Load balancing method and system for registration system in high-concurrency scene

Abstract

The invention provides a load balancing method and system for a registration system in a high-concurrency scene, and belongs to the technical field of load control, and the method specifically comprises the steps: carrying out the determination of a load balancing control target of the registration system in a load balancing control time period through employing an identification method, and carrying out the load balancing control of the registration system based on the load balancing control targets in different hospital departments, the method comprises the following steps: determining an optimized target hospital department in hospital departments according to a load balancing control time period of a load balancing control target in the optimized target department and a load balancing control target in the load balancing control time period according to the load balancing control time period of the load balancing control target in the optimized target department and plan registration data of the hospital departments; the load balancing control method for optimizing the target department is determined, so that the registration availability of different departments is ensured on the basis of reducing the operation load of the registration system.

Description

Load balancing methods and systems for registration systems in high-concurrency scenarios Technical Field

[0001] This invention belongs to the field of load control technology, and particularly relates to a load balancing method and system for a registration system under high concurrency scenarios. Background Technology

[0002] With the rapid development of internet healthcare, hospital registration systems are facing increasing pressure from high-concurrency access. Especially during peak appointment periods and when popular specialist appointments are released, the instantaneous concurrent access volume can reach hundreds of thousands or even millions. Traditional registration systems typically employ single load balancing strategies, such as round-robin or least connections, which are ill-suited to handle complex and ever-changing concurrency scenarios. This can easily lead to server response delays, system crashes, and other problems, severely impacting the patient experience and the hospital's service efficiency.

[0003] To achieve load balancing control, the invention CN202511226851.2, "A Method and System for Optimizing Load Balancing of Measurement Data Query Based on Data Heat Assessment," automatically adjusts the query path according to changes in query load. This improves query efficiency and reduces resource waste through intelligent resource scheduling strategies, making it suitable for high-concurrency, high-load query scenarios. However, the above technical solution has the following technical problems: In high-concurrency scenarios, multiple departments may experience excessive data access pressure from doctors, leading to a decrease in the availability of the entire registration system. Therefore, determining the doctors responsible for traffic control in different departments and establishing a load balancing control strategy for each department based on the impact of traffic control on the department's registration availability becomes a pressing technical problem. This aims to reduce load pressure while ensuring the availability of registration in different departments.

[0004] Therefore, there is an urgent need for a load balancing method and system for registration systems in high-concurrency scenarios. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a load balancing method for a registration system under high concurrency scenarios, specifically including: S1 determining the load balancing control period of the registration system based on the monitoring results of the traffic data of the registration system, and determining the identification method of the load balancing control target of the registration system based on the load balancing control period data and the registration data of the load balancing control period; S2 determining the load balancing control target of the registration system during the load balancing control period using the identification method, and determining the optimized target hospital department based on the load balancing control targets in different hospital departments and the planned registration data of the hospital departments; S3 determining the data of the optimized target hospital department, and determining the load balancing control method of the optimized target department by combining the load balancing control period in which the optimized target department has a load balancing control target and the load balancing control target in the load balancing control period.

[0006] The beneficial effects of this invention are as follows: This application utilizes a method for identifying load balancing control targets in a registration system. Instead of using fixed and unchanging standards to identify "doctors who need load control," it dynamically and intelligently switches between "strict identification" and "relaxed identification" strategies based on the overall system pressure pattern (centralized / dispersed, continuous / intermittent) reflected by the "control period." This reduces load pressure while ensuring system availability, prioritizing the most critical services when resources are extremely scarce, and proactively allocating resources to more potential pressure points when resources are relatively abundant, thereby optimizing overall resource utilization.

[0007] By identifying target hospital departments for optimization, and based on the previously identified "load balancing control targets" (i.e., high-load physicians) within each department, the resource strain and potential service risks are assessed from an overall departmental perspective. This allows for the intelligent determination of which departments need to be marked as "target hospital departments for optimization." Whether a department requires overall scheduling strategy optimization depends not only on the number of high-load physicians, but more importantly on the breadth, continuity, and concentration of the pressure caused by these physicians over the future. This method aims to identify departments that will face widespread, continuous, or concentrated high-load risks in the future, thereby enabling prioritized, system-level resource allocation and process optimization to prevent localized pressure from spreading into systemic service bottlenecks.

[0008] Furthermore, the traffic data is determined based on the monitoring data of the access traffic of the registration system.

[0009] Furthermore, the method for determining the load balancing control period of the registration system is as follows: based on the monitoring results of the traffic data of the registration system, the data traffic of the period at different times is determined; based on the data traffic, the dates of the period that belong to the traffic peak period are determined; based on the date data of the period that belong to the traffic peak period, it is determined whether the period belongs to the load balancing control period of the registration system.

[0010] It should be noted that the dates that fall within the peak traffic period are those where the proportion of monitoring times during which data traffic exceeds the rated traffic threshold does not meet the requirements, i.e., the period during which the proportion of monitoring times during which data traffic exceeds the rated traffic threshold is greater than the preset monitoring time proportion threshold.

[0011] Furthermore, the method for determining the load balancing control method for the optimized target department is as follows: the load balancing control period in which the optimized target department has a load balancing control target is taken as the target control period; using the load balancing control target data in the target control period, the load balancing control targets other than the optimized target department in the target control period are determined and taken as priority control targets; the load balancing control method for the optimized target department is determined using the optimized target hospital department data, the target control period of the optimized target department, and the priority control targets in different target control periods.

[0012] Secondly, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described load balancing method for a registration system under high concurrency scenarios when running the computer program.

[0013] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0014] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0015] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0016] Figure 1 is a flowchart of a load balancing method for a registration system under high concurrency scenarios; Figure 2 is a flowchart of a method for determining the load balancing control period of a registration system; Figure 3 is a flowchart of a method for determining the identification of the load balancing control target of a registration system; Figure 4 is a flowchart of a method for determining the load balancing control method for the optimized target department. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0018] Example 1, as shown in Figure 1, provides a load balancing method for a registration system under high concurrency scenarios. Specifically, it includes: S1 determining the load balancing control period of the registration system based on monitoring results of the system's traffic data; and determining an identification method for the load balancing control target of the registration system based on the load balancing control period data and the registration data within that period; S2 using the identification method to determine the load balancing control target of the registration system under high concurrency scenarios; and determining the optimized target hospital department based on the load balancing control targets in different hospital departments and the planned registration data of those departments; S3 acquiring the load balancing control target during the registration period in real time, and combining this with the load balancing control target data of the optimized target department to determine the load balancing control method for that period.

[0019] Furthermore, the traffic data is determined based on the monitoring data of the access traffic of the registration system.

[0020] Specifically, as shown in Figure 2, the method for determining the load balancing control period of the registration system is as follows: The core decision-making objective of this embodiment is to establish a data-driven intelligent decision-making system based on historical traffic patterns to automatically identify and determine which fixed time periods (e.g., "9:00-11:00 AM every day") in the registration system require the activation of load balancing control strategies. Its core logic lies in analyzing the historical performance of the system's data traffic within fixed time periods to determine whether high load pressure frequently and significantly occurs during those periods. This allows for a scientific decision on whether to deploy or routinely operate a load balancing mechanism to ensure system stability and response speed under high concurrency access, optimize resource utilization, and avoid system congestion or service degradation.

