A transporter scheduling method based on dynamic probability allocation mechanism

Abstract

The present invention belongs to the field of hospital scheduling and relates to a transporter scheduling method based on a dynamic probability allocation mechanism, comprising: extracting historical transport data to obtain a transport time matrix; obtaining a patient transport matrix based on current patient transport data; processing the transporter data based on the patient's scheduling tendency to obtain an initial transporter scheduling matrix; the scheduling tendency is related to the transporter selection for the first transport task and the second transport task; the transporter data includes the transporter's workload and current location; calculating the transporter's scheduling information in the initial transporter scheduling matrix, and dynamically updating the transporter scheduling matrix based on the scheduling information through a dynamic transfer probability model to obtain a final transporter scheduling matrix; the method aims to solve the pain points of unbalanced workload of central transporters and long waiting time for patient examinations under traditional experience-based scheduling, and realize the transformation of central transporter scheduling management from an experience-driven to a data-driven paradigm.

Description

Technical Field

[0001] The present invention relates to the field of hospital scheduling, and particularly discloses a transporter scheduling method based on a dynamic probability allocation mechanism. Background Art

[0002] In modern hospital operations, transporting patients from wards to examination rooms (such as CT and MRI rooms), as well as returning after examinations, is a crucial component of the hospital's central transportation system. Its task allocation and scheduling directly impact the continuity of the patient care process and the quality of medical services. However, current central transportation scheduling suffers from the following issues: Experience-driven: Existing central transporter scheduling and scheduling models primarily rely on manual experience or static rules. While these methods can meet basic scheduling needs to a certain extent, they suffer from numerous limitations in practical application. On the one hand, manual experience fails to fully cover all possible scenarios and is easily influenced by subjective factors, resulting in a lack of scientific basis for scheduling decisions. On the other hand, static rules lack flexibility and cannot adapt to dynamically changing task demands and unexpected situations, such as the insertion of urgent tasks or temporary changes in transporters. Therefore, existing scheduling models lack flexibility in task allocation, balanced resource utilization, and optimized patient wait times, making them unable to meet the requirements of modern hospitals for efficient and accurate scheduling. Unbalanced task distribution: Tasks are unevenly distributed among central transporters, leaving some overloaded and others under-tasked, leading to inefficient resource utilization. This unbalanced task distribution not only impacts transporters' work enthusiasm but also potentially extends patient wait times, impacting the hospital's overall operational efficiency. Long patient wait times: Existing scheduling methods lack optimal consideration for patient wait times, resulting in long wait times for patient examinations and a negative impact on the hospital's overall service quality. Especially during peak hours, increased patient wait times can lead to decreased patient satisfaction, impacting the hospital's overall service quality. Furthermore, prolonged wait times can increase patient anxiety and negatively impact their physical and mental health. Incomplete consideration of scheduling factors: The task allocation process fails to fully account for key scheduling factors, such as the transporter's current location and the time between tasks, impacting the scientific and rational nature of the scheduling. For example, failure to consider the transporter's real-time location can lead to irrational task allocation, increasing wasted travel time for the transporter; failure to consider the time between tasks can lead to poor task transitions, impacting the continuity of patient examinations. Furthermore, existing scheduling methods fail to adequately consider the priority and urgency of tasks, resulting in important tasks being delayed.

[0003] In view of this, the present invention provides a transporter scheduling method based on a dynamic probability allocation mechanism. Based on key technologies such as calculating the transporter transfer probability, dynamically optimizing patient task allocation, and intelligently generating work schedules, it aims to solve the pain points of unbalanced workload of central transporters and long waiting time for patient examinations under traditional experience-based scheduling, and realize the transformation of central transporter scheduling management from experience-driven to data-driven paradigm. Summary of the Invention

[0004] The object of the present invention is to provide a haulier scheduling method based on a dynamic probability allocation mechanism. The specific scheme is as follows, including: extracting historical transportation data to obtain a transportation time matrix; the transportation time matrix includes a first transportation time from a nursing unit to an examination room, an examination time, and a second transportation time from an examination room to a nursing unit; constructing a patient transportation matrix based on a future patient transportation task list; the patient transportation matrix includes a first transportation task from a nursing unit to an examination room, an examination task, and a second transportation task from an examination room to a nursing unit; processing the haulier data based on the scheduling tendencies of patients and hospital transportation managers to obtain an initial haulier scheduling matrix; the scheduling tendencies are related to the haulier selection for the first transportation task and the second transportation task; the haulier data includes the haulier's workload and current location; calculating the haulier's scheduling information in the initial haulier scheduling matrix, and dynamically updating the haulier scheduling matrix based on the scheduling information through a dynamic transition probability model to obtain a final haulier scheduling matrix.

