Elevator real-time scheduling optimization system based on edge calculation

Abstract

The invention discloses an elevator real-time scheduling optimization system based on edge calculation, which comprises a multi-source scheduling data acquisition and preprocessing module for generating a standardized scheduling input vector set; the event sensing door module is used for receiving the input vector and a historical index, calculating a leapfrogging probability and outputting a variable step size value; the variable step size control module is used for controlling hidden state updating and generating a hidden state cache sequence; the leapfrogging attention pooling module is used for generating an abstract vector; the improved SkipRNN hidden state updating module is used for updating a hidden state; the linear readout layer module is used for generating a target floor instruction sequence; the group control scheduling controller interaction module is used for generating a multi-target reward value; and the element reinforcement learning head module is used for updating the event sensing gate and step length mapping parameters to complete closed-loop optimization. According to the method, elevator dispatching intelligence, response rapidness and energy consumption optimization are achieved, and the operation efficiency and the passenger satisfaction degree are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elevator intelligent scheduling, and particularly relates to an elevator real-time scheduling optimization system based on edge computing. BACKGROUND

[0002] The existing elevator scheduling method mainly makes decisions based on fixed rules or centralized control strategies, and often adopts first-come-first-served, shortest distance first or time window first scheduling optimization. In actual application, some systems introduce machine learning algorithms based on cloud deployment to try to improve scheduling efficiency through historical data mining. Such methods rely on centralized servers for state calculation and scheduling decisions, and have problems of large response delay and poor edge adaptability, which cannot meet the real-time requirements under high-frequency passenger flow changes and multi-objective scheduling demands. At the same time, some edge-deployed lightweight models lack the ability of dynamic perception of boarding events, and the scheduling strategy is difficult to jointly optimize according to the elevator state, passenger flow density and energy consumption level, resulting in an imbalance between scheduling action and operating energy consumption.

[0003] Although the existing recurrent neural network structure has sequence modeling capability, its running efficiency is limited under edge computing conditions. The SkipRNN (recurrent neural network with skip state update) structure introduces a skip mechanism to reduce the amount of calculation, but the skip decision is based on fixed gating logic, which cannot dynamically perceive changes in boarding environment and cannot optimize the skip strategy according to performance feedback, which is easy to cause omission of key state updates and affect scheduling accuracy. At the same time, the original SkipRNN lacks a compensation mechanism for input information during skipping, and does not have the ability to model the content of the skipped period, which is easy to cause information breakage and affect the model's ability to capture long-term dependencies. The existing scheduling strategy lacks a closed-loop adaptive optimization mechanism, making it difficult to realize dynamic feedback correction between the scheduling strategy and system operation effect, and cannot meet the intelligent, multi-objective and low-delay group control elevator scheduling requirements.

[0004] Therefore, how to provide an elevator real-time scheduling optimization system based on edge computing is a problem that those skilled in the art need to solve. SUMMARY

[0005] One object of the present application is to provide an elevator real-time scheduling optimization system based on edge computing. The present application fully integrates edge computing and event perception mechanisms, and describes in detail the scheduling process of skip control, state caching and closed-loop reinforcement learning strategy, which has the advantages of low computational overhead, fast response speed and high scheduling accuracy.

[0006] According to the elevator real-time scheduling optimization system based on edge computing, the system comprises: A multi-source scheduling data acquisition and preprocessing module is configured to acquire elevator multi-source scheduling data and perform preprocessing to generate a standardized scheduling input vector set; An event-aware gate module is configured to receive the standardized scheduling input vector set, historical average waiting time indicators and historical average energy consumption indicators, input an event-aware gate, calculate a skip probability and output a variable step value; A variable step length control module is configured to input the variable step value into an improved SkipRNN and control the same through a variable step length control unit. When the variable step value k is greater than zero, the last hidden state is maintained. When the variable step value k is equal to zero, a hidden state update is performed, and a hidden state cache sequence is generated by caching k input vectors. A skip-attention pooling module is configured to perform an attention pooling operation on the hidden state cache sequence to generate an abstract vector. An improved SkipRNN hidden state update module is configured to combine the abstract vector and the next time step standardized scheduling input vector to update the improved SkipRNN hidden state. A linear readout layer module is configured to input the updated hidden state into a linear readout layer to generate a target floor instruction sequence. A group control scheduling controller interaction module is configured to write the target floor instruction sequence into a group control scheduling controller, receive a scheduling controller feedback execution result, calculate real-time average waiting time indicators and real-time average energy consumption indicators, and generate a multi-objective reward value. A meta-reinforcement learning head module is configured to input the multi-objective reward value into a policy optimization structure, update event-aware gate parameters and variable step length mapping parameters, and return to the skip probability calculation step to complete a closed-loop scheduling optimization process.

[0007] Optionally, the elevator multi-source scheduling data in the multi-source scheduling data acquisition and preprocessing module is composed of elevator state data, passenger request data and floor passenger flow density data. The elevator state data records the current floor, running direction, door state and load value in sequence. The passenger request data records the call floor, target floor and request timestamp in sequence. The floor passenger flow density data is a real-time passenger count of the waiting area of each floor. The preprocessing includes, in sequence, an alignment operation based on a unified time reference, a missing data filling operation based on linear interpolation, a noise filtering operation based on a sliding window mean and a numerical normalization operation based on minimum and maximum specifications, and the output standardized scheduling input vector is spliced in the order of elevator state data, passenger request data and floor passenger flow density data.