[0021] This method follows a progressive analysis process: "monitoring historical traffic → identifying busy dates → evaluating busy patterns → deciding whether load balancing is necessary." First, historical traffic data for the target time period is quantitatively monitored. Second, the "busy traffic dates" of that time period are defined and identified. Finally, through multi-dimensional analysis of the occurrence patterns and intensity of these busy dates (consistency, idle time, peak pressure), it is ultimately determined whether the time period should be designated as a "load balancing control period."

[0022] S11 uses the monitoring results of the registration system's traffic data to determine the data traffic at different times during the specified time period, and uses the data traffic as a basis to determine the dates during which the specified time period belongs to the peak traffic period; monitors data traffic and identifies peak traffic dates: "Data traffic": refers to the number of service requests or network data throughput received and processed by the registration system per unit time (e.g., per minute) within the target statistical period (e.g., "9:00-11:00").

[0023] "Rated traffic threshold": A preset system capacity warning line. When data traffic exceeds this threshold, it indicates that the system may begin to face performance pressure, and response time may increase.

[0024] “Monitoring Time”: Divide the target time period into multiple consecutive monitoring time points or time slices (e.g., one point per minute).

[0025] "Busy Traffic Period / Date": For a specific date, if the number of monitoring times during the target time period of that date where the data traffic exceeds the rated traffic threshold, and the proportion of such monitoring times to the total number of monitoring times during that time period exceeds the "preset monitoring time proportion threshold" (e.g., 50%), then that date is marked as a "Busy Traffic Date" for that time period.

[0026] This configuration transforms "busy" from a vague, qualitative concept into a quantifiable and comparable statistical metric. It considers not only whether traffic exceeds a threshold (intensity) but also the duration of that exceedance (density). Being under high load for more than half the time within a given period is clearly a better indicator of sustained system stress than just an occasional traffic spike. This avoids misjudging the entire period based on momentary peaks, ensuring the stability of the identification.

[0027] This step is fundamental to any pattern analysis. It processes the raw, continuous time-series traffic data into discrete sample points labeled "busy" or "not busy," based on date and time period. This allows for subsequent statistical analysis of the behavior patterns of that time period within historical cycles.

[0028] Specific example: We use "9:00 AM - 11:00 AM daily" as the target analysis period, dividing it into 120 monitoring times (one per minute). We set the "rated traffic threshold" to 1000 requests per second and the "preset monitoring time percentage threshold" to 60%. The monitoring data for the past 7 days is as follows: Monday: Out of the 120 times, traffic > 1000 requests per second in 80 times, percentage = 80 / 120 ≈ 66.7% > 60%. Monday is therefore determined to be a "busy traffic day".

[0029] Tuesday: Only 30 times exceeded the threshold, a rate of 25% < 60%. Tuesday is therefore classified as a "non-busy day".

[0030] Wednesday to Sunday: Similar calculations are used, assuming that the proportion of traffic on Wednesday, Thursday, and Friday exceeds 60%, while that on Saturday and Sunday is below 60%. Then, the "busy traffic days" in the past 7 days are {Monday, Wednesday, Thursday, Friday}.

[0031] S12 determines whether the time period belongs to the load balancing control period of the registration system based on the date data of the time period belonging to the busy traffic period.

[0032] It should be noted that the dates that fall within the peak traffic period are those where the proportion of monitoring times during which data traffic exceeds the rated traffic threshold does not meet the requirements, i.e., the period during which the proportion of monitoring times during which data traffic exceeds the rated traffic threshold is greater than the preset monitoring time proportion threshold.

[0033] Understandably, based on date data indicating that the time period falls within a peak traffic period, determining whether the time period belongs to the load balancing control period of the registration system involves the following: Case 1: If the time period falls within a peak traffic period on different dates, then the time period is determined to belong to the load balancing control period of the registration system; First-level judgment (extreme consistency screening): Is the time period busy every day? Check whether the time period is marked as a "peak traffic period" on every single day of all analyzed dates. If so, then directly determine that "the time period belongs to the load balancing control period of the registration system".

[0034] This is the clearest signal. If a period of time historically exhibits high load pressure every day, it indicates that its busyness is an absolute, unparalleled, and systematic pattern (e.g., corresponding to the morning rush hour). For such highly deterministic periods, no further analysis is needed; load balancing control must be enabled routinely, as this is a necessary condition for ensuring basic system availability.

[0035] A rapid decision-making channel has been established. For such obviously "critical" areas, the system can make decisions immediately, saving computing resources and ensuring uninterrupted service during peak periods.

[0036] Specific example (continued from S11): The analysis period "9:00-11:00" was not busy every day in the past 7 days (there were Tuesdays, Saturdays, and Sundays when it was not busy). Therefore, condition 1 is not met, and we proceed to condition 2.

[0037] Scenario 2: If the time period is not uniformly among busy traffic periods on different dates, determine the dates during which no monitoring time exceeds the rated traffic threshold and designate them as idle dates. Determine if the proportion of idle dates exceeds a preset idle date proportion threshold. If so, determine that the time period does not belong to the load balancing control period of the registration system; otherwise, proceed to the next step. Is the time period mostly idle? "Idle Dates": If Scenario 1 is not satisfied, identify the dates during which no monitoring time exceeds the rated traffic threshold. This represents another extreme of busyness.

[0038] "Percentage of free days": Number of free days / Total number of analysis days.

[0039] Calculate the "percentage of idle days". If this percentage is greater than the "preset idle day percentage threshold" (e.g., 70%), then it is determined that "this period does not belong to the load balancing control period".

[0040] This is another clear no-go signal. If a time period shows absolutely no signs of high load for most of the days (below the threshold at all times), then it is an overall "idle period." Regularly enabling load balancing for such periods would result in wasted resources and policy redundancy. Simply classifying it as "no control" is the most cost-effective choice.

[0041] Specific example (continued): In the past 7 days, identify the "free days": Tuesday (30 times exceeding the threshold, not completely free), Saturday, and Sunday (assuming no times exceeding the threshold). Therefore, the free days are {Saturday and Sunday}, with a proportion of 2 / 7 ≈ 28.6%. Assuming the "preset free day proportion threshold" is 70%, 28.6% is not greater than 70%, so the condition is not met, and proceed to the next step in case 2.

[0042] The maximum percentage of monitoring times during which data traffic exceeds the rated traffic threshold on different dates is used to determine whether the time period belongs to the load balancing control period of the registration system.

[0043] It is understood that if the maximum value of the proportion of monitoring times on different dates during the specified time period when the data flow is greater than the rated flow threshold is greater than the preset threshold, then the specified time period is determined to belong to the load balancing control period of the registration system.

[0044] Is the pressure during the busiest period sufficient? Logic and Judgment (second half of Case 2): When busy periods don't occur daily and idle periods are not the norm, more granular metrics are needed. Calculate the maximum value of the metric "proportion of monitoring times where data flow exceeds the rated flow threshold" (i.e., the historical highest busy density) across all analyzed dates for this period. Compare this maximum value with the "preset threshold" (which may be equal to or higher than the "preset monitoring time proportion threshold" in S11).

[0045] If the maximum value is greater than the preset threshold, it is determined that "this period belongs to the load balancing control period".

[0046] This assessment focuses on the upper limit of potential pressure during that period. Even if a period isn't busy every day (scenario 1 hasn't passed) and has many idle days (scenario 2 hasn't passed), as long as it has historically reached extremely high busy density, it indicates that the system has the capacity and scenarios to generate significant pressure during that period (e.g., corresponding to Mondays or specific appointment release days). To cope with this potentially periodic peak pressure, this period must be included in the load balancing control scope to prepare for possible extreme situations and ensure system resilience.