[0005] Furthermore, the transporter data is processed based on the patient's scheduling preference to obtain an initial transporter scheduling matrix, including: determining whether the patient chooses the same transporter to perform the first transport task and the second transport task; if the same transporter is selected, binding the transporters of the patient's first transport task and the second transport task; if different transporters are selected, independently assigning the transporters of the first transport task and the second transport task; generating the initial transporter scheduling matrix based on the transporter binding results of the first transport task and the second transport task.

[0006] Furthermore, the method calculates the dispatch information of the transporters in the initial transporter dispatch matrix, and dynamically updates the transporter dispatch matrix based on the dispatch information through a dynamic transfer probability model, including: determining the dispatch transporter data based on the dispatch information; the dispatch transporter data includes the transporter's workload, task waiting time and current location; constructing a dynamic transfer probability model based on the dispatch transporter data, and calculating the transport transfer probability of the patient to the transporter; sorting the transporters based on the transport transfer probability to obtain a transfer probability matrix of the transporter to the patient; allocating multiple patients to multiple transporters based on the transfer probability matrix to obtain a transport allocation strategy; updating the transporter dispatch matrix based on the transport allocation strategy to obtain a final transporter dispatch matrix.

[0007] Furthermore, the updating of the courier scheduling matrix to obtain the final courier scheduling matrix includes: constructing a parent chromosome based on the updated courier scheduling matrix; generating multiple chromosomes based on the parent chromosome to form a population; each chromosome corresponds to a transportation allocation strategy; constructing a fitness function based on the courier's workload and the average waiting time of the patient, and calculating the fitness value of each chromosome; updating the parent chromosome based on the fitness value, and repeating the screening operation on the parent chromosome until the loop termination condition is met.

[0008] Furthermore, the parent chromosome is constructed based on the updated courier scheduling matrix, including: for patients who choose the same courier for transportation, paired gene encoding is used; the courier of the paired gene encoding remains consistent; the gene that matches the patient and the courier is encoded as 1, and the gene that does not match the patient and the courier is encoded as 0.

[0009] Furthermore, the fitness function is:

[0010] ;

[0011] ;

[0012] ;

[0013] in, represents the fitness function; It represents the working imbalance coefficient; represents the average waiting time of patients; i represents the transporter variable; N Indicates the total number of transporters; Indicates transporter i Current accumulated working hours; represents the average working hours of transporters; Indicates the total number of patients who need to wait; j represents patient variables; Indicates the j The time when each patient is scheduled to be examined; Indicates transport j The end time of each patient.

[0014] Furthermore, the updating of the parent chromosome includes: calculating the transport transfer probability of the patient to the transporter in each transport allocation strategy in the current parent chromosome according to a dynamic transfer probability model; determining the mutation probability of each gene position in the chromosome based on the transport transfer probability; and mutating the parent chromosome based on the mutation probability to update the parent chromosome to obtain a new parent chromosome.

[0015] Furthermore, the dynamic transition probability model is related to the transporter's workload, task time interval and starting transport distance.

[0016] Furthermore, the expression of the dynamic probability model is:

[0017] ;

[0018] in, represents the transport transfer probability; exp represents the exponential function; Indicates the m The total duration of tasks assigned to each transporter during the current scheduling period; Indicates transporter m The time interval between tasks to transport the i-th patient; Indicates transporter m Reaching the patient from the current location i the time required for processing; Indicates the total duration weight; represents the time interval weight; represents the distance weight; represents an empty cell set.

[0019] Furthermore, the task arrangements of patients and transporters are presented through a Gantt chart, and the scheduling decision-making process is presented through a dynamic transfer probability cellular grid diagram.

[0020] The present invention has the following advantages and beneficial effects:

[0021] The central transporter scheduling method based on the dynamic probability allocation mechanism provided by the present invention can significantly shorten the patient waiting time.