[0008] Optionally, the step of calculating the skip probability and outputting the variable step value in the event-aware gate module specifically includes: The current time step input vector in the standardized scheduling input vector set is read in time sequence; The current time step input vector is sequentially concatenated with the historical average waiting time index to generate the first concatenated vector. The first concatenated vector is further concatenated with the historical average energy consumption index to generate an event-aware input vector. The event-aware input vector is input into the linear transformation unit in the event-aware gate. The linear transformation unit performs matrix multiplication and bias addition operations on the event-aware input vector and outputs the intermediate calculation result vector. The intermediate calculation result vector is input into the activation function unit, and the activation function unit performs a Sigmoid function transformation on the intermediate calculation result vector to obtain the normalized skip probability value. The skip probability value is input into the variable step size mapping unit, and the skip probability value is mapped to an integer variable step size value k through a preset interval mapping function. The variable step size value k is limited to between 0 and K, where K is the preset maximum skip step size.

[0009] Optionally, the step of inputting the variable step size value into the improved SkipRNN and controlling it through the variable step size control unit in the variable step size control module specifically includes: Read the variable step size value k of the current time step and input the variable step size value k into the variable step size control unit in the improved SkipRNN; When the variable step size value k is greater than zero, the variable step size control unit disables the hidden state update operation of the current time step, directly passes the hidden state of the previous time step as the hidden state of the current time step, writes the normalized scheduling input vector of the current time step into the skip input buffer, and then appends the normalized scheduling input vector of each time step to the skip input buffer in the next k time steps, forming a skip input buffer sequence of length k. When the variable step size value k equals zero, the variable step size control unit triggers the hidden state update operation, inputs the normalized scheduling input vector of the current time step and the hidden state of the previous time step into the hidden state update unit, performs recursive calculation, and generates the updated hidden state of the current time step. After the length of the skip input buffer sequence reaches the variable step size value k, the skip input buffer sequence is written into the skip attention summarization unit. The skip attention summarization unit performs a weight normalization operation based on the additive attention mechanism on the input vector of each time step in the skip input buffer sequence to generate a summary vector. The digest vector is concatenated with the normalized scheduling input vector of the time step corresponding to the digest vector, and used as the composite input for the next effective hidden state update operation. This is used to compensate for the lack of input information during the skip step and update the hidden state of the improved SkipRNN. After the update is complete, clear the step input buffer and wait for the next round of step control operation.

[0010] Optionally, the hidden state update operation in the improved SkipRNN hidden state update module specifically includes: When the variable step length control unit receives a signal that the variable step length value k is equal to zero, read the normalized scheduling input vector at the current time step ; read the hidden state at the previous time step ; input the state update unit of the improved SkipRNN, and the state update unit performs weighted summation, nonlinear activation and state transformation operations to generate the initial updated hidden state at the current time step ; determine whether the generated summary vector Δs exists in the skip input buffer area, and when the summary vector Δs exists, splice the summary vector Δs with the current time step input vector to form a fusion input vector ; input the fusion input vector and the hidden state at the previous time step to the state update unit again, and repeat the weighted summation, nonlinear activation and state transformation operations to overwrite the initial update result and generate the final fusion updated hidden state ; write the final hidden state to the hidden state register, clear the skip input buffer area and the summary vector buffer area, and prepare for the next time step skip control; when the variable step length value k is greater than zero, skip the hidden state update operation, and directly pass the hidden state at the previous time step to the current time step as the current hidden state.

[0011] Optionally, the generation of the target floor instruction sequence in the linear readout layer module specifically includes: read the updated hidden state at the current time step output by the improved SkipRNN, and input the hidden state to the fully connected transformation unit of the linear readout layer; in the linear readout layer, perform a linear mapping operation on the hidden state vector to generate an unnormalized instruction vector by matrix multiplication with a weight matrix and adding a bias vector; perform activation function transformation on the unnormalized instruction vector to generate a normalized scheduling instruction vector; perform a maximum value selection operation on the normalized scheduling instruction vector to output the target floor number with the maximum activation value as the target scheduling floor at the current time step; On the basis of the target scheduling floor, a target floor instruction sequence is generated in combination with current elevator running direction, door state and load state information; The target floor instruction sequence is written into a task queue of the group control scheduling controller.

[0012] Optionally, the generation of the multi-objective reward value in the group control scheduling controller interaction module specifically includes: The target floor instruction sequence is written into a task execution queue of the group control scheduling controller, triggering the controller to execute an elevator scheduling action; After each scheduling action is completed, actual execution result data returned by the group control scheduling controller is received; According to the call time and elevator arrival time of each passenger request in the actual execution result data, the corresponding waiting time is calculated, and the waiting times of all passenger requests are averaged within a set sliding time window to obtain a real-time average waiting time index of the current time window; According to the motor running power and running time length in the actual execution result data, a cumulative energy consumption value in the current time window is calculated, and the cumulative energy consumption value is divided by the window time length to obtain a real-time average energy consumption index; The real-time average waiting time index and the real-time average energy consumption index are respectively subjected to difference calculation with a preset performance reference value to generate a waiting time deviation term and an energy consumption deviation term; The waiting time deviation term and the energy consumption deviation term are subjected to weighted summation operation to generate a multi-objective reward value of the current time step.

[0013] Optionally, the closed-loop scheduling optimization process in the meta-reinforcement learning head module specifically includes: The multi-objective reward value of the current time step is read, and the multi-objective reward value, the event perception input vector of the corresponding time step and the step probability value are combined to form a state-action-reward triple; The state-action-reward triple is input into a policy updating unit in the meta-reinforcement learning head, and the gradient of the event perception gate parameter is calculated according to the policy gradient optimization method; The linear transformation weight matrix and the bias vector in the event perception gate are updated using the gradient of the event perception gate parameter to generate a new event perception gate parameter; After the event perception gate parameter is updated, the difference between the step probability value and the corresponding reward value is taken as an advantage signal, which is input into a variable step length mapping parameter adjustment unit; The variable step length mapping parameter adjustment unit corrects the mapping function of the step probability to the step length interval according to the advantage signal, and adjusts the step length corresponding to the boundary value of each interval; The updated event perception gate parameter and the variable step length mapping parameter are written into a parameter storage area, and the event perception gate is returned to the step of calculating the step probability for the step control calculation of the next time step; After every preset number of time steps, a round of parameter synchronization operation is performed to replace the parameters in the current event-aware gate and variable step mapping module, so as to complete a strategy evolution period in the closed-loop scheduling optimization process.