[0047] It provides a basis for risk-based decision-making. It ensures that the system does not ignore occasional but highly destructive peak pressure points due to low average load, and sets the line of defense to ensure system stability above the historical highest water level.

[0048] Specific example (final judgment): Calculate the busy density percentage of the "9:00-11:00" time period over the past 7 days: Monday (66.7%), Tuesday (25%), Wednesday (70%), Thursday (65%), Friday (75%), Saturday (0%), Sunday (0%). The highest value is 75% for Friday.

[0049] Let's assume the "preset threshold" here is 70%. Since the maximum value 75% > 70%, this period (9:00-11:00 daily) is determined to be within the load balancing control period of the registration system.

[0050] Specifically, as shown in Figure 3, the method for determining the load balancing control target of the registration system is as follows: The core decision objective of this embodiment is to intelligently select the most suitable load balancing control target identification method based on the characteristics of the busy patterns of multiple identified "load balancing control periods." The core logic is that the root cause of system pressure may be highly concentrated on specific doctors (popular specialists) or relatively dispersed. By analyzing the number of control periods and the overlapping patterns of their busy dates, the system can determine whether the current load pressure pattern is "concentrated" or "dispersed," thereby dynamically adjusting the standard threshold for identifying "high-load doctors" and achieving precise targeting and differentiated control of the "key contradiction" in server resource contention.

[0051] This method follows a progressive analysis process: "statistical control of time periods → analysis of overlapping busy dates → hierarchical decision target identification strategy." First, all time periods requiring load balancing control are summarized. Second, the busy patterns of these time periods across different dates are analyzed. Finally, through a multi-level decision tree (S231-S234), based on dimensions such as "number of time periods," "busy consistency," "busy ratio," and "average pressure on dates," a choice is made between a strict "high threshold" (date quantity threshold) or a more lenient "low threshold" (second date quantity threshold) to filter load balancing control targets (i.e., doctors requiring flow control or resource guarantees).

[0052] S21 uses the load balancing control period data to determine the number of load balancing control periods; determines the total number of load balancing control periods: "Load balancing control period" refers to the set of fixed time windows identified by the aforementioned methods (S11-S12) that require normalized or policy-based activation of the load balancing mechanism, such as {"9:00-11:00", "14:00-16:00", "19:00-20:00"}. "Number of load balancing control periods" is the total number of periods in this set.

[0053] This is a preliminary measure of the breadth of system load distribution over time. A large number of control periods means that the system faces a wide distribution of high load risk throughout the day, which may indicate high overall access pressure, dispersed load sources (popular doctors), or multiple business peaks. This is the primary macro-indicator for determining which refined management strategy to adopt subsequently.

[0054] This step establishes an initial watershed for decision-making. It preliminarily divides the scenario into two categories: "high pressure in multiple periods" and "high pressure in fewer periods," laying the foundation for subsequent in-depth analysis of target recognition strategies under different pressure modes.

[0055] Specific example: Suppose that the system identifies three load balancing control periods based on historical data: T1 (9:00-11:00), T2 (14:00-16:00), and T3 (19:00-20:00). Then the number of load balancing control periods is 3.

[0056] S22 determines the dates that belong to the busy traffic period during the load balancing control period based on the registration data, and designates these dates as busy dates; it then determines the busy date set for each control period: for each load balancing control period, based on its historical data (or real-time monitoring data), it determines which dates belong to the "busy dates" of that period. The definition of a busy date is the same as before (i.e., within that period, the proportion of monitoring times where data traffic exceeds the rated traffic threshold exceeds a preset threshold). Each period will have a corresponding busy date set.

[0057] This step aims to "timestamp" each stress period, revealing the specific date patterns of its occurrence. This forms the basis for analyzing whether busy patterns are synchronized across periods and whether the stress is cyclical. For example, T1 might only be busy on weekdays, T2 might be busy on weekends, and T3 might be busy every day.

[0058] This step links the abstract "time period" with specific "dates," constructing a two-dimensional data matrix to analyze the distribution of stress across both time and date dimensions. This is a prerequisite for subsequent similarity analysis, calculation of busy ratios, and average stress levels across dates.

[0059] Specific example (continued from S21): Analyze the data of the past 30 days to obtain the set of busy dates for each time period (example): T1 (9:00-11:00) busy dates: {all Monday to Friday, a total of 22 days}; T2 (14:00-16:00) busy dates: {all Saturdays and Sundays, a total of 8 days}; T3 (19:00-20:00) busy dates: {all dates, a total of 30 days}; S23 Based on the number of load balancing control time periods and the similarity data of busy dates in different load balancing control time periods, determine the identification method of the load balancing control target of the registration system.

[0060] Furthermore, if the number of load balancing control periods is greater than the preset control period number threshold, then the method for identifying the load balancing control target of the registration system is to use doctors whose access traffic in the most recent preset time period does not meet the requirements and whose number of dates is greater than the date number threshold as load balancing control targets.

[0061] Level 1 Judgment (Pressure Breadth Screening): Is the number of control periods extremely large? If the "number of load balancing control periods" is greater than the "preset control period number threshold" (e.g., 5), it indicates that the system pressure coverage time window is very wide. At this time, the system's default load pressure sources may be relatively dispersed, but the overall environment is highly competitive. Therefore, a strict standard is adopted: the identification method is to "use doctors whose access traffic in the most recent preset time period does not meet the requirements and whose number of dates is greater than the 'date number threshold' as load balancing control targets."

[0062] "Access traffic does not meet requirements": This usually means that the number of access requests (or concurrency) for the doctor in the corresponding time period exceeds the threshold of their individual service capacity.

[0063] "Date Quantity Threshold": A higher standard, such as requiring doctors to have more than 5 days "not meeting the requirements" in the last 7 days.

[0064] When load balancing is required at multiple times throughout the day, it means that resource shortages are a common phenomenon. In this case, the criteria for identifying control targets (high-load doctors) should be more stringent, and only those "absolute hotspots" of doctors who consistently and frequently generate pressure should be listed as control targets. This would avoid unnecessary intervention for a large number of doctors who are only occasionally overloaded, and prevent the strategy from becoming overused.

[0065] Specific example (continued): The number of control periods is 3. Assume the "preset control period number threshold" is 5, and 3 is not greater than 5. Therefore, proceed to the next level of judgment (S231).

[0066] Furthermore, if the number of load balancing control periods is not greater than a preset control period number threshold, the following steps are also included: S231 Determine whether there are load balancing control periods that all belong to busy dates. If so, determine that the identification method for the load balancing control target of the registration system is to use doctors whose access traffic in the most recent preset time period does not meet the requirements and whose number of dates exceeds the date number threshold as load balancing control targets. If not, proceed to step S232; Second-level judgment (pressure consistency screening): Whether there are load balancing control periods where all dates belong to busy dates: Check whether there are dates where all dates in the load balancing control period are simultaneously in a "busy" state. If so (i.e., there are "load balancing control periods that all belong to busy dates"), it indicates that some load balancing control periods are relatively busy on all dates.

[0067] During complex, balanced control periods that all fall within busy days, the data processing pressure is significant. Under this extreme pressure mode, the most stringent criteria must be used to identify and prioritize the sources of pressure for control in order to prevent the system from crashing.

[0068] Specific example (continued): Check if there are dates that make T1, T2, and T3 all busy. According to the example data, T3 is busy every day. Therefore, there exists such a date T3. Thus, the method for identifying the load balancing control target of the registration system is to use doctors whose access traffic within the most recent preset time period does not meet the requirements and whose number of dates exceeds a certain threshold as the load balancing control target.