[0022] The present invention optimizes the patient examination process through a dynamic probability model and task interval adjustment strategy, significantly shortens the patient's waiting time, and improves the patient's medical experience.

[0023] The present invention introduces an inverse workload factor to avoid intensive tasks for some transporters and sparse tasks for others, thereby achieving a balanced allocation of transport resources and improving resource utilization efficiency.

[0024] The present invention takes the geographical location of the transporter into consideration when allocating tasks, prioritizes the shortest path, reduces the transporter's ineffective movement time and patient waiting time, and improves scheduling efficiency.

[0025] The present invention can record the complete scheduling path and probability evolution process, support continuous iterative upgrades of the algorithm, and ensure the scientificity and rationality of the scheduling method.

[0026] The present invention supports displaying the scheduling process and effects in the form of Gantt charts, transfer probability cellular grids, etc., which facilitates managers to review and adjust strategies, and improves the scientific nature and transparency of management decisions.

[0027] The present invention is not only applicable to the scheduling of central transporters for transporting patients undergoing CT and MRI examinations, but is also applicable to the scheduling of other examinations or the transport of goods, and has strong versatility and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 An exemplary flow chart of a transporter scheduling method based on a dynamic probability allocation mechanism provided by the present invention;

[0029] Figure 2 An exemplary Gantt chart of a patient's task schedule provided by the present invention;

[0030] Figure 3 An exemplary Gantt chart of a transporter's task schedule provided by the present invention;

[0031] Figure 4 An exemplary grid diagram of dynamic transition probabilities provided by the present invention. DETAILED DESCRIPTION

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0033] Figure 1 This is an exemplary flow chart of a transporter scheduling method based on a dynamic probability allocation mechanism provided by the present invention. Figure 1 As shown, the transporter scheduling method based on the dynamic probability allocation mechanism provided by the present invention aims to optimize the allocation of transport tasks in the patient examination process within the hospital, balance the workload of transporters, and reduce patient waiting time. It includes the following contents:

[0034] Extract historical transportation data and obtain the transportation time matrix.

[0035] Historical data refers to the recorded data of transportation tasks in the past period of time. For example, historical transportation data may include the transporters, transportation starting points, transportation destinations, and transportation times of all transportation tasks in the hospital's central transportation system in the past year. The transportation time matrix refers to the transportation situation within the hospital stored in a tabular form. The transportation time matrix may include the first transportation time from the nursing unit to the examination room, the examination time, and the second transportation time from the examination room to the nursing unit. The row dimension of the transportation time matrix is ​​the nursing unit, and the column dimension is the examination room. The nursing unit can refer to a ward. The examination room can refer to a place where patients are examined. For example, a CT examination room and an MRI examination room. The first transportation time refers to the regular transportation time from the nursing unit to the examination room. The examination time refers to the time when the patient is examined in the examination room. The second transportation time refers to the regular transportation time from the examination room to the nursing unit.

[0036] Specifically, data extraction can be achieved through the read_time_data function code. The key logic of the code is as follows:

[0037] File path and existence check: exist(file_path, 'file') strictly verifies whether the file exists to avoid subsequent reading failures due to path errors. T1T3 worksheet analysis: readcell reads the original cell data and retains the row and column structure. data.wards extracts the ward list (such as ["01 Nursing Unit", "02 Nursing Unit", ...]). data.check_rooms extracts the examination room list (such as ["CT01 Examination Room", "MRI02 Examination Room", ...]). str2double converts the transportation time in string format to a numerical matrix, such as matlab. T2 worksheet analysis: detectImportOptions automatically detects the table format, and VariableNamingRule is set to 'preserve' to ensure that column names (such as "Examination Name") are not converted to legal variable names (such as the "x_" prefix). check_times stores the standard time consumption of the examination items (such as 30 minutes for a general CT scan and 45 minutes for a general MRI examination).

[0038] Based on the future patient transport task list, a patient transport matrix is ​​constructed.

