An order wave-oriented sorting worker performance real-time evaluation method

Abstract

This invention discloses a real-time evaluation method for sorting personnel efficiency based on order waves, belonging to the field of mechanical processing technology. The method includes: Step 1, extracting the product type count, accumulated value of irregularly shaped part markers, and spatial dispersion of the order wave; obtaining the difficulty increment value as a complexity fingerprint through a lookup table matrix; Step 2, superimposing the difficulty increment value onto the product of the total number of items in the wave and the basic time consumption per item to obtain the standard time budget for the wave; Step 3, reading the actual time consumption, recording the abnormal duration of non-efficiency events, and subtracting the abnormal duration from the actual time consumption to obtain the net actual time consumption. This invention can construct a complexity fingerprint by breaking down the smallest picking action unit, quantifying difficulty factors such as the proportion of irregularly shaped parts and spatial dispersion of the storage location, and converting them into time increments. This effectively avoids the erroneous underestimation of personnel efficiency in high-difficulty wave operations, ensuring that the evaluation results accurately reflect the actual operational capabilities of personnel.

Description

Technical Field

[0001] This invention relates to the field of machining technology, and more specifically, to a method for real-time evaluation of the efficiency of sorting workers based on order waves. Background Technology

[0002] In the production process of a machining plant, the sorting of mechanical parts is a key step that connects the processing steps with the downstream assembly and transfer links, directly affecting the factory's production cycle, parts flow efficiency, and the stability of subsequent processes. Currently, the efficiency evaluation of mechanical parts sorting personnel in machining plants still adopts the traditional single evaluation model, which can no longer adapt to the complexity and special characteristics of mechanical parts sorting scenarios.

[0003] Existing assessment methods often rely solely on the number of items picked and processing time as evaluation indicators, neglecting factors such as differences in order wave difficulty and operational complexity. In the sorting operations of machining plants, the mechanical components of different order waves vary significantly. Some waves consist of small, single-specification mechanical components, with their locations concentrated in the same warehouse aisle, making sorting simple and time-efficient. However, other waves contain mechanical components of various specifications, often including irregularly shaped components such as fragile precision parts and oversized parts, with their locations scattered across multiple warehouse areas. This necessitates frequent switching of work areas during sorting, while simultaneously ensuring the protection and precise handling of irregularly shaped components. The difficulty has increased significantly. For example, in two different sorting waves of mechanical parts, the first wave consists of 30 small mechanical parts of a single specification, with the storage locations concentrated in the same aisle, and the operator can pick 120 pieces per hour. The second wave consists of 20 orders of mechanical parts of multiple specifications, including 10 irregularly shaped parts, with the storage locations scattered across 3 warehouse areas, and the operator can only pick 50 pieces per hour. However, the current assessment uses a uniform standard, with "pieces / hour" as the sole criterion for evaluation. This leads to the operator who completes the second wave being wrongly judged as inefficient, when in fact the difficulty of the work they undertake is far greater than that of the first wave, and the assessment results do not match the actual work capacity.

[0004] It is evident that existing technologies lack a quantifiable and embeddable evaluation mechanism for the inherent complexity of wave sorting. In mechanical parts sorting, the diversity of parts specifications, the time required to handle irregularly shaped parts, and the cost of spatial displacement are fundamental variables affecting operational efficiency. However, existing methods cannot transform these complexity factors into benchmark values ​​that are dynamically correlated with personnel action time. In complex wave sorting scenarios with multiple specifications, irregularly shaped parts, and storage locations scattered across multiple warehouse areas, the pursuit of efficiency indicators such as the number of items picked per unit time leads to the systematic neglect of operational difficulty and the underestimation of the true efficiency of highly skilled personnel, which is not conducive to motivating personnel. Summary of the Invention

[0005] To address the problems existing in the prior art, the purpose of this invention is to provide a real-time evaluation method for sorting personnel efficiency based on order waves. This method can construct a complexity fingerprint by breaking down the smallest picking action unit, quantify difficulty factors such as the proportion of irregularly shaped items and the dispersion of storage space, and convert them into time increments. This effectively avoids the erroneous underestimation of personnel efficiency in high-difficulty wave operations, ensuring that the evaluation results closely reflect the actual operational capabilities of personnel.

[0006] To solve the above problems, the present invention adopts the following technical solution:

[0007] A real-time performance evaluation method for sorting workers based on order waves includes:

[0008] Step 1: Extract the product type count, the cumulative value of the irregular part flag, and the spatial dispersion of the storage location for each wave of orders. Obtain the difficulty increment value by looking up the table matrix, which serves as the complexity fingerprint.

[0009] Step 2: Add the difficulty increment value to the product of the total number of items in the wave and the basic time consumption per item to obtain the standard time budget for the wave.

[0010] Step 3: Read the actual time consumed, record the abnormal duration of non-performance events, and subtract the abnormal duration from the actual time consumed to obtain the net actual time consumed;

[0011] Step 4: Calculate the relative efficiency coefficient as the ratio of the standard time budget per wave to the net actual time spent;

[0012] Step 5: Store the complexity fingerprint, relative efficiency coefficient and operator identifier. When the same complexity fingerprint accumulates to a preset number of times, take the median of the relative efficiency coefficient of the corresponding wave. If it deviates from the preset benchmark value, adjust the difficulty increment value of the lookup table matrix by the reciprocal of the median.

[0013] Step 6: Based on the calibrated memory bank and historical relative efficiency coefficients, retrieve the distribution of historical relative efficiency coefficients for the new wave, assign high-difficulty waves to personnel with high historical relative efficiency coefficients, and low-difficulty waves to personnel with low historical relative efficiency coefficients. When a person's relative efficiency coefficient is lower than the predetermined efficiency threshold for a predetermined number of consecutive times under a specific difficulty fingerprint, output a training instruction for grasping skills practice to that person.

[0014] Further, step 1 includes:

[0015] Step 11: Decompose each order within the wave into a sequence of the smallest picking action units. This sequence includes a walking unit, a positioning unit, a gripping unit, and a placement unit. Assign an action time correction factor to each gripping unit based on the pre-stored irregular part attributes in the product code library.

[0016] Step 12: Based on the action unit sequence and the time correction factor of the grasping unit, calculate the theoretical walking path length and theoretical grasping number, count the number of grasping units corresponding to the irregular parts, and output the standardized action base value including the walking base value, the grasping base value and the irregular part proportion coefficient.

[0017] Step 13: Extract the walking base value and the grasping base value from the standardized action base values, combine them with the total number of items in the wave, construct a nonlinear penalty function for the spatial dispersion of the storage location, and output the spatial dispersion penalty value.

[0018] Furthermore, step 1 also includes:

[0019] Step 14: Weight the spatial discrete penalty value and the grab base value in the standardized action base value, and introduce a dynamic correction coefficient to output the preliminary difficulty integral;

[0020] Step 15: Map the initial difficulty integral to the time increment dimension, output the difficulty increment value through a piecewise linear transformation table, and use the difficulty increment value as the complexity fingerprint.

[0021] Further, step 2 includes:

[0022] Step 21: Obtain the historical waves with the same complexity fingerprint as the current wave, extract the difference between the actual time consumption and the difficulty increment value of each historical wave and divide it by the total number of items in that wave to obtain a sample group of basic time consumption per item, and take the median of the sample group as the basic time consumption per item.

[0023] Step 22: Dynamically determine the weighting of the difficulty increment value based on the ratio of the proportion coefficient of irregular parts in the current wave to the dispersion of the storage space; when the ratio exceeds the predetermined threshold, the difficulty increment value is added to the product of the total number of parts in the wave and the basic time consumption per part with full weight; otherwise, the difficulty increment value is multiplied by the attenuation factor and then added; output the standard time budget for the wave.

[0024] Further, step 3 includes:

[0025] Step 31: Monitor the movement trajectory and dwell time. When the continuous dwell time exceeds the predefined threshold and the dwell location is not within the coordinate range of the picking point, mark the period corresponding to the continuous dwell as a candidate abnormal period.

[0026] Step 32: Verify the overlap between the candidate abnormal time period and the equipment failure time window, take the intersection as the final abnormal duration, subtract the final abnormal duration from the actual time consumption, and output the net actual time consumption.

[0027] Further, step 4 includes:

[0028] Step 41: Divide the net actual time into time segments according to the natural pause point, calculate the ratio of the number of items picked in each segment to the segment duration as the instantaneous efficiency, remove segments with a ratio lower than half or higher than twice the average of the overall instantaneous efficiency, and sum the duration of the remaining segments to output a stable net time.

[0029] Step 42: Use the ratio of stable net time consumption to the standard time budget of the wave as the original efficiency coefficient; obtain multiple original efficiency coefficients calculated in the same way for the last predetermined number of waves under the same complexity fingerprint and sorted by completion time for the same operator; take the median of multiple original efficiency coefficients and the current original efficiency coefficient as a weighted average, and the weight of the current coefficient decreases as the cumulative number of waves with the same fingerprint increases, and output a smooth efficiency coefficient;

[0030] Step 43: Calculate the penalty factor based on the number of occurrences of ineffective events and the total duration of anomalies. The penalty factor is equal to one minus the product of the number of occurrences and the total duration of anomalies divided by the sum of the standard time budget for the wave and the predetermined constant. Multiply the smoothed efficiency coefficient by the penalty factor to output the final relative efficiency coefficient.