[0014] The beneficial effects of the present application are: (1) The elevator scheduling efficiency is improved. By introducing the improved SkipRNN and combining the event-aware gate mechanism, the model skip frequency is dynamically adjusted, the redundant calculation is reduced, the running efficiency and response speed of the model on the edge device are improved, and the high timeliness requirement of real-time elevator scheduling control is met. (2) The energy consumption and user waiting time are reduced. The multi-objective reward mechanism is used to optimize the average energy consumption and average waiting time indicators, and the dynamic balance between elevator operation energy efficiency and passenger service quality is achieved, which has the dual effects of energy saving and emission reduction and user experience improvement in actual deployment. (3) The scheduling strategy is self-adaptively updated. The meta-reinforcement learning mechanism is used to continuously optimize the event-aware gate parameters and variable step mapping parameters, so that the scheduling strategy can learn and adapt to different passenger flow patterns and use scenarios, and has good generalization ability and application expansion. (4) The system design emphasizes the low delay and high responsiveness requirements of scheduling decisions. By deploying a lightweight standardized scheduling input vector generation module, a skip control unit and a linear readout layer module on the elevator edge device, most of the decision-making processes can be completed on the local terminal, reducing the dependence on the central server, and meeting the principle of edge computing processing on the side close to the data source. The group control scheduling controller interaction module can realize real-time two-way communication and feedback collection with the edge group control system, forming a typical edge computing intelligent closed loop. BRIEF DESCRIPTION OF DRAWINGS

[0015] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. In the drawings: Fig. 1 A module structure diagram of an elevator real-time scheduling optimization system based on edge computing is provided for the present application; Fig. 2 An improved SkipRNN skip mechanism structure diagram of an elevator real-time scheduling optimization system based on edge computing is provided for the present application; Fig. 3 A multi-objective reward optimization closed loop structure diagram of an elevator real-time scheduling optimization system based on edge computing is provided for the present application. DETAILED DESCRIPTION

[0016] The application will be described in further detail below with reference to the drawings. These drawings are simplified schematic diagrams and only show the basic structure of the application in a schematic manner, and thus only show the components relevant to the application.

[0017] Reference Figs. 1-3 An edge computing-based elevator real-time scheduling optimization system, comprising: A multi-source scheduling data acquisition and preprocessing module for acquiring elevator multi-source scheduling data, performing preprocessing to generate a standardized scheduling input vector set; An event-aware gate module for receiving the standardized scheduling input vector set, historical average waiting time indicators, and historical average energy consumption indicators, inputting into an event-aware gate, calculating a skip probability, and outputting a variable step value; A variable step control module for inputting the variable step value into an improved SkipRNN, controlling through a variable step control unit, maintaining the previous hidden state when the variable step value k is greater than zero, performing hidden state update when the variable step value k is equal to zero, and simultaneously caching k input vectors to generate a hidden state cache sequence; A skip attention pooling module for performing attention pooling operation on the hidden state cache sequence to generate an abstract vector; An improved SkipRNN hidden state update module for merging the abstract vector with the next time step standardized scheduling input vector to update the improved SkipRNN hidden state; A linear readout layer module for inputting the updated hidden state into a linear readout layer to generate a target floor instruction sequence; A group control scheduling controller interaction module for writing the target floor instruction sequence into a group control scheduling controller, receiving scheduling controller feedback execution results, calculating real-time average waiting time indicators and real-time average energy consumption indicators, and generating a multi-objective reward value; A meta-reinforcement learning head module for inputting the multi-objective reward value into a policy optimization structure to update event-aware gate parameters and variable step mapping parameters, returning to the skip probability calculation step, and completing a closed-loop scheduling optimization process.

[0018] In this embodiment, the modules are implemented through the following methods: Acquiring elevator multi-source scheduling data, performing preprocessing to generate a standardized scheduling input vector set; Concatenating the standardized scheduling input vector set, historical average waiting time indicators, and historical average energy consumption indicators, inputting into an event-aware gate, calculating a skip probability, and outputting a variable step value; The variable step length value is input into the improved SkipRNN, and the variable step length control unit is used for control, when the variable step length value k is greater than zero, the last hidden state is kept, and when the variable step length value k is equal to zero, the hidden state updating is performed, and a hidden state cache sequence is generated by caching the k input vectors; The attention pooling operation is performed on the hidden state cache sequence to generate an abstract vector, the abstract vector is combined with a next time step standardized scheduling input vector, and the improved SkipRNN hidden state is updated; In the embodiment, the attention pooling operation is used for abstract extraction on the skip input cache sequence in the improved SkipRNN, so as to realize the weighted fusion compensation of the input information of the skipped time step, and the specific process is as follows: each standardized scheduling input vector in the skip input cache sequence is combined with a context query vector of a current time step, and is input into a scoring function to generate a corresponding attention score; all scores are normalized by a Softmax function to obtain an attention weight of each time step input vector; all skip input vectors are weighted and summed according to the attention weights to generate an abstract vector; The updated hidden state is input into a linear readout layer to generate a target floor instruction sequence; The target floor instruction sequence is written into a group control scheduling controller, a feedback execution result of the scheduling controller is received, a real-time average waiting time index and a real-time average energy consumption index are calculated, and a multi-objective reward value is generated; The multi-objective reward value is input into the meta-reinforcement learning head, the event perception gate parameter and the variable step length mapping parameter are updated, and the updated event perception gate is returned to the step of calculating the skip probability, and the closed-loop scheduling optimization process is completed.