[0039] The future patient transport task list includes the transport needs in the future. By extracting the patient's scheduled examination information, the current patient transport data can be obtained. For example, if a patient needs to be transported the next day, a task list containing T1 、 T2 and T3The current patient transportation data of the patient task list may include the patient ID, transportation task chain, time of the scheduled examination and transportation time. The transportation task chain includes the transportation starting point and transportation end point. The patient transportation matrix refers to the transportation and examination task information of all current patients stored in the form of a matrix. The patient transportation matrix includes the first transportation task T1 from the nursing unit to the examination room, the examination task T2 and the second transportation task T3 from the examination room to the nursing unit. The first transportation task refers to the task of transporting the patient from the nursing unit to the examination room. The examination task refers to the task of the patient undergoing a specific examination item in the examination room. The second transportation task refers to the return task of transporting the patient from the examination room back to the nursing unit.

[0040] Specifically, the patient transport matrix is ​​generated through the generate_task_table function. The key logic of the code is as follows: Patient task chain generation logic: T1 Task (Ward → Examination Room): Reverses the appointment time to ensure the transporter arrives at the examination room 5 minutes before the appointment. For example: Appointment time: 2023-10-05 09:00; Transport time (including redundancy): 20 minutes; T1 Task duration: 08:35-08:55. T2 Task (Examination): Starts at the appointment time and lasts for the standard duration of the examination item. For example: T2 Task duration: 09:00-09:30 (CT scan, 30 minutes). T3 Task (Examination Room → Ward): Mandatory binding to T1 Task, using the same transporter. For example: T3 Task duration: 09:30-09:50 (Return transport, 20 minutes). Time formatting: split_datetime: Parses the appointment time string into date and time components, supporting various delimiters (such as spaces and slashes). split_time: Extracts the hours and minutes from the time, compatible with formats containing seconds (e.g., 08:30:00). Exception handling: Detects tasks that span multiple days: If the start time of a T1 task is before midnight (T1_start_total < 0), an error message is thrown. Data consistency check: In the match_indices function, verify that the ward and examination room names exist in the time matrix to avoid invalid indexes.

[0041] Based on the scheduling preferences of patients and hospital transport managers, the transporter data is processed to obtain the initial transporter scheduling matrix.

[0042] A hospital transport administrator can be the person responsible for managing patient transport, such as a central administrator or nurse. Scheduling preference refers to the preferences of patients and hospital transport administrators regarding transport task allocation strategies. Hospital transport administrators have higher authority than patients, and the transporters assigned by hospital transport administrators cannot be changed. For transports not assigned by the hospital transport administrator, patients can determine whether to use the same transporter. Users can interactively determine this preference through a graphical user interface (GUI) on a user terminal (e.g., a mobile phone). Scheduling preference is related to the selection of transporters for the first and second transport tasks. Specifically, a dialog box is popped up to prompt the user to select a scheduling strategy. This dialog box is popped up through the questdlg function, prompting the user to select a scheduling strategy. The user can choose to schedule with the same transporter or with different transporters. This interactive design enables the scheduling system to flexibly adjust scheduling strategies based on different needs, improving system applicability and user satisfaction. Transporter data refers to structured information describing the current work status and attributes of transporters. The initial transporter scheduling matrix records the task allocation and time schedules of all transporters. Transporter data includes the transporter's workload and current location. Workload refers to the cumulative working time of a transporter's currently assigned tasks and is used to measure the transporter's task saturation. The current location refers to the specific location of the transporter at the time of dispatch.

[0043] In some embodiments, processing the transporter data based on the patient's scheduling preferences to generate an initial transporter scheduling matrix includes: determining whether the patient has selected the same transporter for both the first and second transport tasks; if the patient has selected the same transporter, binding the transporters for the patient's first and second transport tasks; if different transporters have been selected, independently assigning the transporters for the patient's first and second transport tasks; and generating the initial transporter scheduling matrix based on the transporter binding results for the first and second transport tasks. For example, a particular patient can be assigned a transporter in a fixed manner. Specifically, assignment can be based on area, with a transporter assigned to a specific hospital building; assignment can be based on shifts, with staff assigned according to fixed shifts; assignment can be based on task order, with tasks assigned to transporters in the order of their appointment times. Specifically, based on the scheduling strategy selected by the user, the corresponding scheduling function is called: if the user selects the same transporter for scheduling, the schedule_same_tasks function is called; if the user selects different transporters for scheduling, the schedule_tasks function is called. These functions are based on dynamic probability models, taking into account key factors such as the transporter's current location, task interval time, workload, etc., dynamically optimize task allocation, and generate scheduling results and transfer probability history records.