[0031] Further, step 5 includes:

[0032] Step 51: Store the complexity fingerprint, the final relative efficiency coefficient, and the operator identifier into the difficulty memory bank, and establish an independent coefficient sequence according to the complexity fingerprint and arrange them in chronological order;

[0033] Step 52: When the number of records in the coefficient sequence corresponding to the fingerprint of the same complexity reaches the sample capacity threshold, extract all the final relative efficiency coefficients in the sequence, group them according to the operator identification, calculate the median of the coefficients in each group, and then take the median of the medians of each group as the group efficiency benchmark of the fingerprint.

[0034] Step 53: Divide the difference between the group performance benchmark and the preset standard benchmark value by the standard benchmark value to obtain the deviation ratio.

[0035] Furthermore, step 5 also includes:

[0036] Step 54: When the absolute value of the deviation ratio does not exceed the dead zone threshold, the current difficulty increment value remains unchanged; when it exceeds the dead zone threshold, the adjustment coefficient is set to the negative of the deviation ratio plus one, and the adjustment coefficient is limited to the predetermined adjustment range.

[0037] Step 55: Multiply the adjustment coefficient by the original difficulty increment value corresponding to the fingerprint of that complexity in the lookup table matrix, write the product result back to the corresponding entry in the lookup table matrix, and clear all records in the coefficient sequence corresponding to the fingerprint.

[0038] Further, step 6 includes:

[0039] Step 61: Extract the complexity fingerprint corresponding to the new wave from the calibrated difficulty memory, obtain the historical relative efficiency coefficient sequence of each operator under the fingerprint, assign decreasing weights to the sequence according to time sequence, and calculate the weighted average as the predicted efficiency value.

[0040] Step 62: Sort the predicted performance values ​​of all personnel under the fingerprint, take the first percent as the high-performance group, the last percent as the low-performance group, and the rest as the normal group, and calculate the standard deviation of the predicted performance values. When the standard deviation exceeds the predetermined fluctuation threshold, the personnel in the normal group who are close to the high-performance boundary are classified into the high-performance group, and the personnel who are close to the low-performance boundary are classified into the low-performance group.

[0041] Furthermore, step 6 also includes:

[0042] Step 63: Compare the difficulty increment value of the new wave with the high difficulty threshold and the low difficulty threshold to determine the wave difficulty type. Assign high difficulty waves to the high-efficiency group, low difficulty waves to the low-efficiency group, and normal difficulty waves to the normal group.

[0043] Step 64: Select wave types where the proportion coefficient of irregular parts exceeds the predetermined proportion threshold and the dispersion of the storage space exceeds the predetermined dispersion threshold; monitor the historical relative efficiency coefficient sequence of each person under this wave type; when a person completes this wave type consecutively for a predetermined number of consecutive times and the relative efficiency coefficient of each time is lower than the predetermined efficiency threshold, output the training instruction for grasping skills exercise.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] (1) This solution constructs a complexity fingerprint by breaking down the smallest picking action unit, quantifies the difficulty factors such as the proportion of irregular parts and the dispersion of storage space, and converts them into time increments. This effectively avoids the underestimation of the efficiency of personnel in high-difficulty wave operations and makes the evaluation results consistent with the actual operational capabilities of personnel.

[0046] (2) This solution establishes a closed-loop dynamic calibration system, accumulates multi-dimensional operation data based on the difficulty memory bank, corrects the difficulty increment value of the lookup table matrix through the group effectiveness benchmark, adapts to changes in workshop sorting scenarios and component categories, continuously optimizes the standard time budget, and ensures that the long-term evaluation process is accurate, stable and reliable.

[0047] (3) This solution generates predicted performance values ​​based on historical performance data of personnel and completes intelligent grouping, accurately matches the corresponding performance level of the order waves of different difficulties, reasonably optimizes the allocation of human resources, reduces the bottleneck of high-difficulty operations and the redundancy of manpower in low-difficulty operations, and improves the overall sorting and circulation efficiency.

[0048] (4) This solution can automatically identify high-difficulty operation scenarios with many irregular parts and scattered storage locations, continuously monitor the long-term performance of personnel, and push targeted training instructions on grasping skills to personnel who fail to meet the standards, accurately locate skill shortcomings, and achieve long-term iterative improvement of the team's sorting operation capabilities. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0050] Figure 1 This is a flowchart illustrating the overall steps of the present invention.

[0051] Figure 2 This is a flowchart illustrating the calibration process for the difficulty increment value of this invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0053] Please see Figures 1 to 2 A real-time performance evaluation method for sorting workers based on order waves, the method comprising:

[0054] Step 1: Extract the product type count, the cumulative value of the irregular item flag, and the spatial dispersion of the storage location for each wave of orders. Obtain the difficulty increment value by looking up the table matrix, which serves as the complexity fingerprint. The specific operations are as follows:

[0055] Three key difficulty feature parameters were extracted from the wave order data: product type count, accumulated value of irregularly shaped component markers, and spatial dispersion of storage locations. The product type count reflects the diversity of goods to be sorted within a wave; more types result in higher identification and switching costs during sorting. The accumulated value of irregularly shaped component markers reflects the quantity of irregularly shaped components within a wave; a larger accumulated value increases the difficulty of protection and grasping during sorting. Spatial dispersion reflects the degree of distribution of goods storage locations within a wave; higher dispersion increases the walking and positioning costs during sorting. After extracting these three parameters, they were used as input parameters in the preprocessing... A lookup table matrix is ​​established, which is based on a large amount of historical sorting operation data. By statistically analyzing the differences in sorting difficulty corresponding to different parameter combinations, a pre-set table of the correspondence between parameters and difficulty increment values ​​is used. By looking up the table, the difficulty increment value matching the difficulty of the current wave can be obtained. This difficulty increment value can comprehensively represent the overall sorting difficulty of the current wave. Therefore, it is used as a complexity fingerprint for subsequent wave standard time budget calculation, efficiency coefficient calibration, and wave allocation. This ensures that all subsequent operations are carried out based on a realistic difficulty benchmark, avoiding efficiency misjudgments caused by a single standard in traditional assessments.

[0056] Step 1 further includes the following steps:

[0057] Step 11: Decompose each order within a wave into a sequence of smallest picking action units. This sequence includes a walking unit, a positioning unit, a gripping unit, and a placement unit. Based on the pre-stored irregular part attributes in the product code library, assign an action time correction factor to each gripping unit. The specific operations are as follows:

[0058] For each order within a wave, it is broken down into an indivisible sequence of picking action units according to the actual sorting process. This sequence strictly corresponds to the complete sorting process and specifically includes a walking unit, a positioning unit, a gripping unit, and a placement unit. The walking unit corresponds to the movement of the operator between different storage locations, the positioning unit corresponds to the operator finding and accurately positioning the target storage location, the gripping unit corresponds to the operator grabbing the goods from the storage location, and the placement unit corresponds to the operator placing the grabbed goods into the designated sorting container. Each action unit corresponds to a specific time consumption characteristic. The time consumption of the gripping unit is most affected by the product attributes, especially the gripping time of irregularly shaped items, which is significantly different from that of ordinary items. Therefore, it is necessary to rely on a pre-established product code library, which stores the detailed attributes of all products, including the attribute identifier of irregularly shaped items. By querying the product code library, the irregularly shaped attributes of the products corresponding to each gripping unit can be obtained, and then a corresponding action time correction factor can be assigned to each gripping unit.

[0059] The specific method for obtaining the motion time correction factor is as follows: First, all types of goods are classified, distinguishing between ordinary parts and different types of irregularly shaped parts, such as fragile precision parts and oversized parts. For each type of goods, no less than 30 samples are selected. Under standard operating conditions, such as temperature 22±2℃, humidity 50±5%, and no environmental interference, at least 3 skilled operators conduct repeated grasping tests, recording the grasping time of each sample. The ratio of the grasping time of each type of goods to the grasping time of ordinary parts is calculated. This ratio is the initial motion time correction factor for the grasping unit of that type of goods. Subsequently, the initial correction factor is calibrated by combining actual sorting data from no less than 100 historical waves. After removing outliers, the mean is taken. Outliers are usually data that exceed the mean ± 2 standard deviations. Finally, the mean is used as the final motion time correction factor. This correction factor is used to quantify the impact of irregularly shaped parts on the grasping motion time. The higher the grasping difficulty of irregularly shaped parts, the larger the corresponding time correction factor, thereby achieving refined correction of the grasping motion time.

[0060] Step 12: Based on the action unit sequence and the time correction factor of the grasping unit, calculate the theoretical walking path length and theoretical grasping count, and count the number of grasping units corresponding to the irregular parts. Output the standardized action base value, which includes the walking base value, the grasping base value, and the irregular part proportion coefficient. The specific operation is as follows:

[0061] Based on the minimum picking action unit sequence obtained in step 11, and combined with the action time correction factor corresponding to each gripping unit, the key action parameters of the sorting operation are cumulatively calculated, including the theoretical walking path length and the theoretical number of grips. The theoretical walking path length is calculated based on the coordinate position of each storage location and the distribution of walking units in the action unit sequence. It is the theoretical distance that the operator needs to move to complete the sorting of this wave. Its calculation logic fits the actual distribution of storage locations and walking trajectories in the operation and can reflect the overall load of walking actions. The theoretical number of grips is the cumulative statistics of gripping units in the action unit sequence, which directly corresponds to the total number of items that need to be gripped in the wave, reflecting the overall load of gripping actions. At the same time, the number of irregularly shaped items in all gripping units is counted, that is, the number of gripping units corresponding to irregularly shaped items. Combined with the theoretical number of grips, the irregularly shaped item proportion coefficient can be obtained. This coefficient reflects the proportion of irregularly shaped items in the wave. The higher the proportion, the greater the sorting difficulty.

