the duration of a single unloading operation and the vehicle departure time after unloading. The real-time status monitoring and availability evaluation module includes a status monitoring unit and an availability calculation unit. The status monitoring unit monitors in real time the queue length of vehicles waiting to be unloaded at the entrance of the emergency resource receiving area and the number of vehicles currently unloading, as well as the average passing speed of vehicles leaving the exit. The availability calculation unit recursively generates a real-time availability index based on dynamic concurrent capacity, queue length of vehicles waiting to be unloaded, number of vehicles currently unloading, average passing speed of vehicles leaving the exit, and free flow speed. The scheduling quota and task matching module includes a quota generation unit and a task matching unit. The quota generation unit generates dynamic dispatch quotas based on node flow rate and real-time availability index. The task matching unit truncates the material demand sequence based on the dynamic dispatch quotas and matches candidate vehicle types for each batch of materials after truncation, generating a constrained dispatch task set. Vehicle sorting and instruction generation module: includes a sorting unit and an instruction generation unit; wherein, the sorting unit performs spatial nested sorting of vehicles to be dispatched in the vehicle waiting area to generate a sequence of vehicles to be dispatched; the instruction generation unit extracts target vehicles from the sequence of vehicles to be dispatched according to the dynamic dispatch quota and generates a time-sequenced release instruction. The execution control module includes an execution control unit and an entry verification unit. The execution control unit sends a timed release command to the target vehicle, controlling the target vehicle to enter the emergency resource receiving area in a timed manner to perform unloading operations. The entry verification unit controls whether to release the vehicle to enter the emergency resource receiving area based on the real-time availability index when the target vehicle arrives at the entrance.

Images