[0044] Specifically, a dialog box pops up to prompt the user to select a scheduling strategy: a questdlg function is used to pop up a graphical user interface (GUI) to prompt the user to select a scheduling strategy, and the user can choose "dispatched by the same transporter" or "dispatched by different transporters".

[0045] Based on the user's selection, the system calls the corresponding scheduling function. If the user selects "Scheduled by the same transporter", the schedule_same_tasks function is called; if the user selects "Scheduled by different transporters", the schedule_tasks function is called.

[0046] Code implementation:

[0047] % Call the corresponding scheduling function according to the scheduling strategy selected by the user

[0048] if strcmp(choice, 'yes')

[0049] [schedule, transfer_prob_history] = schedule_same_tasks(patients,transporters);

[0050] elseif strcmp(choice, 'no')

[0051] [schedule, transfer_prob_history] = schedule_tasks(patients,transporters);

[0052] else

[0053] error('No valid scheduling strategy selected!');

[0054] end

[0055] The schedule_same_tasks function implements the logic of scheduling by the same courier. It takes into account key factors such as the courier's workload, task waiting time, current location, etc., and dynamically optimizes task allocation through a dynamic transfer probability model. The key point of the scheduling logic - Dynamic Transfer Probability Model: The scheduling module is based on the dynamic transfer probability model. Transfer probability is a key concept that is used to dynamically assign transportation tasks to different couriers. This probability is calculated based on multiple factors and aims to optimize task allocation, ensure a balanced workload for couriers, and reduce patient waiting time. Task interval rule: The scheduling module applies a 5-minute interval rule to ensure a reasonable time interval between tasks and to guarantee the travel time between the central courier's locations.

[0056] The schedule_tasks function implements the logic for scheduling tasks by different drivers. It also considers key factors such as the driver's workload, task wait time, and current location, dynamically optimizing task allocation through a dynamic transition probability model. The basic logic of the code is the same as the schedule_same_tasks function, but without the restriction that T3 tasks can only be completed by the same person.

[0057] The transporter scheduling information in the initial transporter scheduling matrix is ​​calculated. Based on this information, the transporter scheduling matrix is ​​dynamically updated using a dynamic transition probability model to obtain the final transporter scheduling matrix. Scheduling information refers to the core parameter set used to calculate the transporter task assignment priority. Scheduling information is related to the transporter's shift schedule. For example, scheduling information may include the patient ID that the transporter needs to transport, the transport start time, transport time, transport route, transport start location, transport end location, and transport end time.

[0058] In some embodiments, calculating the transporter scheduling information in the initial transporter scheduling matrix and dynamically updating the transporter scheduling matrix based on the scheduling information using a dynamic transition probability model includes: determining scheduled transporter data based on the scheduling information. Scheduled transporter data refers to structured information describing the real-time work status of the transporter. The scheduled transporter data may include, for example, the transporter's workload, task wait time, and current location. Task wait time refers to the time it takes for the transporter to arrive at the transport starting point and wait to transport the patient. A dynamic transition probability model is constructed based on the scheduled transporter data to calculate the transport transfer probability of the patient to the transporter. The transport transfer probability may refer to the probability of the patient being assigned to the transporter. Transporters are sorted based on the transport transfer probability to obtain a transporter-to-patient transition probability matrix. The transition probability matrix is ​​a matrix-like representation of the patient-to-transporter transition probability, with rows and columns representing the patient-to-transporter correspondence. Based on the transition probability matrix, multiple patients are assigned to multiple transporters to obtain a transport allocation strategy. The transport allocation strategy is a task allocation plan generated based on the transition probability matrix, which specifies the transporter and task time corresponding to each patient. For example, transporters can be sorted by transfer probability, and the transporter with the highest transfer probability for a patient can be assigned to that patient. Based on the transport assignment strategy, the transporter scheduling matrix is ​​updated to obtain the final transporter scheduling matrix. For example, each transporter can be assigned a preset number of patients based on their transfer probability.