[0062] Based on the above calculation results, the output includes standardized action base values, including walking base values, grasping base values, and irregular part proportion coefficients. The walking base value is a parameter obtained by standardizing the theoretical walking path length and is used to quantify the difficulty benchmark of the walking action. The grasping base value is a parameter obtained by standardizing the theoretical grasping number combined with the action time correction factor and is used to quantify the difficulty benchmark of the grasping action. The irregular part proportion coefficient directly reflects the proportion of irregular parts' influence on the sorting difficulty. The three together constitute the basic parameter system for subsequent difficulty calculations.

[0063] Step 13: Extract the walking baseline value and the grasping baseline value from the standardized action baseline values, combine them with the total number of items in each wave, construct a nonlinear penalty function for the spatial dispersion of the storage location, and output the spatial dispersion penalty value. The specific operation is as follows:

[0064] From the standardized action baseline values ​​output in step 12, the walking baseline value and the grasping baseline value are extracted. These two parameters reflect the basic difficulty of the walking and grasping actions, respectively. The impact of the storage space dispersion is mainly reflected in the additional time spent on walking actions and the switching cost of grasping actions. Therefore, it is necessary to construct a nonlinear penalty function that can quantify this additional load by combining the total number of items per wave. The higher the storage space dispersion, the more dispersed the operators need to move between more storage locations, and the walking distance and positioning time will increase nonlinearly. At the same time, the switching frequency of grasping actions will also increase, resulting in a nonlinear increase in the overall sorting difficulty. Therefore, using a nonlinear penalty function can more accurately fit this difficulty change pattern. The derivation logic of this nonlinear penalty function is as follows: the walking baseline value reflects the basic walking load, the grasping baseline value reflects the basic grasping switching load, and the total number of items per wave reflects the total amount of work. The three together determine the weight of the impact of storage space dispersion on sorting difficulty. The higher the dispersion, the larger the penalty value, and the penalty intensity increases nonlinearly with the increase of dispersion, so as to fit the influence of dispersion on time consumption in actual operations. Its formula is: ;

[0065] The derivation logic of this formula is as follows: Considering that the impact of the dispersion D of the storage space on walking motion increases logarithmically, that is, the impact is gradual when the dispersion is small and increases sharply when the dispersion is large, a logarithmic term is introduced. Combined with the walking baseline value W; simultaneously, the impact of dispersion on grasping switching is directly proportional to the dispersion and inversely proportional to the total number of pieces N in the wave, therefore, it is introduced... Combined with the base value G, and then through weighting coefficients , Balancing the effects of both, the spatial dispersion penalty value P is finally obtained; in the formula, P is the spatial dispersion penalty value, which is used to quantify the additional difficulty brought about by the spatial dispersion of the cargo location; The walking baseline weighting coefficient is pre-calibrated using historical data, with a value range of 0.6 to 0.8, reflecting the degree of influence of the walking baseline on the penalty value; W is the walking baseline in the standardized action baseline; D is the spatial dispersion of the cargo location, characterized by the standard deviation calculated from the cargo location coordinates. The base value weighting coefficient is pre-calibrated using historical data, with a value range of 0.2 to 0.4, reflecting the degree of influence of the base value on the penalty value; G is the base value of the gripping action; N is the total number of items in the wave; the spatial dispersion penalty value calculated by this formula can accurately quantify the additional impact of the spatial dispersion of the storage location on the sorting difficulty.

[0066] Step 14: Weight the spatial discrete penalty value and the grab base value in the standardized action base value, and introduce a dynamic correction coefficient to output the preliminary difficulty integral. The specific operation is as follows:

[0067] The spatial dispersion penalty value quantifies the additional difficulty caused by the dispersed distribution of storage locations, while the grasping baseline value quantifies the basic difficulty of the grasping action. Both are factors affecting the difficulty of wave sorting, but their impact on the overall difficulty has different weights. Therefore, a weighted sum is needed to comprehensively reflect their combined influence. The weights for the weighted sum are set based on historical sorting operation data. By statistically analyzing the impact of the spatial dispersion penalty value and the grasping baseline value on the actual sorting time in different waves, the weight ratios of the two are pre-set to ensure that the summation result closely reflects the actual difficulty situation. Furthermore, considering that the impact of the spatial dispersion penalty value and the grasping baseline value may vary slightly under different sorting scenarios, such as different warehouse layouts and different product types, a dynamic correction coefficient is introduced. The specific method for obtaining this coefficient is as follows:

[0068] First, all common sorting scenarios were analyzed, categorized by warehouse layout (single-channel, multi-channel, cross-channel) and product type (primarily general-purpose, irregularly shaped, mixed). For each scenario, at least 50 historical wave data points were selected, and the deviation between the base difficulty score and the time increment corresponding to the actual difficulty in each wave was calculated. Linear regression analysis was then used to obtain the initial dynamic correction coefficient for that scenario. Subsequently, a rolling calibration method was adopted, updating the initial correction coefficient every 20 waves of the same scenario. After removing abnormal wave data (e.g., waves with equipment malfunctions or human error), the linear regression coefficient was recalculated. The updated dynamic correction coefficient can be dynamically adjusted according to the actual situation of the current sorting scenario. The adjustment is based on factors such as the layout characteristics of the current warehouse area and the overall attributes of the goods, ensuring that the calculation of the preliminary difficulty score can adapt to the differences in different scenarios and improve the accuracy of difficulty quantification. First, the spatial discrete penalty value and the grasping base value are multiplied by their respective weights to obtain the weighted components of the two. Then, the two weighted components are added together to obtain the basic difficulty score. Finally, the basic difficulty score is multiplied by the dynamic correction coefficient to obtain the preliminary difficulty score. The preliminary difficulty score can comprehensively reflect the basic action difficulty and location discrete difficulty of the wave sorting, and is a preliminary quantification of the overall difficulty of the wave.

[0069] Step 15: Map the initial difficulty integral to the time increment dimension, output the difficulty increment value through a piecewise linear transformation table, and use this difficulty increment value as the complexity fingerprint. The specific operation is as follows:

[0070] The initial difficulty integral is a quantification of the sorting difficulty of a wave, but this integral does not correspond to a specific time dimension and cannot be directly used to calculate the standard time budget for a wave. Therefore, it needs to be mapped to a time increment dimension, that is, the difficulty integral is converted into a corresponding time increment. This time increment reflects the additional time length that needs to be added to the basic sorting time due to the wave difficulty. The mapping process is implemented through a piecewise linear transformation table, which is constructed based on a large amount of historical sorting operation data. It pre-defines the linear correspondence between different initial difficulty integral intervals and their corresponding time increments. Its construction logic is as follows: statistically analyze the actual additional time required for different initial difficulty integrals in historical waves. The time consumption is divided into multiple intervals, and a linear relationship between the points and the time increment is established within each interval to ensure that the converted time increment can accurately reflect the actual time consumption impact corresponding to the difficulty points. Based on the initial difficulty points value, the piecewise linear transformation table is consulted to find the corresponding time increment interval. The corresponding difficulty increment value is calculated using the linear transformation formula for that interval. Since this difficulty increment value can comprehensively characterize all key difficulty features of the wave, such as the diversity of product types, the number of irregularly shaped items, and the spatial dispersion of storage locations, and has a clear time dimension, it can uniquely reflect the overall sorting difficulty of the current wave. Therefore, this difficulty increment value is defined as the complexity fingerprint.

[0071] In a preferred embodiment of the present invention, step 2 is further included: adding the difficulty increment value to the product of the total number of items in the wave and the basic time consumption per item to obtain the standard time budget for the wave. The specific operation is as follows:

[0072] The base time for sorting in a wave is obtained by multiplying the total number of items in the wave by the base time per item. This base time reflects the baseline time required to complete the wave without considering additional difficulty. The difficulty increment value obtained in step 1 is then added to this base time. This difficulty increment value corresponds to the additional time in the wave caused by factors such as product type, number of irregularly shaped items, and location dispersion. The sum of these values ​​yields a standard time budget that comprehensively reflects the actual difficulty of the wave. The formula for calculating the standard time budget for a wave is: ;

[0073] The derivation logic of this formula is: the total number of pieces N in a wave and the basic time consumption per piece. The product of these values ​​corresponds to the basic sorting time for a normal difficulty wave, reflecting the basic workload of the sorting operation. The difficulty increment value ΔT corresponds to the additional time brought about by the difficulty of the wave. Its impact on the total standard time is dynamically adjusted by superimposing a weight ω, ensuring that the standard time budget for different difficulty waves can accurately match the actual operational needs and avoiding performance misjudgments caused by a single basic time. In the formula, The standard time budget for each wave is in minutes; N is the total number of items in the wave, with no unit. The time taken per unit is the basic time, expressed in minutes per unit. The weighting factor is the sum of the difficulty increment values, and its value ranges from 0 to 1; The difficulty increment value obtained in step 1 is expressed in minutes. This formula achieves a dynamic match between the standard time budget and the wave difficulty by superimposing the base time and the additional difficulty time.