[0069] S232 determines the busy ratio of the load balancing control period based on the proportion of busy dates in the load balancing control period, and determines whether there is a load balancing control period with a busy ratio greater than a preset ratio threshold. If yes, proceed to step S233; otherwise, determine that the identification method for the load balancing control target of the registration system is to take doctors whose access traffic in the most recent preset time period does not meet the requirements and whose number of dates is greater than a second date number threshold as load balancing control targets. It should be noted that the number of dates is less than the second date number threshold.

[0070] Does it have periods that are almost always busy? "Busy Ratio" = Number of busy days in a given load balancer control period / Total number of analyzed days. It measures the frequency of busy periods within that timeframe.

[0071] Logic and Judgment (first half of S232): Calculate the "busy ratio" for each control period. If the busy ratio for a certain period is greater than the "preset ratio threshold" (e.g., 80%), it indicates that this period is a highly stable, almost daily stress point. Further analysis is needed (proceed to S233).

[0072] If a period of time is busy almost every day, then it is itself a source of "certain" stress. For this kind of "time-specific" certainty stress, it's necessary to determine whether it's an isolated or widespread phenomenon to decide whether to raise or lower the threshold for identifying doctors.

[0073] Specific example (continued): Assuming T3 has only 28 days, calculate the busy percentage for each time period (based on 30 days): T1: 22 / 30 ≈ 73.3%, T2: 8 / 30 ≈ 26.7%, T3: 28 / 30 = 93.3%. Assuming the "preset percentage threshold" is 80%, the busy percentage for T3 is 93.3% > 80%, therefore the condition is met. Mark T3 as the "filter control period" and proceed to S233.

[0074] S233 identifies load balancing control periods with a busy ratio greater than a preset threshold as filter control periods. It then determines whether the number of filter control periods exceeds a preset filter period number threshold. If so, the method for identifying the load balancing control target of the registration system is to identify doctors whose access traffic within the most recent preset time period does not meet the requirements and whose number of dates exceeds a date number threshold, as load balancing control targets. If not, proceed to step S234. The number of consistently busy periods is counted: the number of "filter control periods" (i.e., periods with a busy ratio > 80%) is counted. If this number exceeds the "preset filter period number threshold" (e.g., 2), it indicates that the system has multiple periods of high pressure almost daily, and the pressure pattern is still biased towards "concentrated and continuous." Therefore, a strict standard (date number threshold) is still needed to identify the control target.

[0075] If multiple time periods exhibit characteristics of being busy almost every day, then the system faces a continuous high-pressure environment consisting of multiple "daily hot periods," where competition remains fierce and targets need to be strictly screened.

[0076] Specific example (continued): There is only one "filter control period" T3, with a quantity of 1. Assume the "preset filter period quantity threshold" is 2, and 1 is not greater than 2. Therefore, proceed to the final judgment S234.

[0077] S234 uses similar data of busy days in different load balancing control periods to determine the average number of load balancing control periods belonging to busy days in different dates, and uses the average number of load balancing control periods belonging to busy days in different dates to determine the identification method of the load balancing control target of the registration system.

[0078] Furthermore, if the average number of load balancing control periods belonging to busy days in different dates is greater than the preset control period number threshold, then the method for identifying the load balancing control target of the registration system is to use doctors whose access traffic does not meet the requirements in the most recent preset time period is greater than a date number threshold as load balancing control targets. If the average number of load balancing control periods belonging to busy days in different dates is not greater than the preset control period number threshold, then the method for identifying the load balancing control target of the registration system is to use doctors whose access traffic does not meet the requirements in the most recent preset time period is greater than a second date number threshold as load balancing control targets.

[0079] Level 5 Judgment (Date-Dimensional Average Pressure Screening): On average, how many periods are busy each day? "The average number of load balancing control periods that are busy on different dates" is a key integrated indicator. It calculates, on average, how many control periods are busy each day over the entire analysis period. For example, if the total number of busy periods over the past 30 days is (T1:22 + T2:8 + T3:28) = 58, then the average number of busy periods per day = 58 / 30 = 1.93.

[0080] Logic and Judgment (S234): If the average value is greater than the "preset control period number threshold" (this threshold may be the same as the first level's "preset control period number threshold", for example, 2.5), it indicates that from the date dimension, there are more stressful periods on average every day, and the stressful environment is still "intense", so a strict standard (date number threshold) is adopted.

[0081] If the average value is not greater than the threshold, it indicates that there are fewer busy periods on average each day, and the pressure is relatively "sparse" or concentrated in a few periods. In this case, a more lenient standard (the second date quantity threshold) can be used to identify the control target. The second date quantity threshold is less than the date quantity threshold.

[0082] This is the most comprehensive and balanced consideration. It no longer looks at the attributes of the time period itself, but rather at the "density" of stress periods from a "daily" perspective. More average busy periods per day mean that doctors face competition in more time windows, making them more likely to become high-workload doctors, and thus the selection process should be more stringent. Conversely, fewer average busy periods per day mean a limited window of competition, allowing for more relaxed selection criteria to include more doctors who may cause localized stress within the scope of regulation, achieving more refined preventative management.

[0083] Specific example (final judgment): Calculate the average: (22 + 8 + 28) / 30 = 58 / 30 = 1.93 (doctors / day). Assume the "preset control period quantity threshold" here is 2.5. Since 1.93 is not greater than 2.5, the method for identifying the load balancing control target of the registration system is: doctors whose access traffic in the most recent preset time period does not meet the requirements on a number greater than the 'second date quantity threshold' are used as load balancing control targets.

[0084] In a possible specific embodiment, "recent preset time" is set to the most recent 7 days (one calendar week). "Access traffic not meeting requirements" is defined as the doctor's average number of access requests (concurrent users) per minute exceeding a certain threshold during their consultation hours. In this example, this threshold is set to 50 requests per second. That is, if a doctor's average number of requests per minute is >50 during a particular day's consultation hours, that day is recorded as a "not meeting requirements" instance.

[0085] "Second Date Quantity Threshold": Based on the decision result of step S234, it is set to 3 days. This is a relatively lenient standard, meaning that if a doctor has a high workload for more than 3 days in the past week, they are marked as a control target, and the date quantity threshold is 2 days.

[0086] Objective: Over the past 7 days (Monday to Sunday), identify which doctors should be listed as "load balancing control targets" so that the system can handle their registration requests specially (such as queuing, rate limiting, or allocation to a dedicated server cluster).

[0087] This embodiment constructs an adaptive load balancing control target identification strategy selection system based on stress pattern analysis. Its core innovation lies in not using fixed, unchanging standards to identify "high-load doctors," but rather dynamically and intelligently switching between "strict identification" (high threshold) and "lenient identification" (low threshold) strategies based on the overall system stress pattern (concentrated / dispersed, continuous / intermittent) reflected by the "control period." Through multi-level funnel-shaped analysis of "quantity → consistency → proportion → mean," the system can accurately perceive the "concentration" of the current load environment and select a matching screening granularity.

[0088] Achieve refined management of load regulation: focus on core issues when pressure is high (strict screening), and expand the scope of prevention when pressure is low (relaxed screening), making the intervention of load balancing control strategies more precise and effective, and avoiding a "one-size-fits-all" approach.