[0059] In some embodiments, the dynamic transfer probability model is related to the transporter's workload, task time interval and starting transport distance. Task time interval refers to the time interval between the end time of the transporter's previous transport task and the start time of the next transport task. Starting transport distance refers to the distance between the end position of the transporter's previous transport task and the start position of the next transport task. The workload is calculated based on the transporter's current accumulated working time. The higher the workload, the lower the allocation probability; the time interval from the transporter completing the current task to the next task, the shorter the waiting time, the higher the allocation probability; the moving time from the transporter's current position to the starting point of the task, the closer the distance, the higher the allocation probability. Specifically, the expression of the dynamic probability model is:

[0060] ;

[0061] in, represents the transport transfer probability; exp represents the exponential function; Indicates the m The total duration of tasks assigned to each transporter during the current scheduling period; Indicates transporter m Transport i The time interval between tasks for each patient; Indicates transporter m The time required to reach patient i from the current location; Indicates the total duration weight; represents the time interval weight; represents the distance weight; represents the empty cell set, that is, the set of all transporters currently to be assigned; ∑ is the sum of the empty cell set. Among them, the total duration weight, time interval weight, and distance weight can be 1.

[0062] In some embodiments, updating the courier scheduling matrix to obtain a final courier scheduling matrix includes:

[0063] Based on the updated courier scheduling matrix, a parent chromosome is constructed. The matching relationship between the courier and the patient task is mapped to a gene sequence, and each gene position represents the assignment relationship of "patient ID → courier ID". In some embodiments, the construction of the parent chromosome based on the updated courier scheduling matrix includes: for patients who choose the same courier for transportation, paired gene encoding is used; the courier of the paired gene encoding remains consistent; the gene encoding of the patient and the courier is 1, and the gene encoding of the patient and the courier is 0.

[0064] Based on the parent chromosome, multiple chromosomes are generated to form a population; each chromosome corresponds to a transportation allocation strategy. By randomly perturbing the initial scheduling matrix, multiple chromosomes are generated to form the initial population.

[0065] Based on the workload of the transporter and the average waiting time of the patient, a fitness function is constructed and the fitness value of each chromosome is calculated. The pros and cons of the chromosome corresponding scheduling scheme are quantitatively evaluated, integrating the workload balance and the patient waiting time. In some embodiments, the fitness function is:

[0066] ;

[0067] ;

[0068] ;

[0069] in, represents the fitness function; Indicates the working imbalance coefficient, the unit is time; represents the average waiting time of patients; i represents the transporter variable; N Indicates the total number of transporters; Indicates transporter i Current accumulated working hours; represents the average working hours of transporters; Indicates the total number of patients who need to wait; j represents patient variables; Indicates the j The time when each patient is scheduled to be examined; Indicates transport j The calculate_transporter_work_time function calculates the cumulative working time of each transporter to assess whether the transporter's workload is balanced. The calculate_patients_waiting_time function calculates the average waiting time of patients, providing a quantitative indicator of scheduling efficiency and helping to evaluate the optimization of scheduling methods for patient waiting time.

[0070] Specifically, calculate the cumulative working time and work imbalance coefficient of the transporter. Use the calculate_transporter_work_time function to calculate the cumulative working time of each transporter. This function traverses the schedule table, counts the task duration of each transporter, and calculates their cumulative working time. Its output transporter_work_time is an array containing the cumulative working time of each transporter. The first column of the array is the transporter ID, and the second column is the corresponding cumulative working time (unit: hour). This indicator helps evaluate whether the transporter's workload is balanced and whether some transporters are overloaded. This function also calculates the work imbalance coefficient , used to assess the balance of transporter workloads. A lower workload imbalance coefficient indicates a more balanced workload distribution. Use the calculate_patients_waiting_time function to calculate the average patient waiting time. This function traverses the schedule table, counts the actual waiting time for each patient, and calculates the average waiting time. Its output, patients_waiting_time, is a numeric value representing the average patient waiting time (in hours). If no records meet the criteria, NaN is returned. This metric helps evaluate the effectiveness of scheduling methods in optimizing patient waiting times, ensuring that patients receive timely examinations.

[0071] Based on the fitness value, the parent chromosome is updated, and the screening operation on the parent chromosome is repeated until the loop termination condition is met. The roulette method is used to select the parent chromosome according to the fitness ratio, and invalid solutions with inconsistent IDs of the transporters in the task cluster are excluded. Updating the parent chromosome may refer to performing a crossover mutation operation on the selected chromosome. The loop termination condition may include that the fitness value is less than a preset threshold or the number of iterations reaches the maximum number of iterations. In some embodiments, the updating of the parent chromosome includes: calculating the transport transfer probability of the patient to the transporter in each transport allocation strategy in the current parent chromosome according to the dynamic transfer probability model; determining the mutation probability of each gene position in the chromosome based on the transport transfer probability; and mutating the parent chromosome based on the mutation probability to obtain a new parent chromosome. The transporter with a higher transport transfer probability has a higher probability of being selected during the mutation, so that the mutation direction is optimized towards the goals of load balancing and reducing the distance between the transporter and the patient.