[0074] Step 2 further includes the following steps:

[0075] Step 21: Obtain historical waves with the same complexity fingerprint as the current wave. Extract the difference between the actual time taken and the difficulty increment value for each historical wave and divide it by the total number of items in that wave to obtain a sample group of basic time taken per item. Take the median of this sample group as the basic time taken per item. The specific operation is as follows:

[0076] From the pre-established difficulty memory, all historical waves with the same complexity fingerprint as the current wave are retrieved. The complexity fingerprint serves as the unique quantitative representation of wave difficulty. Historical waves with the same complexity fingerprint are consistent with the current wave in key difficulty characteristics such as product type, number of irregularly shaped items, and spatial dispersion of storage locations. Their basic sorting time is of reference value. For each retrieved historical wave, its actual time and difficulty increment value are extracted. The actual time is the total time of sorting operations for that historical wave, and the difficulty increment value is the additional time corresponding to that historical wave. The difference between the two is the basic time of that historical wave after removing the influence of additional difficulty. This basic time only reflects the basic load of the sorting operation itself and is unrelated to the wave difficulty.

[0077] Dividing the base time by the total number of items in that historical wave yields the base time per item for that historical wave. The base times per item from all historical waves constitute the sample group for base times per item. To avoid the impact of outliers on the accuracy of the base time per item, the median of the sample group is selected as the final base time per item. The median effectively avoids time deviations caused by abnormal situations such as human error or temporary equipment failure in individual historical waves, ensuring the stability and accuracy of the base time per item. The calculation of the base time per item must meet the sample group size requirement; the number of retrieved historical waves must be no less than 30. If less than 30 waves are retrieved, data from waves of the same type and similar difficulty must be added until the sample size meets the requirement, ensuring the reliability of the statistical results. Those skilled in the art can directly calculate the base time per item using this method.

[0078] Step 22: Dynamically determine the weighting of the difficulty increment value based on the ratio of the proportion coefficient of irregularly shaped parts in the current wave to the dispersion of the storage space; when this ratio exceeds a predetermined threshold, the difficulty increment value is added with full weight to the product of the total number of parts in the wave and the basic time consumption per part; otherwise, the difficulty increment value is multiplied by a decay factor before being added; output the standard time budget for the wave. The specific operation is as follows:

[0079] First, by using the ratio of the proportion coefficient of irregularly shaped parts to the dispersion of storage space in the current wave, the dominant influence of irregularly shaped parts and storage space dispersion on the wave difficulty is determined, thus determining the value of the superimposed weight. Then, the ratio of the proportion coefficient of irregularly shaped parts to the dispersion of storage space in the current wave is calculated. The proportion coefficient of irregularly shaped parts comes from the standardized action base value output in step 12, reflecting the proportion of irregularly shaped parts within the wave; the dispersion of storage space comes from the calculation result in step 13, reflecting the degree of dispersion of storage space distribution. This ratio quantifies the relative influence of the two difficulty factors. The larger the ratio, the more significant the influence of irregularly shaped parts on the wave difficulty; the smaller the ratio, the more significant the influence of storage space dispersion on the wave difficulty. The specific acquisition of the predetermined threshold... The method is as follows: Select no less than 100 historical batches with different difficulties, calculate the ratio of the proportion of irregularly shaped parts to the dispersion of the storage space for each batch, and simultaneously analyze the correlation between this ratio and the proportion of additional difficulty time in the actual sorting time. Determine the threshold through linear fitting. When the ratio exceeds the threshold, it indicates that irregularly shaped parts are the dominant factor in the batch difficulty, and the difficulty increment value should be fully weighted, i.e., the weighting ω is set to 1, to ensure that the additional time caused by irregularly shaped parts is fully included in the standard time budget. When the ratio does not exceed the threshold, it indicates that the dispersion of the storage space is the dominant factor in the batch difficulty, and the additional time it causes is relatively controllable. The difficulty increment value should be multiplied by a decay factor before being superimposed to avoid the standard time budget being too high.

[0080] The specific method for obtaining the attenuation factor is as follows: For historical waves dominated by the spatial dispersion of the cargo location, select no less than 50 samples, calculate the deviation between the difficulty increment value and the actual additional time consumption in each sample, and determine the initial value of the attenuation factor through statistical analysis. The initial value ranges from 0.5 to 0.8. Then, a rolling calibration method is adopted. After accumulating 30 waves of the same type, abnormal data is removed and the attenuation factor is recalculated to ensure that it can fit the actual operation situation. After the superposition is completed, the final standard time budget of the wave is output. This budget can be dynamically adjusted according to the wave difficulty characteristics, which further improves the accuracy of subsequent performance evaluation.

[0081] In a preferred embodiment of the present invention, step 3 is further included: reading the actual time consumed, recording the abnormal duration of non-performance events, and subtracting the abnormal duration from the actual time consumed to obtain the net actual time consumed. The specific operation is as follows:

[0082] First, the actual time taken by the operator to complete the current wave of sorting is read. This actual time is the total time from the start of the sorting task to the completion of all sorting actions and placement of goods into designated containers. It includes normal operation time and abnormal time caused by various non-efficiency events. Then, the abnormal time corresponding to all non-efficiency events is accurately recorded. Non-efficiency events refer to events unrelated to the operator's own operational ability and that cannot reflect their efficiency level, mainly including operation interruption time caused by equipment failure, handling of sudden anomalies, etc. Finally, the recorded abnormal time is deducted from the actual time. The time obtained after deduction is the net actual time. This time only includes the time taken by the operator to perform normal sorting actions, objectively and truthfully reflecting their actual efficiency level in the current wave of sorting, providing an accurate time benchmark for subsequent calculation of relative efficiency coefficients, and ensuring the fairness and accuracy of efficiency evaluation results. The calculation logic of net actual time is as follows: ;

[0083] The derivation logic of this formula is as follows: actual time consumption consists of two parts: normal operation time and abnormal operation time. The net actual time consumption only needs to retain the normal operation time; therefore, it is obtained by subtracting the abnormal operation time from the actual time consumption. In the formula, The actual net time consumed is in minutes. The actual time taken is in minutes. This is the final abnormal duration, or the total abnormal duration, expressed in minutes.

[0084] Step 3 further includes the following steps:

[0085] Step 31: Monitor the movement trajectory and dwell time. When the continuous dwell time exceeds a predefined threshold and the dwell location is not within the coordinate range of the picking point, mark the period corresponding to the continuous dwell time as a candidate abnormal period. The specific operation is as follows:

[0086] By pre-deploying positioning devices on workers, such as RFID locators and Bluetooth positioning modules, the movement trajectory and dwell time of workers are monitored in real time. The positioning accuracy of the positioning devices must meet the requirements of sorting operations, with an error of no more than 0.5 meters, to ensure the accuracy of movement trajectory and dwell position monitoring. The specific method for obtaining the predefined threshold is as follows: Select no less than 50 sets of sample data from normal sorting operations, and statistically analyze the continuous dwell time of workers during normal sorting processes, such as locating storage locations, grabbing goods, and placing goods. Calculate the mean and standard deviation of the dwell time of all samples. Add twice the standard deviation to the mean as the predefined threshold. This threshold is usually set between 2 and 3 minutes. If the continuous dwell time exceeds this threshold, it indicates that the dwelling behavior exceeds the reasonable range of normal sorting operations and may be due to a non-functional event.

[0087] The method for determining the coordinate range of picking points is as follows: the coordinate information of all picking points is pre-entered into the system, and a circular area with a radius of 0.5 to 1 meter is delineated as the coordinate range of the picking point, based on the center coordinates of each picking point. This range can cover the activity area of ​​the operator performing normal sorting operations at the picking point. During the monitoring process, when it is detected that the continuous dwell time of the operator exceeds the predefined threshold and the coordinates of the dwell position are not within the coordinate range of any picking point, it indicates that the dwell time is not required for normal sorting operations and is likely caused by a non-functional event. Therefore, this dwell time is marked as a candidate abnormal period.

[0088] Step 32: Verify the overlap between the candidate abnormal time period and the equipment failure time window, take the intersection as the final abnormal duration, subtract this final abnormal duration from the actual time, and output the net actual time. The specific operation is as follows:

[0089] By overlapping and verifying the candidate abnormal time period with the equipment failure time window, candidate abnormal time periods that are not caused by equipment failure are eliminated, and only the overlapping part is retained as the final abnormal duration. The equipment failure time window refers to the start and end time interval of the equipment failure. This time window is recorded in real time by the factory's equipment control system and synchronized to this evaluation system. It includes key information such as the number of the faulty equipment, the time of failure, and the time of failure resolution, to ensure the accuracy and real-time nature of the time window.