[0089] Improve the fairness and efficiency of system resource allocation: By dynamically adjusting the identification threshold, ensure that the most critical services (strictly selected doctors) are prioritized when resources are extremely scarce, while when resources are relatively abundant, preventative resource allocation is carried out in advance for more potential pressure points (relatively selected doctors), thereby optimizing the overall resource utilization rate.

[0090] Enhance the system's adaptability to different pressure scenarios: The system can automatically distinguish between different scenarios such as "daily peak", "intermittent peak" and "all-weather high pressure", and call up the corresponding target identification and handling plans, thereby enhancing the automation level of operation and maintenance and the resilience of the system.

[0091] Provides in-depth insights for capacity planning and performance optimization: The identified different stress patterns and their corresponding target identification strategies are valuable inputs for analyzing system bottlenecks, expanding server capacity, and optimizing architecture, making technical decisions more aligned with actual business load characteristics.

[0092] This greatly simplifies the configuration and management of load balancing strategies. Operations personnel no longer need to manually set complex rules for different dates and time periods; the system can automatically derive the best target identification method currently applicable based on historical data, reducing operational complexity.

[0093] Specifically, the method for determining the target hospital departments for optimization is as follows: The core decision-making objective of this embodiment is to assess the resource strain and potential service risks of each department from an overall departmental perspective, based on the identification of the "load balancing control targets" (i.e., high-load doctors), and then intelligently determine which departments need to be marked as "target hospital departments for optimization." The core logic is that whether a department needs overall scheduling strategy optimization depends not only on the number of high-load doctors, but more importantly on the breadth, continuity, and concentration of the pressure caused by these doctors over the future. This method aims to screen out departments that will face widespread, continuous, or concentrated high-load risks in the future, thereby prioritizing and optimizing their resources and processes at the systemic level to prevent localized pressure from spreading into systemic service bottlenecks.

[0094] This method follows a progressive decision-making process: "quantifying the number of high-risk doctors in a department → predicting future stress dates → assessing risk patterns → determining which department to optimize." First, the number of doctors within the department marked as load balancing control targets is counted. Second, based on the scheduling plan, the dates on which these high-load doctors will appear are predicted. Finally, through a decision tree (S332-S333) containing multi-level thresholds, the "number scale," "future impact frequency," "continuity," and "overall impact intensity" of high-risk doctors are comprehensively analyzed to ultimately determine whether the department needs to be upgraded to an "optimized target hospital department."

[0095] S31 determines the number of load balancing control targets in the hospital departments based on the load balancing control targets in the hospital departments; and determines the number of high-load doctors within the departments: "Load balancing control targets" refers to individual doctors identified through the aforementioned methods who require focused traffic control. "The number of load balancing control targets in the hospital departments" refers to the total number of doctors marked as such targets within a specific department (such as "cardiovascular medicine").

[0096] This is the most direct indicator for assessing the scale of internal stressors within a department. If multiple doctors in a department are under high workload, it indicates that there is widespread, multi-point concurrent resource competition pressure within the department, rather than just a problem with a few star doctors.

[0097] This step provides an initial quantification of the department's resource strain. It forms the basis for all subsequent analyses and is one of the strongest risk signals. A high number of control targets directly implies the necessity of departmental intervention.

[0098] Specific example: Taking the "Cardiovascular Medicine" department as an example, using the load balancing target identification method, it was found that there are 15 doctors in this department, of which 5 doctors (such as D1, D3, D7, D9, and D12) are marked as "load balancing control targets". Therefore, the "number of load balancing control targets" for this department is 5.

[0099] S32 determines, based on the planned registration data of the hospital department, the dates on which the hospital department will have a load control target within a preset time period in the future, and uses these dates as the affected dates; predicts the dates on which there will be high-load doctors in the future: "Planned registration data": refers to the doctor's schedule, available appointment slots, etc. in the hospital information system for a future period of time.

[0100] "Predicted duration of the future": The time window for predictive analysis, such as "the next 14 days".

[0101] "Impact Dates": Dates on which at least one physician from the department's "load balancing control target" is scheduled to see patients within a predetermined timeframe. A date is considered to have a potential department-level high-load risk if any high-load physician is scheduled to see patients on that day.

[0102] This step combines a static list of doctors with a dynamic schedule, enabling spatiotemporal prediction of risk. It answers a key question: "On which specific dates in the coming weeks will this department experience 'high-pressure' conditions?" This reflects the actual temporal distribution of service pressure better than simply counting the number of doctors, and forms the basis for accurate resource prediction and intervention scheduling.

[0103] This step shifts the perspective from "who has a problem" to "when will a problem occur." It transforms departmental risk from attribute labels into a clear future risk calendar, providing core input for subsequent assessments of risk frequency, continuity, and other time patterns.

[0104] Specific example (continued from S31): Analyze the schedule of "Cardiovascular Medicine" for the next 14 days (days 1 to 14): Days 1, 3, 5, 7, 9, 11, and 13: There is at least one overloaded doctor (some of the 5) on duty.

[0105] Days 2, 4, 6, 8, 10, 12, and 14: No high-load doctors are on duty (or they are resting).

[0106] Therefore, the "impact dates" for this department are {days 1, 3, 5, 7, 9, 11, 13}, a total of 7 days.

[0107] S33 determines whether a hospital department is an optimization target hospital department based on the number of load balancing control targets in the hospital department and the data on the date of impact.

[0108] Specifically, if the number of load balancing control targets in the hospital department exceeds the preset control target number threshold, port restrictions may cause some users to have difficulty registering in a single department. Therefore, the hospital department is identified as the target hospital department for optimization.

[0109] Are there too many doctors with high workload? Compare the "number of load balancing control targets" with the "preset control target number threshold". If it is greater, then directly determine "this department is an optimization target hospital department".

[0110] This is a circuit breaker mechanism. If a department has more than a threshold number of doctors requiring individual load control, it indicates that the supply-demand imbalance in that department is not an isolated phenomenon, but a structural and widespread problem. Controlling individual doctors solely through port rate limiting would lead to widespread difficulties for users registering with all doctors in that department, resulting in a terrible user experience. Therefore, it is directly marked as an optimization target.

[0111] Specific example (continued): Let the "preset control target quantity threshold" be 4. The cardiology department has 5 control targets, 5 > 4. Therefore, the cardiology department is directly determined as the target hospital department for optimization. The subsequent process ends.

[0112] (To demonstrate the complete logic, assume the threshold is 6; if 5 is not greater than 6, proceed to the next level of judgment.) It should also be noted that if the number of load balancing control targets in the hospital department is not greater than the preset control target number threshold, the following steps are also included: S331 Obtain the impact date data of the hospital department, and determine whether the proportion of the impact dates of the hospital department within the preset future duration is greater than the preset date proportion threshold. If yes, then determine that the hospital department is an optimization target hospital department; otherwise, proceed to step S332; Whether most future dates are at risk: Calculate the proportion of the number of "impact dates" to the total number of days in the "preset future duration". If this proportion is greater than the "preset date proportion threshold" (e.g., 60%), then determine that "this department is an optimization target hospital department".

[0113] Even if the number of doctors with high workloads is not large, if their scheduling becomes very intensive in the future, causing the department to be under high pressure and risk most days, then the stability of the department's services will face long-term challenges. Users will face the pressure of vying for appointments almost every day. In order to ensure the accessibility and fairness of the department's services, it is necessary to optimize it as a whole.

[0114] The affected date ratio = 7 days / 14 days = 50%. Set the "preset date ratio threshold" to 60%, where 50% is no greater than 60%. Proceed to S332.