[0019] In the embodiment, the elevator multi-source scheduling data in the multi-source scheduling data acquisition and preprocessing module is composed of elevator state data, passenger request data and floor passenger flow density data; the elevator state data records, in sequence, a current floor, a running direction, a door state and a load value; the passenger request data records, in sequence, a call floor, a target floor and a request timestamp; and the floor passenger flow density data is a real-time passenger count of a waiting area of each floor. The preprocessing includes, in sequence, an alignment operation based on a unified time reference, a missing data filling operation based on linear interpolation, a noise filtering operation based on a sliding window mean and a numerical normalization operation based on a minimum-maximum specification, and the output standardized scheduling input vector is spliced according to the sequence of the elevator state data, the passenger request data and the floor passenger flow density data.

[0020] In the embodiment, the step of calculating the skip probability and outputting the variable step length value in the event perception gate module includes: The current time step input vector in the standardized scheduling input vector set is read in a time sequence; The current time step input vector is sequentially spliced with a historical average waiting time index to generate a first spliced vector; the historical average waiting time index refers to a statistical result of an average user waiting time of the elevator system for all elevator requests within a set time sliding window, and is used to reflect a service response capability of the current elevator dispatching strategy within a period of time; The first spliced vector is continuously spliced with a historical average energy consumption index to generate an event-aware input vector; the historical average energy consumption index refers to a statistical result of an average user waiting time of the elevator system for all elevator requests within a set time sliding window, and is used to reflect a service response capability of the current elevator dispatching strategy within a period of time; The event-aware input vector is input into a linear transformation unit in the event-aware gate, and the linear transformation unit performs matrix multiplication and bias addition operations on the event-aware input vector to output an intermediate calculation result vector; The intermediate calculation result vector is input into an activation function unit, and the activation function unit performs Sigmoid function transformation on the intermediate calculation result vector to obtain a normalized skip probability value; The skip probability value is input into a variable step length mapping unit, and the skip probability value is mapped into an integer variable step length value k through a preset interval mapping function, and the variable step length value k is limited between 0 and K, wherein K is a preset maximum skip step length.

[0021] In the embodiment, the step of inputting the variable step length value into the improved SkipRNN and controlling the variable step length value through the variable step length control unit in the variable step length control module includes: The current time step variable step length value k is read, and the variable step length value k is input into the variable step length control unit in the improved SkipRNN; In the case that the variable step length value k is greater than zero, the variable step length control unit disables the hidden state update operation of the current time step, directly transmits the hidden state of the previous time step as the hidden state of the current time step, and writes the standardized dispatching input vector of the current time step into the skip input cache area, and then appends the standardized dispatching input vector of each time step to the skip input cache area in turn to form a skip input cache sequence with a length of k in the next k time steps; When the variable step length value k is equal to zero, the variable step length control unit triggers the hidden state update operation, inputs the standardized dispatching input vector of the current time step and the hidden state of the previous time step into the hidden state update unit together, performs recursive calculation, and generates the updated hidden state of the current time step; After the length of the skip input cache sequence reaches the variable step length value k, the skip input cache sequence is written into the skip attention summary unit, and the skip attention summary unit performs weight normalization operation based on the additive attention mechanism on the input vectors of each time step in the skip input cache sequence to generate a summary vector. In this embodiment, to solve the problem of missing input information during skipping, an additive attention mechanism-based skipping input summary method is introduced to extract key feature information from the skipping input cache sequence and generate a summary vector to compensate for the context information break caused by skipping hidden state update. The additive attention mechanism uses learnable parameters to weight and score the input vectors at each time step during skipping. Specifically, first, each input vector in the cache sequence and the query vector at the current time step are input into the additive attention scoring function together to calculate the attention score of each input vector. Then, all scores are normalized to generate normalized attention weights. Finally, the input vectors in the skipping cache sequence are weighted and summed according to the corresponding weights to generate a summary vector. The summary vector is a compact representation of the skipping interval information and is concatenated with the input vector at the current time step to update the hidden state. Through the additive attention mechanism described above, the present application can realize dynamic extraction and fusion of input information under the skipping update strategy, retain the computational efficiency advantage brought by the skipping strategy, improve the expression ability of the model for key input segments, and enhance the accuracy of the scheduling strategy and the robustness of the system.

[0022] The summary vector and the normalized scheduling input vector corresponding to the summary vector are concatenated as the composite input for the next effective hidden state update operation to compensate for the missing input information during skipping and update the improved SkipRNN hidden state. After updating, the skipping input cache area is emptied, and the next round of skipping control operation is waited.

[0023] In this embodiment, the improved SkipRNN described in the present application introduces three structural innovations of variable step length control, skipping input caching, and attention summary compensation in the skipping mechanism compared to the original SkipRNN. The original SkipRNN only determines whether to skip the current update based on the current input and hidden state through fixed gating logic, while the present application generates a dynamically variable skipping step length k through the event-aware gate, caches the continuous input sequence during skipping, and performs summary fusion on the cached sequence through the attention mechanism at the effective time step. This reduces the computational load while maintaining the ability to extract key information, improving the adaptability and decision-making accuracy of the model to complex dynamic scheduling scenarios.