[0090] The specific process of overlap verification is as follows: The start and end times of each candidate abnormal time period are compared with the start and end times of all equipment failure time windows to determine if there is an overlap. If there is an overlap, it indicates that part or all of the duration of the candidate abnormal time period is caused by equipment failure, belonging to the abnormal time consumption corresponding to a non-efficiency event. If there is no overlap, it indicates that the candidate abnormal time period may be caused by the operator's own reasons, such as unauthorized absence from duty or rest, and does not belong to a non-efficiency event, so it needs to be eliminated. Subsequently, the overlap between all candidate abnormal time periods and equipment failure time windows is extracted, and the durations of these overlaps are summed to obtain the final abnormal duration. This duration only includes the work interruption time caused by equipment failure, accurately reflecting the impact of non-efficiency events on the actual time consumption. Finally, the final abnormal duration is deducted from the actual time consumption read in step 3, and the net actual time consumption is output. This net actual time consumption completely eliminates the interference of non-efficiency factors such as equipment failure, truly reflecting the actual sorting efficiency of the operators.

[0091] In a preferred embodiment of the present invention, step 4 is further included, which is the ratio of the standard time budget of a wave to the net actual time spent. The specific operation is as follows:

[0092] Based on the standard time budget for each wave obtained in step 2 and the net actual time consumed in step 3, the relative efficiency coefficient is calculated by the ratio of the two. This coefficient can eliminate the influence of differences in wave difficulty and non-efficiency factors, and achieve a unified assessment of the efficiency of workers under different waves and different difficulties. The formula for calculating the relative efficiency coefficient is as follows: ;

[0093] The derivation logic of this formula is as follows: the standard time budget for a wave is a reasonable time benchmark that fits the difficulty of the current wave; the net actual time is the actual working time of the operators after excluding non-efficiency factors; the ratio of the two can intuitively reflect the difference between the actual efficiency of the operators and the benchmark efficiency. A ratio greater than 1 indicates that the efficiency of the operators is higher than the benchmark level, and a ratio less than 1 indicates that the efficiency is lower than the benchmark level; in the formula, This is a relative efficiency coefficient, without units; The standard time budget for the wave output from step 2, in minutes; The net actual time output in step 3 is in minutes. To further improve the accuracy and stability of the coefficient, step 4 also optimizes the net actual time, smooths the coefficient, and performs penalty calibration through sub-steps to ensure that the final output relative efficiency coefficient can truly and objectively reflect the actual working ability of the operator.

[0094] Step 4 further includes the following steps:

[0095] Step 41: Divide the net actual time into time segments according to the natural pause points, calculate the ratio of the number of items picked to the segment duration in each segment as the instantaneous efficiency, and remove segments with a ratio lower than half or higher than twice the overall instantaneous efficiency average. Accumulate the remaining segment durations to output a stable net time. The specific operation is as follows:

[0096] By identifying natural pauses in the net actual time, the net actual time is divided into multiple continuous time segments. Then, by filtering out abnormal efficiency segments, the remaining effective segment durations are finally added together to obtain a stable net time.

[0097] The method for determining natural pause points is as follows: By monitoring the movement trajectory and action status of operators in real time, when it is detected that an operator has no movement for 10 to 15 consecutive seconds and has no picking, placing, or other sorting operations, this moment is determined to be a natural pause point. This judgment standard is calibrated through historical normal sorting operation data to ensure accurate differentiation between normal operation intervals and abnormal pauses. The net actual time is divided into multiple time segments according to the natural pause points. Each time segment corresponds to a continuous normal sorting operation process. Then, the ratio of the number of picked items in each time segment to the duration of that segment is calculated. This ratio is the natural pause point for that time segment. The instantaneous efficiency of time segments reflects the real-time working status of the operator during that time period. The overall average instantaneous efficiency of all time segments is calculated. Segments with instantaneous efficiency below half or above twice the average are identified as abnormal segments. These segments usually correspond to situations such as brief lapses in concentration, operational errors, or sudden brief abnormalities by the operator, and cannot reflect stable working efficiency, so they need to be eliminated. Finally, the durations of the remaining non-abnormal time segments are summed up, and the summed duration is the stable net time consumption. This duration can accurately reflect the actual working time consumed by the operator under normal and stable conditions.

[0098] Step 42: Use the ratio of stable net time to the standard time budget for each wave as the original efficiency coefficient; obtain multiple original efficiency coefficients calculated using the same method for the last predetermined number of waves for the same operator under the same complexity fingerprint, sorted by completion time; take the median of the multiple original efficiency coefficients and the current original efficiency coefficient as a weighted average, with the weight of the current coefficient decreasing as the cumulative number of waves with the same fingerprint increases, and output a smoothed efficiency coefficient. The specific operation is as follows:

[0099] The ratio of the stable net time obtained in step 41 to the standard time budget of the wave obtained in step 2 is calculated to obtain the original efficiency coefficient of the current wave. The original efficiency coefficient directly reflects the efficiency level of the workers in the current wave, but it is easily affected by the fluctuation of the single operation status and has insufficient stability. Subsequently, the historical wave data of the same worker under the same complexity fingerprint is obtained. The same complexity fingerprint means that the difficulty of the historical wave is consistent with that of the current wave, and its efficiency data has reference value.

[0100] The specific method for obtaining the predetermined quantity is as follows: Based on the stability analysis of historical performance data, select 3-5 waves as the predetermined quantity. This quantity can ensure the reference value of historical data while avoiding excessive dilution of the current performance coefficient due to too much historical data. It can be flexibly selected within the range of 3 to 5 according to the actual sorting scenario. Sort by completion time, select the last predetermined quantity of fingerprint waves with the same complexity completed by the operator, calculate the original performance coefficient of each wave, and obtain multiple historical original performance coefficients. The median of these historical raw performance coefficients is taken and averaged with the current raw performance coefficient. The weight of the current coefficient decreases as the number of fingerprint waves of the same complexity completed by the operator increases. The specific weight setting logic is as follows: when the cumulative number of completions is n, the weight of the current coefficient is 1 / n, and the weight of the historical median is (n-1) / n. The more cumulative the number, the smaller the impact of the current single operation, and the more stable the smoothing effect. The result obtained by this weighted average is the smoothing performance coefficient. This coefficient can effectively eliminate the impact of fluctuations in a single operation and more objectively reflect the stable performance level of the operator under the difficulty wave.

[0101] Step 43: Calculate the penalty factor based on the number of occurrences of ineffective events and the total duration of anomalies. The penalty factor is equal to one minus the product of the number of occurrences and the total duration of anomalies, divided by the sum of the standard time budget for each wave and a predetermined constant. Multiply the smoothing efficiency coefficient by the penalty factor to output the final relative efficiency coefficient. The specific operation is as follows:

[0102] Based on the number of non-performance events and the total abnormal duration, a penalty factor is calculated. The smoothed performance coefficient is then multiplied by the penalty factor to achieve reasonable calibration of the performance coefficient. The number of non-performance events is counted in real-time by the system, specifically the number of abnormal durations caused by equipment failures confirmed in step 32; each equipment failure corresponds to one non-performance event. The total abnormal duration is the final abnormal duration output in step 32, reflecting the overall degree of interference of non-performance events on the operation. The formula for calculating the penalty factor is: ;

[0103] The derivation logic of this formula is as follows: the degree of impact of ineffective events is related to the number of occurrences k and the total duration of abnormalities. The product is directly proportional to the product; the larger the product, the more severe the interference from ineffective events, and the greater the punishment should be; wave standard time budget. Reflecting the overall workload of a wave, a predetermined constant C is used to avoid an abnormal penalty factor due to a denominator that is zero or too small. By using the sum of both as the denominator, the impact of wave load and ineffective events can be balanced, making the penalty factor more reasonable; in the formula, The penalty factor is dimensionless and ranges from 0 to 1; k is the number of non-effective events that occur and is dimensionless. Total duration of abnormality, in minutes; C is the standard time budget for a wave, in minutes; C is a predetermined constant, in minutes. It is obtained by selecting no less than 100 historical waves of different difficulties, calculating the standard time budget for each wave, and taking the minimum value of all budget values ​​as the predetermined constant. The value is usually in the range of 5 to 10 minutes to ensure that the denominator is always positive and reasonable.

[0104] The closer the calculated penalty factor is to 1, the smaller the impact of non-effective events and the smaller the penalty on the smoothing efficiency coefficient; conversely, the smaller the penalty factor, the more serious the interference of non-effective events and the greater the penalty on the smoothing efficiency coefficient. Finally, the smoothing efficiency coefficient obtained in step 42 is multiplied by the penalty factor to obtain the final relative efficiency coefficient. This coefficient reflects both the stable efficiency level of the operators and takes into account the impact of non-effective events, enabling a more comprehensive and objective assessment of the actual sorting efficiency of the operators.

[0105] In a preferred embodiment of the present invention, step 5 is further included: storing complexity fingerprints, relative efficiency coefficients, and operator identifiers. When the same complexity fingerprint accumulates to a preset number of times, the median of the relative efficiency coefficients for the corresponding wave is taken. If it deviates from the preset benchmark value, the difficulty increment value of the lookup table matrix is ​​adjusted by the reciprocal of the median. The specific operation is as follows:

[0106] First, key data from each sorting operation is stored. After accumulating a certain sample size, statistical analysis is used to obtain the group performance benchmark for that difficulty level. This benchmark is then compared with a preset standard benchmark value to determine the deviation. If the deviation exceeds a reasonable range, the corresponding difficulty increment value in the lookup table matrix is ​​adjusted. After adjustment, the corresponding sample data is cleared to prepare for the next round of calibration. The entire process forms a closed-loop calibration mechanism that can continuously adapt to changes in actual sorting operations, eliminate deviations between the initially set difficulty increment value and the actual operation scenario, and ensure that the complexity fingerprint can truly reflect the actual difficulty of each wave.