[0115] S332 uses the distribution data of affected dates to group all periods within a preset future timeframe that are affected dates into the same group. It then determines whether the maximum number of dates in the group exceeds a preset number. If so, the hospital department is identified as the target hospital department for optimization; otherwise, proceed to step S333. The question then examines whether there exists a long, continuous period of risk: "Group": On the future timeline, all consecutive affected dates are grouped into one group. For example, if the affected dates are {1,2,3,7,8,10}, then the groups are: G1{1,2,3}, G2{7,8}, G3{10}.

[0116] "Maximum number of dates": refers to the number of dates in the group that contains the most consecutively affected dates among all groups.

[0117] Logic and Judgment (S332): Based on the distribution of the affected dates, divide the consecutive date groups and find the number of dates in the largest group.

[0118] If this maximum value is greater than the "preset value for the number of days" (e.g., 3 days), then it is determined that "this department is a department of the target hospital for optimization".

[0119] This assessment focuses on a concentrated outbreak pattern of risk. Multiple consecutive days of high physician workload indicate that the department will face a sustained, uninterrupted period of pressure (e.g., a week-long specialist outpatient week). This pattern easily leads to accumulated patient anxiety, daily increases in systemic pressure, and can trigger a "snowball effect" (if you can't get an appointment today, even more people will be scrambling for it tomorrow). Therefore, even if the overall proportion is not high, a prolonged period of continuous risk necessitates departmental optimization to smoothly navigate this period.

[0120] Specific example (continued): The dates affected by cardiovascular medicine are {1, 3, 5, 7, 9, 11, 13}. They are not consecutive (1 day apart). Therefore, they can be divided into 7 independent groups, each with only 1 day. The largest group has 1 day.

[0121] Set the "Preset value for the number of dates" to 3. 1 If it is not greater than 3, proceed to S333.

[0122] S333 determines the influence coefficient of the hospital department based on the proportion of the influence dates of the hospital department within a preset future time period and the ratio of the number of influence dates to the number of groups, and determines whether the hospital department is an optimization target hospital department based on the influence coefficient.

[0123] It is understood that if the influence coefficient is greater than the preset influence coefficient threshold, then the hospital department is determined to be the target hospital department for optimization.

[0124] How are the intensity and concentration of the risk? The "impact coefficient" is a comprehensive indicator designed to simultaneously measure the frequency and concentration of risk. Its calculation formula is: Impact Coefficient = α × (Proportion of Influenced Dates) + β × (Number of Influenced Dates / Number of Groups); where α and β are weighting coefficients (e.g., α = 0.6, β = 0.4). The ratio of "Number of Influenced Dates / Number of Groups" can be understood as the average number of days contained in each consecutive risk period. The larger the ratio, the more concentrated the risk dates tend to be compared to their dispersed nature.

[0125] Calculate the "influence coefficient" of the department. If this coefficient is greater than the "preset influence coefficient threshold", then the department is determined to be "the department of the target hospital for optimization".

[0126] This is the most refined assessment, used to capture moderate risk patterns that were not covered by the previous layers of judgment. For example, a department may have a moderate proportion of future risk days (40%), without exceptionally long consecutive periods (maximum 2 days), but these risk days occur in a "short, frequent" pattern (e.g., every other day). This pattern can still significantly disrupt the department's daily operations. The "impact coefficient," by combining proportion and concentration, can quantify the overall risk intensity of such "frequent but short" or "dispersed but substantial" risks.

[0127] Specific example (final judgment): For the cardiology department: the proportion of affected dates = 7 / 14 = 0.5, the number of affected dates = 7, the number of groups = 7 (each affected date is in an independent group), and the average concentration = 7 / 7 = 1.0; assuming α = 0.6 and β = 0.4, then: the influence coefficient = 0.6 × 0.5 + 0.4 × 1.0 = 0.3 + 0.4 = 0.7, and assuming the "preset influence coefficient threshold" is 0.65. Because 0.7 > 0.65, the cardiology department is determined to be the target hospital department for optimization.

[0128] This embodiment constructs a multi-level departmental risk perception and optimization decision-making system, encompassing micro-level (doctors) to meso-level (departments), current to future, and quantity to pattern. Its innovation lies in its four-layer progressive analysis—"quantity threshold → frequency threshold → continuity threshold → comprehensive coefficient"—which accurately distinguishes different types such as "structurally high-pressure departments," "long-term risk departments," "short-term outbreak departments," and "intermittently disruptive departments," and initiates departmental-level optimization strategies for the first three. This ensures that resource optimization is no longer a "one-size-fits-all" approach, but rather matches the actual risk patterns faced by the department.

[0129] To achieve precise and hierarchical allocation of medical resources: optimize resources (such as server computing power, network bandwidth, and management priority) and focus them on the departments that need them most and have the highest risk, thereby maximizing the efficiency of resource utilization.

[0130] Improving patient experience and fairness: By identifying and optimizing "target departments", the difficulty of registering for appointments in these departments can be systematically alleviated, avoiding negative experiences caused by repeated frustrations in a single department and promoting fairness in the allocation of appointment slots.

[0131] Enhance the overall stability of the registration system: By optimizing high-risk departments in a proactive and holistic manner, we can effectively prevent system-wide avalanche effects caused by congestion in individual departments, and ensure the stable operation of the registration service throughout the hospital.

[0132] Specifically, as shown in Figure 4, the method for determining the load balancing control of the optimized target department is as follows: The core decision-making objective of this embodiment is to further formulate refined and differentiated load balancing control execution strategies for departments identified as "optimized target departments". The core logic is that not all "optimized target departments" require immediate proactive port traffic control during all "target control periods". This method analyzes the overall competitive landscape across departments within the hospital, particularly the resource competition between the "optimized target department" and other departments (excluding the optimized target department) during the same high-pressure period, to determine whether to adopt a strategy of "immediate control", "delayed control (triggered after observation)", or "no control" for the "optimized target department". The core idea is that when hospital-wide resource competition is extremely fierce, the availability of the "optimized target department" should be prioritized; conversely, stricter control can be applied.

[0133] This method follows a progressive analysis process: "identifying target time periods → identifying hospital-level competitors → assessing competition intensity → deciding on control strategies." First, it identifies high-risk time periods for the target department itself. Second, within these time periods, it identifies doctors in other departments that also have high workload risk but are not part of the target department (priority control targets). Proactive workload control for these doctors can reduce workload pressure in some high-risk time periods. Finally, by analyzing the number and distribution of these "priority control targets," it assesses the degree of relief from current workload pressure and intelligently selects the most appropriate control method for the target department: immediate flow restriction, delayed triggering of flow restriction, or remaining open.

[0134] S41 defines the load balancing control period during which the optimized target department has a load balancing control target as the target control period; and identifies the high-risk period for the optimized target department: the "target control period" refers to the time window during which, for a certain "optimized target hospital department," there is both a hospital-wide "load balancing control period" and a doctor in that department with a "load balancing control target" within that period. That is, this is a "double high-pressure" period where the department faces both internal and external high pressure during the hospital-wide high-pressure period.

[0135] This setup is designed to focus on the points of convergence where conflicts are most prominent and decisions are most needed. Complex decisions are not required at all times; only when the entire hospital is at its peak (during load balancing control periods), and the target department itself is also at its peak (with target doctors seeing patients), should a careful consideration of control strategies be undertaken. This avoids excessive analysis during unnecessary periods.