[0024] In this embodiment, the hidden state update operation in the improved SkipRNN hidden state update module specifically includes: When the variable step length control unit receives a signal that the variable step length value k is equal to zero, the normalized scheduling input vector at the current time step is read At the same time, the hidden state of the previous time step is read ; standardized scheduling input vector with the hidden state The state update unit of the improved SkipRNN inputs the initial updated hidden state of the current time step, and the state update unit performs a weighted summation, a nonlinear activation, and a state transformation operation to generate a final updated hidden state ; It is judged whether the generated summary vector As exists in the skip input buffer, and when the summary vector As exists, the summary vector As and the current time step input vector are spliced to form a fusion input vector ; The fusion input vector and the hidden state of the previous time step are input into the state update unit again, and the weighted summation, the nonlinear activation, and the state transformation operation are repeatedly performed to overwrite the initial update result to generate a final updated hidden state after fusion ; The final hidden state is written into the hidden state register, and the skip input buffer and the summary vector buffer are emptied to prepare for the next time step skip control; In the embodiment, the hidden state register is used to store the hidden state generated by the improved SkipRNN at each time step, and is a key intermediate structure for realizing skip control and recursive state preservation. The register is used to write the latest generated hidden state after each execution of the hidden state update operation for the next time step, and the hidden state of the previous time step is kept unchanged during the skip, ensuring that the hidden state has sustainable tracking capability under skip control; Specifically, when the variable step value k output by the event-aware gate is equal to zero, the current time step hidden state generated by the state update unit will be immediately written into the hidden state register as the starting state for the next time step state calculation after triggering the hidden state update operation. When the variable step value k is greater than zero, the state update operation is skipped, and at this time the hidden state register does not occur. The system directly reads the hidden state saved in the register of the previous time step for transmission, avoiding unnecessary calculation. The register can be mapped to a high-priority buffer in the memory address space in the edge computing device, and has the characteristics of low read-write delay and high access frequency, which can support the high-frequency state calling demand in the fast reasoning scene. By setting the hidden state register, the application realizes the breakpoint reservation and lossless continuation of the recursive neural network state under the skip mechanism, supports the collaborative execution of skip and update operations, and enhances the engineering deployment feasibility of the model structure and the stability of the scheduling response.

[0025] In the case where the variable step value k is greater than zero, the hidden state update operation is skipped, and only the hidden state of the previous time step is directly passed to the current time step and used as the current hidden state.

[0026] In the embodiment, the generation of the target floor instruction sequence in the linear readout layer module specifically comprises: reading the updated hidden state of the current time step output by the improved SkipRNN, and inputting the hidden state into the full connection transformation unit of the linear readout layer; performing a linear mapping operation on the hidden state vector in the linear readout layer, generating an unnormalized instruction vector by matrix multiplication with a weight matrix and adding a bias vector, the weight matrix being a trainable two-dimensional real matrix, denoted as , the dimension being , wherein L is the total number of floors that can be reached by the elevator control system, and H is the dimension of the hidden state vector, the hidden state vector being input, first performing a matrix multiplication operation with the weight matrix , and then adding the bias vector to form an unnormalized instruction vector, each element in the unnormalized instruction vector corresponding to a candidate target floor scheduling score, and after an activation function such as a Softmax function, a normalized scheduling probability distribution is formed, and finally the scheduling target floor of the current time step is obtained through the maximum activation value index.

[0027] performing an activation function transformation on the unnormalized instruction vector to generate a normalized scheduling instruction vector; performing a maximum value selection operation on the normalized scheduling instruction vector to output the target floor number with the maximum activation value as the target scheduling floor of the current time step; on the basis of the target scheduling floor, in combination with the current elevator running direction, door state and load state information, a target floor instruction sequence is generated; writing the target floor instruction sequence into the task queue of the group control scheduling controller.

[0028] For example, in a 20-story office building, the elevator scheduling system is deployed with the improved SkipRNN model and the linear readout layer of the application, the system sets the hidden state vector dimension to 64, the total number of floors that can be reached by the elevator is 20, the weight matrix in the linear readout layer has a dimension of , and the bias vector has a dimension of .

[0029] When the model runs to a certain time step t, the current hidden state output by the SkipRNN is a real number vector with a dimension of 64 ; the system inputs The input is fed into the linear readout layer, and the following operations are performed: The score vector is calculated: ; where, is a 20-dimensional real number vector, representing the dispatch score for each target floor.

[0030] is fed into the Softmax activation function, generating a normalized dispatch probability distribution; Assuming the 12th layer has the highest score, i.e., ; The system determines that, under the current dispatch state, the elevator should be dispatched to the 12th floor first, and then generates a complete dispatch instruction sequence based on the current running direction "up", door state "closed", and load state "non-full load" as follows: Target floor: 12th floor; Running direction: up; Door operation: keep closed; Task queue action: insert priority dispatch position; The instruction sequence is immediately written to the group control dispatch controller, triggering the elevator control logic to execute the corresponding actions. The entire reasoning process is completed in the edge node, with an average delay of less than 150ms, meeting the real-time requirements.

[0031] In this embodiment, the generation of the multi-objective reward value in the group control dispatch controller interaction module specifically includes: Write the target floor instruction sequence to the task execution queue of the group control dispatch controller, triggering the controller to execute the elevator dispatch action; After each dispatch action is completed, receive the actual execution result data returned by the group control dispatch controller. The result data includes the execution start time, arrival time, number of passengers taking the elevator, corresponding floor record, motor running time and power record of the dispatch instruction; According to the call time and elevator arrival time of each passenger request in the actual execution result data, calculate the corresponding waiting time, and average the waiting time of all passenger requests within the set sliding time window to obtain the real-time average waiting time index of the current time window; According to the motor running power and running time in the actual execution result data, calculate the cumulative energy consumption value in the current time window, and divide by the window duration to obtain the real-time average energy consumption index; Calculate the difference between the real-time average waiting time index and the real-time average energy consumption index and the preset performance reference value to generate the waiting time deviation term and the energy consumption deviation term; Perform weighted summation operation on the waiting time deviation term and the energy consumption deviation term to generate the multi-objective reward value of the current time step.

[0032] ​In this embodiment, in order to realize the dynamic trade-off between service efficiency and energy consumption level of the elevator dispatching strategy, a multi-objective reward value generated based on real-time performance feedback is set, the multi-objective reward value is used as an input signal of the meta-reinforcement learning module, and is used for guiding the update of the event perception gate and the variable step mapping parameter, and the multi-objective reward value is composed of two parts: a waiting time deviation term and an energy consumption deviation term, which respectively measure the relative performance of the dispatching response performance and the energy efficiency level in the current time window.