[0107] Step 5 further includes the following steps:

[0108] Step 51: Store the complexity fingerprint, the final relative efficiency coefficient, and the operator identifier into the difficulty memory. Establish an independent coefficient sequence based on the complexity fingerprint and arrange them in chronological order. The specific operations are as follows:

[0109] The difficulty memory bank is a pre-built database used to store sorting operation-related data for all waves. It has data reading, writing, classification, retrieval, and deletion functions, which can meet the needs of subsequent data statistical analysis. The specific data stored includes the complexity fingerprint obtained in step 1, the final relative efficiency coefficient output in step 43, and the operator identification of the sorting operation in that wave. These three types of data are indispensable. The complexity fingerprint is used to distinguish waves of different difficulties, the final relative efficiency coefficient is used to reflect the actual efficiency of the operators under that difficulty, and the operator identification is used for statistical analysis by grouping by personnel.

[0110] After storage, all data are grouped according to complexity fingerprints, and an independent coefficient sequence is established for each complexity fingerprint. Each sequence contains only all the final relative efficiency coefficients corresponding to that complexity fingerprint and the associated operator identifiers. At the same time, the coefficient sequences are arranged in the order of the completion time of the tasks to ensure that the data in the sequence can reflect the time change trend of efficiency under that difficulty wave, providing an orderly and standardized data foundation for subsequent sample size judgment and group efficiency benchmark calculation.

[0111] Step 52: When the number of records in the coefficient sequence corresponding to the fingerprint of the same complexity reaches the sample size threshold, extract all the final relative efficiency coefficients in the sequence, group them according to the operator identification, calculate the median of the coefficients in each group, and then take the median of the medians of each group as the group efficiency benchmark of the fingerprint. The specific operation is as follows:

[0112] Once the sample size of a fingerprint of a certain complexity reaches a preset standard, the population performance benchmark for that difficulty is obtained through group statistics and median calculation; the specific method for obtaining the sample size threshold is as follows:

[0113] To meet the reliability requirements of statistical analysis, a sample size threshold of 30 to 50 records is selected. This threshold is validated using historical data to ensure that the sample size reflects the true level of group effectiveness and avoids excessive bias in statistical results due to an insufficient sample size. The number of records can be flexibly selected within the range of 30 to 50 based on the actual number of sorting waves. When the number of records in the coefficient sequence corresponding to the same complexity fingerprint reaches the sample size threshold, it indicates that the sample size meets the statistical requirements. At this point, all final relative effectiveness coefficients in the sequence are extracted and grouped according to the operator's identification. That is, all final relative effectiveness coefficients of the same operator under the same complexity fingerprint are grouped together to ensure that each group of data can reflect the stable effectiveness level of an individual operator under that difficulty. Subsequently, the median of all final relative effectiveness coefficients in each group is calculated. The median can effectively avoid the influence of individual abnormal effectiveness data and ensure the representativeness of each group of data. Finally, the medians of all groups are summarized, and the median of these medians is calculated. This value is the group effectiveness benchmark corresponding to the complexity fingerprint, which can objectively reflect the overall effectiveness level of all operators under that difficulty wave.

[0114] Step 53: Divide the difference between the group effectiveness benchmark and the preset standard benchmark value by the standard benchmark value to obtain the deviation ratio. The specific operation is as follows:

[0115] The method for obtaining the preset standard baseline value is as follows:

[0116] Select at least 200 historical waves of varying difficulty, calculate the group performance benchmark for each wave, and take the mean of all group performance benchmarks as the preset standard benchmark value, typically set to 1. This benchmark value reflects the average performance level of workers across all difficulty waves and serves as a reference standard for judging the reasonableness of the group performance benchmark for a single complexity fingerprint. The formula for calculating the deviation ratio is as follows: ;

[0117] The derivation logic of this formula is as follows: the difference between the group effectiveness benchmark and the preset standard benchmark value reflects the absolute degree of deviation between the two. Dividing this difference by the preset standard benchmark value converts it into the relative degree of deviation, i.e., the deviation ratio, which can more accurately reflect the severity of the deviation and facilitate subsequent judgment on whether the difficulty increment value needs to be adjusted; in the formula, This is the deviation ratio, without units; The group effectiveness benchmark obtained in step 52 is unitless; The standard baseline value is a unitless value and is usually set to 1. When the deviation ratio is positive, it indicates that the group efficiency corresponding to the complexity fingerprint is higher than the average level, which means that the current difficulty increment value may be too low and the quantification of wave difficulty is insufficient. When the deviation ratio is negative, it indicates that the group efficiency is lower than the average level, which means that the current difficulty increment value may be too high and the quantification of wave difficulty is excessive.

[0118] Step 54: When the absolute value of the deviation ratio does not exceed the dead zone threshold, maintain the current difficulty increment value unchanged. When it exceeds the dead zone threshold, set the adjustment coefficient to the negative of the deviation ratio plus one, and limit the adjustment coefficient to a predetermined adjustment range. The specific operation is as follows:

[0119] By setting a dead zone threshold, it is determined whether the deviation is within a reasonable range. If it exceeds the range, an adjustment coefficient is calculated and a limiting effect is applied to ensure that the adjusted difficulty increment value meets the actual operational requirements. The specific method for obtaining the dead zone threshold is as follows:

[0120] Select at least 100 historical calibration data points of varying difficulty, statistically analyze the deviation ratio for each calibration, and calculate the standard deviation of the deviation ratio. Use 1.5 times the standard deviation as the dead zone threshold, typically ranging from 0.1 to 0.2. This threshold defines a reasonable range for deviation. When the absolute value of the deviation ratio does not exceed this threshold, it indicates that the deviation between the current difficulty increment and the actual operational difficulty is within an acceptable range, requiring no adjustment; maintain the current difficulty increment unchanged. When the absolute value of the deviation ratio exceeds the dead zone threshold, an adjustment factor needs to be calculated. The formula for calculating the adjustment factor is: ;

[0121] The derivation logic of this formula is as follows: the adjustment coefficient must be inversely related to the deviation ratio. When the deviation ratio is positive, the adjustment coefficient is less than 1, reducing the difficulty increment; when the deviation ratio is negative, the adjustment coefficient is greater than 1, increasing the difficulty increment. This inverse correction is achieved by subtracting the deviation ratio from 1, ensuring that the adjusted difficulty increment can reduce the deviation between the group effectiveness benchmark and the preset standard benchmark. In the formula, This is an adjustment factor, and has no unit. The deviation ratio obtained in step 53 is unitless. To avoid abnormal difficulty increment values ​​caused by excessively large or small adjustment coefficients, the adjustment coefficients need to be limited to a predetermined adjustment range. The predetermined adjustment range is obtained by combining historical adjustment data and setting it to 0.8 to 1.2. This range has been verified in practice to ensure that the adjusted difficulty increment value will not deviate too much from the actual difficulty, thus achieving deviation correction while ensuring the stability and rationality of the adjustment.

[0122] Step 55: Multiply the adjustment coefficient by the original difficulty increment value corresponding to the fingerprint of this complexity in the lookup table matrix, write the product back to the corresponding entry in the lookup table matrix, and clear all records in the coefficient sequence corresponding to this fingerprint. The specific operation is as follows:

[0123] The calibrated difficulty increment value is obtained by multiplying the adjustment coefficient by the original difficulty increment value, updating the lookup table matrix, and clearing the corresponding coefficient sequence to avoid interference from old sample data in subsequent calibration. The original difficulty increment value corresponding to the current complexity fingerprint is retrieved from the lookup table matrix. This original difficulty increment value is the initial or previous calibration value used in step 1 to calculate the complexity fingerprint. The original difficulty increment value is multiplied by the adjustment coefficient obtained in step 54, and the product is the calibrated difficulty increment value. This value can correct the deviation between the original difficulty increment value and the actual task difficulty, making the subsequent calculation of the complexity fingerprint more accurate. Subsequently, the calibrated difficulty increment value is written back to the entry in the lookup table matrix corresponding to the complexity fingerprint, overwriting the original difficulty increment value, thus updating the lookup table matrix and ensuring that subsequent new waves of complexity fingerprint calculations can use the calibrated difficulty increment value. Finally, all records in the coefficient sequence corresponding to the complexity fingerprint are cleared. This is because the sample data in the sequence has already been used for this calibration, and continuing to retain it would lead to duplication of subsequent sample size statistics, affecting the accuracy of the next round of calibration. After clearing, the operation data for this complexity fingerprint is accumulated again to prepare for the next calibration after reaching the sample size threshold, forming a continuous and closed-loop difficulty increment value calibration mechanism, ensuring that the entire performance evaluation system can continuously adapt to changes in actual operation scenarios.