[0072] The result display module is an important part of the present invention. It uses visualization tools to intuitively display the scheduling results, helping managers and schedulers to quickly understand the task allocation and scheduling decision-making process. Figure 2 、 Figure 3 and Figure 4 As shown, the transporter scheduling method based on the dynamic probability allocation mechanism provided by the present invention also includes displaying the task arrangement of patients and transporters through a Gantt chart and displaying the scheduling decision process through a dynamic transition probability grid diagram. Figure 2 As shown, the transportation arrangement for each patient includes T1 first transportation task time, T2 examination task time and T3 second transportation task time. Figure 3As shown, each transporter's task schedule includes multiple patients. Transporter 1 transports, from left to right, the T1 time of patient 3, the T1 time of patient 11, the T3 time of patient 3, the T3 time of patient 11, the T1 time of patient 8, the T1 time of patient 30, the T1 time of patient 4, the T3 time of patient 8, the T3 time of patient 30, and the T3 time of patient 4. Transporter 2 transports, from left to right, the T1 time of patient 1, the T3 time of patient 1, the T1 time of patient 15, the T1 time of patient 6, the T1 time of patient 48, the T1 time of patient 5, the T3 time of patient 6, the T3 time of patient 48, the T3 time of patient 15, and the T3 time of patient 5. The tasks for the other transporters are similar to those for transporters 1 and 2, except for the differences in the assigned transport times, transported patients, and transport tasks. The filled squares in the figure are the actual working time, the unfilled squares are the non-waiting time, the white squares are the time when the patient was not transported to the examination room, and the gray squares are the time when the patient was transported back to the ward.

[0073] Use Gantt charts to display the task arrangements of patients and transporters. Use the plot_gantt_chart and plot_transporter_gantt_chart functions to draw Gantt charts of patients and transporters' tasks, which intuitively display the start time, end time, and task type of the task, helping managers and dispatchers quickly understand the task allocation situation.

[0074] Draw a dynamic transfer probability cell grid diagram. Use the plot_dynamic_transfer_probability_grids function to draw a dynamic transfer probability cell grid diagram to show the changes in transfer probability during the scheduling decision process, making the scheduling process more transparent and traceable.

[0075] Use a Gantt chart to display the patient and transporter task schedules: A Gantt chart is used to display the task schedules for patients and transporters, including the start and end times and task types. Use the plot_gantt_chart function to create a patient task Gantt chart. This function generates a patient task Gantt chart based on the data in the schedule table, showing each patient's task schedule. Figure 2 Create a timeline diagram for the patient examination process. Use the plot_transporter_gantt_chart function to create a Gantt chart of transporter tasks. This function generates a Gantt chart of transporter tasks based on the data in the schedule table, showing the task schedule for each transporter. Figure 3 Create a Gantt chart for the transporter tasks to show their progress. These functions plot the start and end times of tasks, along with the task type, to help managers and schedulers quickly understand the distribution of tasks.

[0076] Draw a dynamic transfer probability grid: A dynamic transfer probability grid is used to visualize changes in transfer probabilities during the scheduling decision-making process, helping to understand the dynamic scheduling process. Use the plot_dynamic_transfer_probability_grids function to draw a dynamic transfer probability grid. This function generates a dynamic transfer probability grid based on the transfer_prob_history transfer probability history, showing the changes in transfer probabilities for each task assignment. This function provides visibility into the scheduling decision-making process, making it more transparent and traceable. Figure 4 This is a schematic diagram of the visualization interface of the scheduling decision transfer probability cellular grid, showing the visualization interface of the scheduling decision dynamic transfer probability cellular grid.