[0033] Specifically, the system receives the execution result returned by the group control dispatching controller at each time step, calculates the average waiting time of all valid elevator requests in the current sliding time window , compares with the set target waiting time , and obtains the waiting time deviation term by difference calculation . According to the motor running power and the running time during the dispatching execution process, the average energy consumption per unit time is calculated , and the target energy consumption level is obtained by difference, and the energy consumption deviation term is obtained . The above two deviation terms are multiplied by adjustable weight coefficients and , and finally form the multi-objective reward value of the current time step , and the calculation formula is ; wherein and indicate the attention degree weight of the service response priority and the energy saving target, and satisfy , and can be dynamically configured according to the building dispatching strategy demand; The multi-objective reward value can reflect the contribution of the dispatching behavior to the user experience and the energy consumption optimization, has real-time and adjustability, and is an important quantitative index supporting the adaptive strategy evolution of the closed-loop dispatching optimization system of the application; by introducing the reward mechanism, the system can automatically strengthen the beneficial strategy and suppress the inefficient behavior according to the current dispatching effect, and improve the intelligence and stability of the overall dispatching system.

[0034] In this embodiment, the closed-loop dispatching optimization process in the meta-reinforcement learning head module specifically includes: The multi-objective reward value of the current time step is read, and the multi-objective reward value, the event perception input vector and the jump probability value of the corresponding time step form a state-action-reward triple; The state-action-reward triple is input into the policy update unit in the meta-reinforcement learning head, and the gradient of the event perception gate parameter is calculated according to the policy gradient optimization method; The gradients of the event-aware gate parameters are used to update the linear transformation weight matrix and bias vector in the event-aware gate, generating new event-aware gate parameters. After the event-aware gate parameters are updated, the difference between the skip probability value and the corresponding reward value is used as an advantage signal and input to the variable step size mapping parameter adjustment unit. The variable step size mapping parameter adjustment unit corrects the mapping function from the step probability to the step size interval based on the dominant signal, and adjusts the step length corresponding to the boundary value of each interval. In this embodiment, to achieve dynamic matching between the skip step length and the current scheduling performance, a variable step size mapping function adjustment mechanism based on the advantage signal is designed. This mechanism is applied to the probability-step size mapping process in the skip step control flow. The core of this mechanism is a set of learnable interval boundary parameters, which are used to normalize the skip step probability value output by the event-aware gate. Mapped to discrete integer step size values ,in This represents the maximum step size. Initially, the interval boundaries are defined as a monotonically increasing set of real numbers: ; The mapping function is defined as: if the skip step probability Falling into the range The corresponding step size ; In each scheduling feedback cycle, the system uses the multi-objective reward value obtained at the current time step. and the probability of skipping steps The corresponding actual step size The difference between the expected reward and the expected reward constitutes an advantage signal: ; in, Indicates the step size value The corresponding historical average reward; When the advantage signal When the current step length results in better-than-average performance, the system will increase the mapping probability range of that interval through a fine-tuning mechanism; conversely, if the current step length results in worse-than-average performance... If so, the range of that interval is narrowed, and the strategy is adjusted by adjusting the corresponding interval boundaries. Apply perturbation: ; in, The learning rate is used to control the magnitude of boundary adjustments. The interval boundary is adjusted and normalized to ensure that all boundaries meet the conditions of monotonic increase and fixed start and end range, and to maintain the integrity and effectiveness of the probability mapping function within the domain. This mechanism can realize the adaptive evolution of the step length distribution to the scheduling environment, automatically adjust the step strategy in different scenarios, and improve the dynamic response ability and long-term strategy optimization level of system performance. The updated event-aware gate parameters and variable step length mapping parameters are written into the parameter storage area, and the event-aware gate is returned to the step of calculating the skip probability for skip control calculation at the next time step. After a preset number of time steps, a round of parameter synchronization operation is performed to replace the parameters in the current event-aware gate and variable step length mapping module, to complete the strategy evolution period in a closed-loop scheduling optimization process.

[0035] Embodiment 1: In order to verify the feasibility of the application in implementation, the application is applied to the elevator group control system of a high-rise building for scheduling optimization experiment. The high-rise building has five elevators covering thirty floors. The system collects multi-source scheduling data of multiple elevators in real time through edge collection devices, including elevator running state, floor button request, up and down passenger number change, door opening and closing state, etc. To ensure data quality, the collected data is first aligned by timestamp, missing data is completed, and normalized processing is performed to construct a standardized scheduling input vector set, which is input into the event-aware gate module.

[0036] In actual operation, the event-aware gate receives the input vector and combines the historical average waiting time index and the historical average energy consumption index to dynamically calculate the skip probability and output the variable step length value. The skip value is used to control the state update rhythm of the improved SkipRNN, thereby skipping redundant state transitions and improving computational efficiency.

[0037] Experiments show that when the variable step length dynamically adjusts between 1~4, the elevator state prediction accuracy improves by about 7.3%, and the redundant calculation decreases by 34.8%. In the skip state, the system performs attention pooling operation on the cached multi-segment input vector to compensate for the input information during the skip period and enhance the hidden state expression ability.

[0038] The updated hidden state generates floor instructions through the linear readout layer and is sent to the group control scheduler. The system quantitatively analyzes the feedback information of the controller and the scheduling effect at the current time (real-time average waiting time and energy consumption), and outputs a multi-objective reward value. The reward value enters the meta-reinforcement learning head module, driving the optimization and update of the event-aware gate parameters and variable step length mapping parameters, realizing closed-loop adaptive scheduling optimization.

[0039] The experimental scene of the application particularly investigates the performance comparison between the rush hour (8:00-9:00 in the morning and 5:00-6:00 in the evening) and the ordinary time period (10:00-11:00 and 14:00-15:00). During the morning and evening rush hours, the traditional fixed step RNN scheduling system has a long response lag due to frequent state updates, resulting in an average waiting time of 56.8 seconds and an increase in passenger complaint rate. After using the application, the average waiting time is reduced to 39.2 seconds, the energy saving rate is increased to 17.6%, and the overall operation stability and user satisfaction are significantly improved.