[0124] In a preferred embodiment of the present invention, step 6 is further included: based on the calibrated memory bank and historical relative efficiency coefficients, the historical relative efficiency coefficient distribution is retrieved for the new wave, and high-difficulty waves are assigned to personnel with high historical relative efficiency coefficients, while low-difficulty waves are assigned to personnel with low historical relative efficiency coefficients. When a person's relative efficiency coefficient is lower than a predetermined efficiency threshold for a predetermined number of consecutive times under a specific difficulty fingerprint, a training instruction for grasping skills practice is output to that person. The specific operation is as follows:

[0125] Based on the difficulty memory bank calibrated in step 5, the complexity fingerprint and related historical performance data corresponding to the new wave are extracted. By analyzing the historical relative performance coefficient distribution of the operators, the operators are grouped by performance. Then, according to the difficulty type of the new wave, waves of different difficulties are assigned to personnel with corresponding performance levels. At the same time, for specific wave types with higher difficulty, the performance of the operators is continuously monitored. Training instructions are issued to personnel whose performance is consistently below standard. This achieves rationalization of wave allocation and dynamic improvement of personnel skills, ultimately improving the overall efficiency and stability of the entire sorting operation. It avoids the inefficiency caused by assigning high-difficulty waves to low-efficiency personnel and also avoids the waste of manpower caused by assigning low-difficulty waves to high-efficiency personnel.

[0126] Step 6 further includes the following steps:

[0127] Step 61: Extract the complexity fingerprint corresponding to the new wave from the calibrated difficulty memory, obtain the historical relative efficiency coefficient sequence of each operator under this fingerprint, assign decreasing weights to the sequence according to time sequence, and calculate the weighted average as the predicted efficiency value. The specific operation is as follows:

[0128] Efficiency data matching the new wave is extracted from the calibrated difficulty memory. A weighted average is used to calculate the predicted efficiency value, highlighting the reference value of recent efficiency data. The complexity fingerprint corresponding to the new wave is accurately extracted from the calibrated difficulty memory. This complexity fingerprint has passed the closed-loop calibration in step 5 and can accurately reflect the actual difficulty of the new wave. Based on this complexity fingerprint, the historical relative efficiency coefficient sequence for each worker under this fingerprint is retrieved. This sequence contains the final relative efficiency coefficients corresponding to all completion records for each worker under this difficulty wave, arranged in chronological order. Since recent efficiency data better reflects the current skill level and work status of workers, the historical relative efficiency coefficient sequence is assigned decreasing weights according to chronological order. The specific weighting method is as follows:

[0129] The four to six most recent records in the sequence are selected and assigned weights of 0.4, 0.3, 0.15, 0.1, and 0.05 respectively, ranked from most recent to oldest. If fewer than six records are selected, the weights are adjusted proportionally to ensure that the sum of all weights is 1. The more recent the record, the higher its weight; the more distant the record, the lower its weight, thus highlighting the impact of recent performance. The predicted performance value is calculated using a weighted average formula, which is: ;

[0130] The derivation logic of this formula is as follows: historical performance data from different times have varying reference value for current predictions, with recent data having higher reference value. By assigning decreasing weights, each historical performance coefficient is multiplied by its corresponding weight and then summed to obtain a predicted performance value that reflects the current expected performance of personnel, thus avoiding prediction bias caused by single historical or average data. In the formula, The predicted performance value of the workers is unitless; n is the number of historical relative performance coefficients selected, unitless. The weight of the i-th historical relative efficiency coefficient is dimensionless, and the sum of the weights is 1. , where is the i-th historical relative efficiency coefficient, without units; the predicted efficiency value calculated by this formula can accurately reflect the expected efficiency level of each worker under the new wave of difficulty.

[0131] Step 62: Sort all personnel's predicted performance values ​​under this fingerprint, take the top 1% as the high-performance group, the bottom 1% as the low-performance group, and the rest as the normal group, and calculate the standard deviation of the predicted performance values. When the standard deviation exceeds a predetermined fluctuation threshold, personnel in the normal group who are close to the high-performance boundary are assigned to the high-performance group, and personnel who are close to the low-performance boundary are assigned to the low-performance group. The specific operation is as follows:

[0132] The specific method for obtaining the first percentage is as follows: Combining the manpower allocation requirements of sorting operations and historical efficiency grouping data, select 15% to 20% as the first percentage. This percentage, through actual verification, can ensure that the number of personnel in the high-efficiency group and the low-efficiency group is reasonable. This ensures that there are enough high-efficiency personnel to handle high-difficulty waves, while avoiding efficiency bottlenecks caused by too many low-efficiency personnel. The percentage can be flexibly selected within the range of 15% to 20% according to the actual scenario. Sort all operators' predicted efficiency values ​​under this complexity fingerprint in descending order. Take the first percentage of personnel before the sorting and assign them to the high-efficiency group. Take the first percentage of personnel after the sorting and assign them to the low-efficiency group. The remaining personnel are assigned to the regular group, thus achieving the initial grouping of personnel.

[0133] Subsequently, the standard deviation of the predicted performance values ​​for all operators is calculated. The standard deviation reflects the dispersion of the predicted performance values; the greater the dispersion, the more significant the differences in personnel performance. The predetermined fluctuation threshold is obtained by selecting no fewer than 100 sets of predicted performance value data under fingerprints of different complexity, calculating the standard deviation of each set of data, and taking the mean of all standard deviations as the predetermined fluctuation threshold, typically ranging from 0.15 to 0.25. When the calculated standard deviation exceeds this predetermined fluctuation threshold, it indicates that the differences in personnel performance are significant, and the boundaries of the initial grouping are not reasonable enough. In this case, the personnel in the regular group need to be adjusted. Personnel in the regular group whose predicted performance values ​​are close to the boundary of the high-performance group are assigned to the high-performance group, and personnel whose predicted performance values ​​are close to the boundary of the low-performance group are assigned to the low-performance group. The boundary judgment criteria are: if the difference between the predicted performance value of a regular group member and the lowest value of the high-performance group is less than 0.05, the regular group member is assigned to the high-performance group; if the difference between the predicted performance value and the highest value of the low-performance group is less than 0.05, the regular group member is assigned to the low-performance group. This adjustment ensures that the grouping results can more accurately match the actual performance level of the personnel.

[0134] Step 63: Compare the difficulty increment value of the new wave with the high difficulty threshold and the low difficulty threshold to determine the wave difficulty type. Assign high difficulty waves to the high-efficiency group, low difficulty waves to the low-efficiency group, and normal difficulty waves to the normal group. The specific operation is as follows:

[0135] The high-difficulty and low-difficulty thresholds are obtained as follows: From the calibrated lookup table matrix, extract the difficulty increment values ​​corresponding to all complexity fingerprints. Statistically sort these difficulty increment values, taking the 75th percentile as the high-difficulty threshold and the 25th percentile as the low-difficulty threshold. This quantile selection ensures a reasonable number of high-difficulty and low-difficulty waves, conforming to the difficulty distribution pattern of actual sorting operations. Those skilled in the art can calculate these two thresholds using conventional statistical methods. The difficulty increment value of the new wave is compared with both the high-difficulty and low-difficulty thresholds. The difficulty increment value of the new wave comes from the difficulty increment value corresponding to the complexity fingerprint obtained in step 1. Then, the difficulty type of the new wave is determined: when the difficulty increment value is greater than or equal to the high-difficulty threshold, it is determined to be a high-difficulty wave; when the difficulty increment value is less than or equal to the low-difficulty threshold, it is determined to be a low-difficulty wave; when the difficulty increment value is between the two, it is determined to be a normal-difficulty wave.

[0136] Subsequently, tasks were assigned according to the correspondence between wave difficulty type and personnel performance group. High-difficulty waves were assigned to high-performance group personnel, who have a history of excellent performance and can efficiently complete high-difficulty sorting operations, avoiding excessive time consumption and increased error rates due to excessive difficulty. Low-difficulty waves were assigned to low-performance group personnel, who can gradually improve their skills in low-difficulty scenarios, while avoiding waste of manpower. Regular-difficulty waves were assigned to regular group personnel, achieving a reasonable allocation of human resources and ensuring that each wave is undertaken by suitable personnel, thereby improving the efficiency and stability of the overall sorting operation.

[0137] Step 64: Select wave types where the proportion coefficient of irregularly shaped parts exceeds a predetermined proportion threshold and the dispersion of the storage space exceeds a predetermined dispersion threshold; monitor the historical relative efficiency coefficient sequence of each person in this wave type; when a person completes this wave type consecutively a predetermined number of times, and the relative efficiency coefficient of each time is lower than the predetermined efficiency threshold, output a training instruction for grasping skills exercise. The specific operation is as follows:

[0138] The predetermined percentage threshold and predetermined dispersion threshold are obtained as follows: The predetermined percentage threshold is determined by statistically analyzing historical irregular part sorting operation data, selecting the 60th percentile of the irregular part percentage coefficient as the predetermined percentage threshold, typically ranging from 0.3 to 0.4; the predetermined dispersion threshold is determined by statistically analyzing historical storage space dispersion data, selecting the 70th percentile of the storage space dispersion as the predetermined dispersion threshold, ensuring that the selected wave types are high-difficulty types with a high proportion of irregular parts and dispersed storage locations; wave types where the irregular part percentage coefficient exceeds the predetermined percentage threshold and the storage space dispersion exceeds the predetermined dispersion threshold are selected. These wave types are difficult to sort and require high levels of skill and operational proficiency from the operators; subsequently, the historical relative efficiency coefficient sequence of each operator under this type of wave is continuously monitored, recording the number of times they complete this type of wave and the corresponding relative efficiency coefficient.