[0136] This step completes the construction of the key risk matrix from "department optimization goals" to "department-time period". All subsequent decisions will be based on this table.

[0137] Assume that the Department of Cardiology has been identified as the target department for optimization. The hospital-wide load balancing control periods are: T1 (9:00-11:00) and T2 (14:00-16:00). The target physicians for load balancing control in the Department of Cardiology are {D1, D2, D3}. Specifically, D1 and D2 see patients during the T1 period, and D3 see patients during the T2 period. Therefore, for the Department of Cardiology, its "target control periods" are {T1, T2}.

[0138] S42 uses the load balancing control target data within the target control period to determine the load balancing control targets other than the optimized target department within the target control period, and designates them as priority control targets. A "priority control target" refers to a doctor belonging to another department (not the current optimized target department) within a given "target control period." For example, in time period T1, besides doctors D1 and D2 in cardiology, doctors E1 in neurology and F1 in orthopedics may also be marked as control targets. Therefore, in time period T1, for cardiology, the "priority control targets" are {E1, F1}.

[0139] This is the key to achieving global coordination in this method. By identifying "priority control targets," the system can perceive how many other popular departments and doctors at the hospital level are also competing for system resources during the same time period, while the target department is facing internal pressure. These "priority control targets" represent the main pressure sources from other departments that form cross-departmental competition with the target department. Since load balancing control targets require port traffic control during the balancing control period, such as limiting access to around 10 people per second, the traffic of the registration system is effectively controlled. If there are many priority control targets, the load pressure in the entire period is relatively low. Therefore, even if load balancing control is not performed on the target department, the reliability of registration for all departments in the entire period can be guaranteed.

[0140] Specific example (continued from S41): During the T1 period (9:00-11:00), the following doctors in other departments that are not part of the optimization target department are marked as control targets: Neurology E1, Orthopedics F1, and Pediatrics G1. Therefore, for Cardiology, the "priority control targets" during the T1 period are {E1, F1, G1}, with a quantity of 3.

[0141] During the T2 period (14:00-16:00), the control target physicians in other departments throughout the hospital that are not part of the optimization target department are: Gastroenterology H1. Therefore, for Cardiology, the "priority control target" during the T2 period is {H1}, with a quantity of 1.

[0142] S43 determines the load balancing control method for the optimized target department based on the optimized target hospital department data, the target control period of the optimized target department, and the priority control targets in different target control periods.

[0143] Furthermore, based on the optimized target hospital department data, the number of optimized target hospital departments is determined. If the number of optimized target departments is less than a preset optimized department number threshold, then the load balancing control method for the optimized target departments is determined to be no load control.

[0144] A tiered control strategy is formulated based on the overall competitive landscape: Decision entry point (global optimization demand screening): Is there a small number of departments requiring optimization? The total number of "target departments for optimization" across the entire hospital is counted. If this number is less than the "preset threshold for the number of departments to be optimized" (e.g., 3), it indicates that only a very small number of departments in the entire hospital are identified as high-risk, and the overall resource competition environment may not be tense. In this case, proactive load control is not implemented on these few departments ("no load control"), allowing them to freely compete for resources. The impact on the load pressure of the entire registration system is not very high.

[0145] If only one or two departments in the entire hospital are high-risk "frontrunners," then the overall system resources may be relatively abundant. In this case, no flow control should be imposed on them.

[0146] Suppose there are two target departments for optimization across the hospital: Cardiology and Orthopedics. The preset threshold is 3.2 < 3; therefore, the system directly determines the load balancing control method for Cardiology (and Orthopedics) as "no load control". The subsequent process ends.

[0147] (To demonstrate the complete logic, assume there are 4 or more target departments for optimization in the entire hospital, > 3, proceeding to S431) It should also be noted that if the number of target departments for optimization is not less than a preset threshold for the number of target departments, the process includes: S431 obtaining the number of priority control targets in the target control period of the target department for optimization, determining whether there is a target control period where the number of priority control targets is greater than the preset threshold for the number of control targets. If yes, proceed to step S432; otherwise, determine that the load balancing control method for the target department for optimization is not to perform load control until the cumulative duration of access traffic to the target department for optimization in the load balancing control period does not meet the requirements and exceeds the preset duration threshold, at which point port traffic control processing is required; Logic and Judgment (S431): For each "target control period" (T1, T2) of the current target department for optimization (e.g., cardiology), check its "number of priority control targets". If there is a target control period where the number of priority control targets is greater than the "preset threshold for the number of control targets" (e.g., 4), proceed to S432 for more refined judgment.

[0148] If the number of priority control targets in all target control periods does not exceed this threshold, a "delayed control" strategy is adopted: load control is not immediately implemented for this department, but a monitoring mechanism is activated. Port traffic control for this department is only triggered when the cumulative duration of the access traffic of the load balancing control target doctors (D1, D2, D3) in the load balancing control period (T1, T2) continuously "does not meet the requirements" exceeds the "preset duration threshold" (e.g., a cumulative 30 minutes).

[0149] This judgment distinguishes between two external competitive environments: During the target control period when the number of priority control targets is greater than the preset threshold for the number of control targets: At this time, the number of priority control targets is large. Since the priority control targets belong to the optimization target departments, load control processing needs to be carried out immediately. Therefore, the load pressure in the target control period will be too small. For these periods, the optimization target departments can be directly determined as not requiring load control.

[0150] During target control periods where the number of priority control targets does not exceed a preset threshold, it indicates that the number of priority control targets requiring immediate load control is relatively small, and there is still significant load optimization pressure. In this case, adopting an "observation-triggered" strategy for the optimization target departments is reasonable: allow them to operate freely initially, and only intervene when their internal pressure is genuinely and consistently manifested (cumulative duration exceeds the threshold). This avoids premature flow restriction when competition is not intense, demonstrating the prudence of the strategy.

[0151] Specific example (assuming there are 4 optimization target departments in the whole hospital, proceed to this step): For the Department of Cardiology: Priority control target number in T1 time period = 3, preset threshold = 4 → 3 is not greater than 4.

[0152] During the T2 time period, the priority control target number is 1, and the preset threshold is 4 → 1 is not greater than 4.

[0153] Neither time period exceeded the threshold. Therefore, the "Delay Control" branch was initiated: no immediate control was implemented for the cardiology department, but visitor flow for D1 / D2 at T1 and D3 at T2 was monitored. If the cumulative time for which the flow exceeded the limit (i.e., more than 50 people per second) during these time periods exceeded 30 minutes, flow control was triggered to limit it to below 10 people per second, thus not affecting the registration experience of other departments.

[0154] S432 If the number of priority control targets in the target control period is greater than the preset control target number threshold, then the load balancing control method for the optimized target department is determined to be no load control. If the number of priority control targets in the target control period is not greater than the preset control target number threshold, the target control period for which no load control is performed is obtained and used as the registration availability period. S433 Determine whether the proportion of the registration availability period in the target control period is greater than the availability period proportion threshold. If so, then load balancing control processing is performed in all target control periods where the number of priority control targets is not greater than the preset control target number threshold, i.e., port traffic control processing is performed. If not, then the load balancing control method for the optimized target department is determined to be no load control until the cumulative duration of access traffic of the optimized target hospital department in the load balancing control period does not meet the requirements and exceeds the preset duration threshold, at which point port traffic control processing is required.

[0155] Logic and Judgment (S432): This step deals with the high-intensity competition periods found in S431 where "the number of priority control targets is greater than the preset control target number threshold" (assuming there are 5 priority control targets > 4 in period T1).