[0077] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A transporter scheduling method based on a dynamic probability allocation mechanism, characterized in that: include: Extracting historical transportation data to obtain a transportation time matrix; the transportation time matrix includes a first transportation time from the nursing unit to the examination room, an examination time, and a second transportation time from the examination room to the nursing unit; Constructing a patient transportation matrix based on the future patient transportation task list; the patient transportation matrix includes a first transportation task from the nursing unit to the examination room, an examination task, and a second transportation task from the examination room to the nursing unit; Based on the scheduling tendency of patients and hospital transport managers, the transporter data is processed to obtain the initial transporter scheduling matrix, including: determining the dispatched transporter data based on the scheduling information; the dispatched transporter data includes the transporter's workload, task waiting time and current location; constructing a dynamic transfer probability model based on the dispatched transporter data to calculate the transport transfer probability of patients to transporters; sorting the transporters based on the transport transfer probability to obtain the transporter-to-patient transfer transfer probability matrix; based on the transfer probability matrix, assigning multiple patients to multiple transporters respectively to obtain the transport allocation strategy; based on the transport allocation strategy, updating the transporter scheduling matrix to obtain The final courier scheduling matrix includes: constructing a parent chromosome based on the updated courier scheduling matrix; generating multiple chromosomes based on the parent chromosome to form a population; each chromosome corresponds to a transportation allocation strategy; constructing a fitness function based on the courier's workload and the average waiting time of the patient, and calculating the fitness value of each chromosome; updating the parent chromosome based on the fitness value, and repeating the screening operation of the parent chromosome until the loop termination condition is met; the scheduling tendency is related to the courier selection for the first transportation task and the second transportation task; the courier data includes the courier's workload and current location; the fitness function is: ; ； ； in, represents the fitness function; It represents the working imbalance coefficient; represents the average waiting time of patients; i represents the transporter variable; N represents the total number of transporters; represents the current cumulative working time of transporter i; represents the average working hours of transporters; represents the total number of patients who need to wait; j represents the patient variable; represents the time point at which the jth patient is scheduled to undergo examination; represents the end time of transporting the jth patient; Calculate the dispatch information of the transporters in the initial transporter dispatch matrix, and dynamically update the transporter dispatch matrix based on the dispatch information through a dynamic transition probability model to obtain the final transporter dispatch matrix; the expression of the dynamic transition probability model is: ； in, represents the transport transfer probability; exp represents the exponential function; Indicates the total duration of the tasks assigned to the mth transporter during the current scheduling period; represents the task time interval of transporter m transporting patient i; represents the time required for transporter m to reach patient i from the current location; Indicates the total duration weight; represents the time interval weight; represents the distance weight; represents an empty cell set.

2. The method for dispatching transporters based on a dynamic probability allocation mechanism according to claim 1 is characterized in that: The transporter data is processed based on the patient's scheduling tendency to obtain an initial transporter scheduling matrix, including: Determine whether the patient selects the same transporter for the first transport task and the second transport task; If the same transporter is selected, the transporter for the first and second transport tasks of the patient will be bound; If different transporters are selected, the transporters for the first and second transport tasks will be assigned independently; The initial transporter scheduling matrix is ​​generated according to the transporter binding results of the first transport task and the second transport task.

3. The method for dispatching transporters based on a dynamic probability allocation mechanism according to claim 1 is characterized in that: The parent chromosome is constructed based on the updated transporter scheduling matrix, including: For patients who choose the same transporter for transportation, paired gene coding is used; the transporters of paired gene coding are kept consistent; Genes where the patient matched the transporter were coded as 1, and genes where the patient did not match the transporter were coded as 0.

4. The method for dispatching transporters based on a dynamic probability allocation mechanism according to claim 1 is characterized in that: The updating of the parent chromosome comprises: According to the dynamic transfer probability model, the transport transfer probability of the patient to the transporter in each transport allocation strategy in the current parent chromosome is calculated; Determining the mutation probability of each gene position in the chromosome based on the transport transfer probability; Based on the mutation probability, the parent chromosome is mutated and updated to obtain a new parent chromosome.

5. The method for dispatching transporters based on a dynamic probability allocation mechanism according to claim 1 is characterized in that: The dynamic transition probability model is related to the transporter's workload, task time interval and starting transport distance.

6. The method for dispatching transporters based on a dynamic probability allocation mechanism according to claim 1 is characterized in that: The task arrangement of patients and transporters is presented through a Gantt chart, and the scheduling decision-making process is presented through a dynamic transfer probability cellular grid diagram.

Images