[0139] The predetermined number of consecutive operations is determined as follows: 3 to 4 operations are selected based on the skill improvement cycle of the personnel. This number ensures the accuracy of the judgment and avoids misjudgment caused by low efficiency in a single operation. The predetermined efficiency threshold is determined as follows: 0.8 to 0.9 of the preset standard benchmark value is taken as the qualified efficiency line for this type of wave. If it is lower than this threshold, it indicates that the personnel's efficiency has not reached the qualified level. When a worker completes this type of wave consecutively for the predetermined number of consecutive operations, and the relative efficiency coefficient of each operation is lower than the predetermined efficiency threshold, it indicates that the worker has a skill deficiency in sorting operations that require high difficulty and high grasping skills. At this time, training instructions for grasping skills are issued to the worker. The training instructions specifically include the grasping posture of irregular parts, protection methods, and precise grasping techniques to help them improve their grasping skills and thus improve their efficiency performance.

[0140] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for real-time evaluation of sorting worker efficiency based on order waves, characterized in that, include: Step 1: Extract the product type count, the cumulative value of the irregular part flag, and the spatial dispersion of the storage location for each wave of orders. Obtain the difficulty increment value by looking up the table matrix, which serves as the complexity fingerprint. Step 2: Add the difficulty increment value to the product of the total number of items in the wave and the basic time consumption per item to obtain the standard time budget for the wave. Step 3: Read the actual time consumed, record the abnormal duration of non-performance events, and subtract the abnormal duration from the actual time consumed to obtain the net actual time consumed; Step 4: Calculate the relative efficiency coefficient as the ratio of the standard time budget per wave to the net actual time spent; Step 5: Store the complexity fingerprint, relative efficiency coefficient and operator identifier. When the same complexity fingerprint accumulates to a preset number of times, take the median of the relative efficiency coefficient of the corresponding wave. If it deviates from the preset benchmark value, adjust the difficulty increment value of the lookup table matrix by the reciprocal of the median. Step 6: Based on the calibrated memory bank and historical relative efficiency coefficients, retrieve the distribution of historical relative efficiency coefficients for the new wave, assign high-difficulty waves to personnel with high historical relative efficiency coefficients, and low-difficulty waves to personnel with low historical relative efficiency coefficients. When a person's relative efficiency coefficient is lower than the predetermined efficiency threshold for a predetermined number of consecutive times under a specific difficulty fingerprint, output a training instruction for grasping skills practice to that person.

2. The method for real-time evaluation of sorting worker efficiency based on order wave as described in claim 1, characterized in that, Step 1 includes: Step 11: Decompose each order within the wave into a sequence of the smallest picking action units. This sequence includes a walking unit, a positioning unit, a gripping unit, and a placement unit. Assign an action time correction factor to each gripping unit based on the pre-stored irregular part attributes in the product code library. Step 12: Based on the action unit sequence and the time correction factor of the grasping unit, calculate the theoretical walking path length and theoretical grasping number, count the number of grasping units corresponding to the irregular parts, and output the standardized action base value including the walking base value, the grasping base value and the irregular part proportion coefficient. Step 13: Extract the walking base value and the grasping base value from the standardized action base values, combine them with the total number of items in the wave, construct a nonlinear penalty function for the spatial dispersion of the storage location, and output the spatial dispersion penalty value.

3. The method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 2, is characterized in that... Step 1 further includes: Step 14: Weight the spatial discrete penalty value and the grab base value in the standardized action base value, and introduce a dynamic correction coefficient to output the preliminary difficulty integral; Step 15: Map the initial difficulty integral to the time increment dimension, output the difficulty increment value through a piecewise linear transformation table, and use the difficulty increment value as the complexity fingerprint.

4. The method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 3, is characterized in that... Step 2 includes: Step 21: Obtain the historical waves with the same complexity fingerprint as the current wave, extract the difference between the actual time consumption and the difficulty increment value of each historical wave and divide it by the total number of items in that wave to obtain a sample group of basic time consumption per item, and take the median of the sample group as the basic time consumption per item. Step 22: Dynamically determine the weighting of the difficulty increment value based on the ratio of the proportion coefficient of irregular parts in the current wave to the dispersion of the storage space; when the ratio exceeds the predetermined threshold, the difficulty increment value is added to the product of the total number of parts in the wave and the basic time consumption per part with full weight; otherwise, the difficulty increment value is multiplied by the attenuation factor and then added; output the standard time budget for the wave.

5. The method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 4, is characterized in that... Step 3 includes: Step 31: Monitor the movement trajectory and dwell time. When the continuous dwell time exceeds the predefined threshold and the dwell location is not within the coordinate range of the picking point, mark the period corresponding to the continuous dwell as a candidate abnormal period. Step 32: Verify the overlap between the candidate abnormal time period and the equipment failure time window, take the intersection as the final abnormal duration, subtract the final abnormal duration from the actual time consumption, and output the net actual time consumption.

6. The method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 5, is characterized in that... Step 4 includes: Step 41: Divide the net actual time into time segments according to the natural pause point, calculate the ratio of the number of items picked in each segment to the segment duration as the instantaneous efficiency, remove segments with a ratio lower than half or higher than twice the average of the overall instantaneous efficiency, and sum the duration of the remaining segments to output a stable net time. Step 42: Use the ratio of stable net time consumption to the standard time budget of the wave as the original efficiency coefficient; obtain multiple original efficiency coefficients calculated in the same way for the last predetermined number of waves under the same complexity fingerprint and sorted by completion time for the same operator; take the median of multiple original efficiency coefficients and the current original efficiency coefficient as a weighted average, and the weight of the current coefficient decreases as the cumulative number of waves with the same fingerprint increases, and output a smooth efficiency coefficient; Step 43: Calculate the penalty factor based on the number of occurrences of ineffective events and the total duration of anomalies. The penalty factor is equal to one minus the product of the number of occurrences and the total duration of anomalies divided by the sum of the standard time budget for the wave and the predetermined constant. Multiply the smoothed efficiency coefficient by the penalty factor to output the final relative efficiency coefficient.

7. A method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 6, is characterized in that... Step 5 includes: Step 51: Store the complexity fingerprint, the final relative efficiency coefficient, and the operator identifier into the difficulty memory bank, and establish an independent coefficient sequence according to the complexity fingerprint and arrange them in chronological order; Step 52: When the number of records in the coefficient sequence corresponding to the fingerprint of the same complexity reaches the sample capacity threshold, extract all the final relative efficiency coefficients in the sequence, group them according to the operator identification, calculate the median of the coefficients in each group, and then take the median of the medians of each group as the group efficiency benchmark of the fingerprint. Step 53: Divide the difference between the group performance benchmark and the preset standard benchmark value by the standard benchmark value to obtain the deviation ratio.

8. A method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 7, is characterized in that... Step 5 further includes: Step 54: When the absolute value of the deviation ratio does not exceed the dead zone threshold, the current difficulty increment value remains unchanged; when it exceeds the dead zone threshold, the adjustment coefficient is set to the negative of the deviation ratio plus one, and the adjustment coefficient is limited to the predetermined adjustment range. Step 55: Multiply the adjustment coefficient by the original difficulty increment value corresponding to the fingerprint of that complexity in the lookup table matrix, write the product result back to the corresponding entry in the lookup table matrix, and clear all records in the coefficient sequence corresponding to the fingerprint.

9. A method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 8, is characterized in that... Step 6 includes: Step 61: Extract the complexity fingerprint corresponding to the new wave from the calibrated difficulty memory, obtain the historical relative efficiency coefficient sequence of each operator under the fingerprint, assign decreasing weights to the sequence according to time sequence, and calculate the weighted average as the predicted efficiency value. Step 62: Sort the predicted performance values ​​of all personnel under the fingerprint, take the first percent as the high-performance group, the last percent as the low-performance group, and the rest as the normal group, and calculate the standard deviation of the predicted performance values. When the standard deviation exceeds the predetermined fluctuation threshold, the personnel in the normal group who are close to the high-performance boundary are classified into the high-performance group, and the personnel who are close to the low-performance boundary are classified into the low-performance group.

10. A method for real-time evaluation of sorting worker efficiency based on order waves, as described in claim 9, is characterized in that... Step 6 also includes: Step 63: Compare the difficulty increment value of the new wave with the high difficulty threshold and the low difficulty threshold to determine the wave difficulty type. Assign high difficulty waves to the high-efficiency group, low difficulty waves to the low-efficiency group, and normal difficulty waves to the normal group. Step 64: Select wave types where the proportion coefficient of irregular parts exceeds the predetermined proportion threshold and the dispersion of the storage space exceeds the predetermined dispersion threshold; monitor the historical relative efficiency coefficient sequence of each person under this wave type; when a person completes this wave type consecutively for a predetermined number of consecutive times and the relative efficiency coefficient of each time is lower than the predetermined efficiency threshold, output the training instruction for grasping skills exercise.

Images