[0040] The following table shows the performance comparison data of the traditional method and the method of the application in different time periods, including average waiting time, average energy consumption, passenger satisfaction score and other dimensions.

[0041] Table 1: Comparison of elevator scheduling optimization system performance evaluation table

[0042] The above table 1 shows the comparison of the traditional fixed step RNN scheduling method and the elevator scheduling optimization system based on the improved SkipRNN and meta-reinforcement learning mechanism proposed by the application in different time periods in terms of multiple key performance indicators, covering average waiting time, average single energy consumption, skip rate and passenger satisfaction score; In the morning rush hour of 8:00, the average waiting time of the traditional method is 56.8 seconds, the energy consumption is 14.7 kJ, the skip rate is 0, and the satisfaction score is only 2.8, which shows that the traditional method has a long response lag and low energy utilization efficiency under high load conditions. In contrast, the method of the application reduces the average waiting time to 39.2 seconds, the energy consumption to 12.1 kJ, while achieving a skip rate of 48.5%, and the satisfaction score is improved to 4.3, indicating that the system effectively alleviates the congestion problem during peak hours, improves the operation efficiency and user experience.

[0043] In the relatively stable noon period of 11:00, the method of the application also shows superior performance; the average waiting time is reduced from 34.7 seconds of the traditional method to 24.6 seconds, the energy consumption is reduced from 11.3 kJ to 9.8 kJ, the skip rate is increased to 52.1%, and the satisfaction score is increased to 4.6, which shows that even in non-peak hours, the system can reduce redundant calculation through the skip mechanism, achieve energy efficiency optimization and service improvement.

[0044] In the evening rush hour of 17:30, the traditional method again shows a long waiting time (58.3 seconds) and high energy consumption (15.0 kJ), while the application scheme shortens the waiting time to 40.5 seconds, reduces the energy consumption to 12.4 kJ, and maintains a skip rate of 46.9%, and the satisfaction score reaches 4.2, which shows that the system has stable scheduling performance and strong resource control ability under high load scenarios.

[0045] Through comprehensive analysis, the skip mechanism and the parameter adaptive optimization means of reinforcement learning proposed in the application show obvious scheduling effect better than the traditional scheme in multiple time periods, can flexibly adjust the calculation strategy under different load states, improve the scheduling response efficiency and reduce the system energy consumption, enhance the user experience and system economy.

[0046] The above describes only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art within the technical range disclosed by the present application, according to the technical scheme and the inventive concept of the present application, equivalent replacement or change, should be covered within the protection scope of the present application.

Claims

1. A real-time elevator scheduling and optimization system based on edge computing, characterized in that, include: The multi-source scheduling data acquisition and preprocessing module is used to acquire multi-source elevator scheduling data and perform preprocessing to generate a standardized set of scheduling input vectors. The event-aware gate module is used to receive a set of standardized scheduling input vectors, historical average waiting time indicators, and historical average energy consumption indicators, input them into the event-aware gate, calculate the skipping probability, and output a variable step size value. The variable step size control module is used to input the variable step size value into the improved SkipRNN, and control it through the variable step size control unit to generate the hidden state buffer sequence. The skip-step attention pooling module is used to perform attention pooling operations on the hidden state cache sequence to generate a summary vector; An improved SkipRNN hidden state update module is used to merge the summary vector with the normalized scheduling input vector of the next time step and update the hidden state of the improved SkipRNN. The linear readout layer module is used to input the updated hidden state into the linear readout layer to generate the target floor instruction sequence; The group control and dispatch controller interaction module is used to write the target floor instruction sequence into the group control and dispatch controller, receive the execution results fed back by the dispatch controller, calculate the real-time average waiting time index and the real-time average energy consumption index, and generate multi-objective reward values. The meta-reinforcement learning head module is used to input multi-objective reward values ​​into the policy optimization structure, update the event-aware gate parameters and variable step size mapping parameters, return to the skip step probability calculation step, and complete the closed-loop scheduling optimization process.

2. The elevator real-time scheduling and optimization system based on edge computing according to claim 1, characterized in that, The multi-source scheduling data acquisition and preprocessing module consists of elevator status data, passenger request data, and floor passenger density data. The elevator status data sequentially records the current floor, direction of travel, door status, and load value. The passenger request data sequentially records the calling floor, target floor, and request timestamp. The floor passenger density data is the real-time passenger count in the waiting area of ​​each floor. The preprocessing includes sequentially performing alignment operations based on a unified time reference, missing data imputation operations based on linear interpolation, noise filtering operations based on sliding window mean, and numerical normalization operations based on min-maximum norm. The output standardized scheduling input vector is concatenated according to the order of elevator status data, passenger request data, and floor passenger density data.

3. The elevator real-time scheduling and optimization system based on edge computing according to claim 2, characterized in that, The specific steps in the event-aware gate module for calculating the skip probability and outputting the variable step size value include: Read the current time step input vector from the standardized scheduling input vector set step by step in chronological order; The current time step input vector is sequentially concatenated with the historical average waiting time index to generate the first concatenated vector. The first concatenated vector is further concatenated with the historical average energy consumption index to generate an event-aware input vector. The event-aware input vector is input into the linear transformation unit in the event-aware gate. The linear transformation unit performs matrix multiplication and bias addition operations on the event-aware input vector and outputs the intermediate calculation result vector. The intermediate calculation result vector is input into the activation function unit, and the activation function unit performs a Sigmoid function transformation on the intermediate calculation result vector to obtain the normalized skip probability value. The skip probability value is input into the variable step size mapping unit, and the skip probability value is mapped to an integer variable step size value k through a preset interval mapping function. The variable step size value k is limited to between 0 and K, where K is the preset maximum skip step size.