[0156] For these time periods, the system decision is: "Do not implement load control for the target department during these time periods," and mark these time periods as the department's "registration available time periods."

[0157] When a large number of other popular doctors (priority control targets) are online simultaneously in the hospital, the priority control target is not considered an optimized target department and requires immediate load control. Therefore, the load pressure during the entire period is too low. In order to ensure the reliability of registration in optimized target departments, load control is not required.

[0158] Level 4 Judgment (Control Coverage Screening): Whether the Exemption Period is Dominant: The proportion of "Registration Available Periods" = (Number of Target Control Periods Exempt from Control) / (Total Number of Target Control Periods).

[0159] Logic and Judgment (S433): Calculate the proportion of "Available Registration Period" in the total target control period.

[0160] If this ratio is greater than the "available time period ratio threshold" (e.g., 50%), it means that the department has been exempted from control for most critical time periods. As a balance, in the remaining target control periods with high load pressure (number of priority control targets ≤ threshold), the system will adopt an "immediate control" strategy, that is, directly perform port traffic control during these periods.

[0161] If this proportion is not greater than the threshold, it indicates that the period for which exemption control is applied is small. In this case, the "delay control" strategy will still be adopted for the department as a whole (same as the "No" branch of S431) to ensure the availability of appointments for the target department.

[0162] Specific example (assuming T1 is the only high-intensity competition period): According to S432, T1 is marked as "registration available period", and the proportion of registration available period = 1 (T1) / 2 (total target period) = 50%.

[0163] Set the "Available Time Period Ratio Threshold" to 50%. According to the rules, immediate control is triggered only when it is "greater than" and 50% is not greater than 50%. Therefore, the "Delay Control" branch is entered, and the overall strategy is the same as the "No" branch of S431.

[0164] "Priority control targets" refer to load balancing control targets that do not belong to the current optimization target department but belong to other departments. For these "priority control targets," the system will perform immediate load control (such as port rate limiting). This means that if there are a large number of "priority control targets" within a certain period, then the competing traffic from other departments has been actively suppressed by the system, so the overall load pressure during that period is relatively low. In this case, the currently analyzed optimization target department does not need to undergo load control and can enjoy a more relaxed resource environment.

[0165] Prioritizing multiple control targets → External competition is actively suppressed by the system → Overall time-period load pressure is low → Optimized target departments can relax control (or even stop control).

[0166] Few priority control targets → External competition not suppressed → Overall workload pressure may be high during the period → Departments with optimized targets need to consider strengthening control.

[0167] In this application, global collaboration and leverage effect are employed: the system creates a "low-competition environment" for the optimized target department (K1) that needs protection by prioritizing the control of high-load doctors in other departments (priority control targets). This is an efficient resource allocation strategy that uses control over a few "critical nodes" (priority control targets) in exchange for the release of the overall service capacity of the key department (optimized target department).

[0168] Precise resource allocation: Given a limited total amount of resources, by analyzing competitive relationships, the system intelligently decides "who to prioritize" and "who to protect," achieving a precise and dynamic allocation of resources to the departments that are being optimized.

[0169] The causal relationship of the strategy is clearer: "prioritizing the control of a large number of targets" directly leads to the result of "relaxing control over the optimized target departments", making the logical chain more direct and powerful.

[0170] Improving overall system efficiency: This strategy avoids imposing indiscriminate and strict restrictions on all popular departments during peak hours. Instead, it selectively suppresses and relaxes restrictions to maintain system stability while meeting as many patients' registration needs for different departments as possible. This improves overall service throughput and satisfaction, while also ensuring the availability of registration for the optimized target departments.

[0171] In a second aspect, according to Embodiment 2, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described load balancing method for a registration system under a high-concurrency scenario when running the computer program.

[0172] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0173] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0174] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A load balancing method for a registration system under high concurrency scenarios, characterized in that, Specifically, it includes: Based on the monitoring results of the registration system's traffic data, the load balancing control period of the registration system is determined. Using the load balancing control period data and the registration data within that period, a method for identifying the load balancing control target of the registration system is determined. This identification method is then used to determine the load balancing control target of the registration system during the load balancing control period. Based on the load balancing control targets in different hospital departments and the planned registration data for those departments, optimized target hospital departments are identified. Finally, the data for the optimized target hospital departments is determined, and combined with the load balancing control period in which the optimized target department has a load balancing control target and the load balancing control target within that period, a load balancing control method for the optimized target department is determined.

2. The load balancing method for a registration system in a high-concurrency scenario as described in claim 1, characterized in that, The traffic data is determined based on the monitoring data of the access traffic of the registration system.

3. The load balancing method for a registration system in a high-concurrency scenario as described in claim 1, characterized in that, The method for determining the load balancing control period of the registration system is as follows: based on the monitoring results of the traffic data of the registration system, the data traffic of the period at different times is determined; based on the data traffic, the dates of the period that belong to the traffic peak period are determined; based on the date data of the period that belong to the traffic peak period, it is determined whether the period belongs to the load balancing control period of the registration system.

4. The load balancing method for a registration system in a high-concurrency scenario as described in claim 3, characterized in that, The time period is defined as the date when the proportion of monitoring times during which the data flow exceeds the rated flow threshold does not meet the requirement, i.e., the time period when the proportion of monitoring times during which the data flow exceeds the rated flow threshold is greater than the preset monitoring time proportion threshold.

5. The load balancing method for a registration system in a high-concurrency scenario as described in claim 3, characterized in that, Based on the date data of the time period belonging to the peak traffic period, determine whether the time period belongs to the load balancing control period of the registration system. Specifically, if the time period belongs to the peak traffic period on different dates, then determine that the time period belongs to the load balancing control period of the registration system.

6. The load balancing method for a registration system in a high-concurrency scenario as described in claim 1, characterized in that, The method for determining the load balancing control target of the registration system is as follows: using the load balancing control period data, determine the number of load balancing control periods; based on the registration data of the load balancing control periods, determine the dates of the load balancing control periods that belong to the busy traffic periods, and take them as busy dates; based on the number of load balancing control periods and similar data of busy dates of different load balancing control periods, determine the method for identifying the load balancing control target of the registration system.

7. The load balancing method for a registration system in a high-concurrency scenario as described in claim 6, characterized in that, If the number of load balancing control periods is greater than the preset control period number threshold, then the method for identifying the load balancing control target of the registration system is to take doctors whose access traffic in the most recent preset time period does not meet the requirements and whose number of dates is greater than the date number threshold as load balancing control targets.

8. The load balancing method for a registration system in a high-concurrency scenario as described in claim 1, characterized in that, The method for determining the load balancing control method for the optimized target department is as follows: the load balancing control period in which the optimized target department has a load balancing control target is taken as the target control period; Based on the load balancing control target data during the target control period, determine the load balancing control targets other than the optimized target departments during the target control period, and take them as priority control targets. The load balancing control method for the optimized target hospital department is determined based on the optimized target department's departmental data, the target control period of the optimized target department, and the priority control targets in different target control periods.

9. The load balancing method for a registration system in a high-concurrency scenario as described in claim 8, characterized in that, Based on the optimized target hospital department data, the number of optimized target hospital departments is determined. If the number of optimized target departments is less than a preset optimized department number threshold, then the load balancing control method for the optimized target departments is determined to be no load control.

10. A computer system, comprising: A memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, characterized in that, when the processor runs the computer program, it executes a load balancing method for a registration system in a high-concurrency scenario as described in any one of claims 1-9.

Images