4. The elevator real-time scheduling and optimization system based on edge computing according to claim 3, characterized in that, The steps in the variable step size control module that input the variable step size value into the improved SkipRNN and control it through the variable step size control unit specifically include: Read the variable step size value k of the current time step and input the variable step size value k into the variable step size control unit in the improved SkipRNN; When the variable step size value k is greater than zero, the variable step size control unit disables the hidden state update operation of the current time step, directly passes the hidden state of the previous time step as the hidden state of the current time step, writes the normalized scheduling input vector of the current time step into the skip input buffer, and then appends the normalized scheduling input vector of each time step to the skip input buffer in the next k time steps, forming a skip input buffer sequence of length k. When the variable step size value k equals zero, the variable step size control unit triggers the hidden state update operation, inputs the normalized scheduling input vector of the current time step and the hidden state of the previous time step into the hidden state update unit, performs recursive calculation, and generates the updated hidden state of the current time step. After the length of the skip input buffer sequence reaches the variable step size value k, the skip input buffer sequence is written into the skip attention summarization unit. The skip attention summarization unit performs a weight normalization operation based on the additive attention mechanism on the input vector of each time step in the skip input buffer sequence to generate a summary vector. The summary vector is concatenated with the standardized scheduling input vector of the time step corresponding to the summary vector, and used as the composite input for the next effective hidden state update operation. This is used to compensate for the missing input information during the skip step and update the hidden state of the improved SkipRNN. After the update is complete, clear the step input buffer and wait for the next round of step control operation.

5. The elevator real-time scheduling and optimization system based on edge computing according to claim 4, characterized in that, The hidden state update operation in the improved SkipRNN hidden state update module specifically includes: When the variable step size control unit receives a signal that the variable step size value k is equal to zero, it reads the normalized scheduling input vector of the current time step. Simultaneously read the hidden state from the previous time step. ; Standardize the scheduling input vector With hidden state Input the state update unit of the improved SkipRNN. The state update unit performs weighted summation, nonlinear activation, and state transition operations to generate the initial updated hidden state for the current time step. ; Determine if a generated summary vector Δs exists in the jump step input buffer. If the summary vector Δs exists, combine it with the current time step input vector. The vectors are concatenated to form a fused input vector. ; fuse input vectors Hidden state from the previous time step Input the state update unit again, repeat the weighted summation, nonlinear activation, and state transition operations, overwrite the initial update result, and generate the final hidden state after fusion update. ; The final hidden state Write to the hidden state register, clear the jump input buffer and the digest vector buffer, and prepare for the jump control of the next time step; If the variable step size k is greater than zero, skip the hidden state update operation and only update the hidden state of the previous time step. It is passed directly to the current time step and used as the current hidden state.

6. The elevator real-time scheduling and optimization system based on edge computing according to claim 5, characterized in that, The generation of the target floor instruction sequence in the linear readout layer module specifically includes: Read the hidden state updated at the current time step by the output of the improved SkipRNN, and input the hidden state into the fully connected transformation unit of the linear readout layer; In the linear readout layer, a linear mapping operation is performed on the hidden state vector. By performing matrix multiplication with the weight matrix and adding the bias vector, an unnormalized instruction vector is generated. The activation function transformation is performed on the unnormalized instruction vector to generate a normalized scheduling instruction vector; Perform a maximum value selection operation on the normalized scheduling instruction vector and output the target floor number with the maximum activation value as the target scheduling floor for the current time step. Based on the target floor, and combined with the current elevator running direction, door status and load status information, a target floor instruction sequence is generated; Write the target floor instruction sequence into the task queue of the group control and scheduling controller.

7. The elevator real-time scheduling and optimization system based on edge computing according to claim 6, characterized in that, The generation of multi-objective reward values ​​in the group control scheduling controller interaction module specifically includes: Write the target floor instruction sequence into the task execution queue of the group control and scheduling controller, triggering the controller to execute elevator scheduling actions; After each scheduling action is completed, receive the actual execution result data returned by the group control scheduling controller; Based on the actual execution results data, the corresponding waiting time is calculated for each passenger's requested elevator call time and elevator arrival time. The waiting time of all passengers' requests is averaged within the set sliding time window to obtain the real-time average waiting time index for the current time window. Based on the motor operating power and operating time in the actual execution results data, calculate the cumulative energy consumption value within the current time window, divide it by the window duration, and obtain the real-time average energy consumption index. The difference between the real-time average waiting time index and the real-time average energy consumption index and the preset performance reference value is calculated to generate waiting time deviation item and energy consumption deviation item. Perform a weighted summation operation on the waiting time deviation term and the energy consumption deviation term to generate the multi-objective reward value for the current time step.

8. The elevator real-time scheduling and optimization system based on edge computing according to claim 7, characterized in that, The closed-loop scheduling optimization process in the meta-reinforcement learning head module specifically includes: Read the multi-objective reward value of the current time step, and combine the multi-objective reward value with the event-aware input vector and the step-skip probability value of the corresponding time step to form a state-action-reward triplet; The state-action-reward triplet is input into the policy update unit in the meta-reinforcement learning head, and the gradient of the event-aware gate parameters is calculated according to the policy gradient optimization method. The gradients of the event-aware gate parameters are used to update the linear transformation weight matrix and bias vector in the event-aware gate, generating new event-aware gate parameters. After the event-aware gate parameters are updated, the difference between the skip probability value and the corresponding reward value is used as an advantage signal and input to the variable step size mapping parameter adjustment unit. The variable step size mapping parameter adjustment unit corrects the mapping function from the step probability to the step size interval based on the dominant signal, and adjusts the step length corresponding to the boundary value of each interval. Write the updated event-aware gate parameters and variable step size mapping parameters into the parameter storage area, and return the event-aware gate to the step of calculating the skip probability for use in the skip control calculation of the next time step; After each preset number of time steps, a round of parameter synchronization operation is performed to replace the parameters in the current event-aware gate and variable step size mapping module, so as to complete the strategy evolution cycle in a closed-loop scheduling optimization process.

Images