Automatic express classification system based on RFID technology

Abstract

The invention relates to the technical field of image recognition, in particular to an automatic express classification system based on the RFID technology, and the system comprises a tag and image reading module, an error verification and signal decoding module, an execution adjustment and target orientation module, and a performance evaluation and adjustment module. According to the invention, through information reading and error checking mechanisms of the RFID and the camera, the accuracy and efficiency of the express automatic classification system are remarkably improved, after the tag information is read preliminarily, the consistency of the image and the RFID data is checked, classification errors caused by tag damage or information errors are effectively reduced, and when the classification errors are found, the accuracy and efficiency of the express automatic classification system are improved. The system can automatically trigger the re-classification operation, ensures that each parcel is correctly classified, reduces the requirement of manual re-classification, improves the automation level of operation, and can adapt to different operation environments by continuously monitoring the processing speed and the correct rate and optimizing the operation parameters according to the evaluation result; and the stability and reliability of the whole system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and particularly to an express automatic classification system based on RFID technology. Background Art

[0002] The technical field of image recognition focuses on enabling computers and systems to recognize and process information in images. This technology involves the development of algorithms and models that can identify patterns, features, and objects from images, including applications such as object recognition, facial recognition, scene reconstruction, and action and event detection. The implementation of image recognition relies on machine learning technologies, especially deep learning, which trains neural networks to process and analyze large amounts of image data to identify and classify the content of images. Image recognition technology is widely used in fields such as express automatic classification, security monitoring, autonomous driving vehicles, medical imaging analysis, agricultural monitoring, and industrial automation.

[0003] Among them, an express automatic classification system is a system that uses image recognition technology to automatically process and classify express parcels. The main purpose of this system is to automatically classify and sort express deliveries by identifying label information on the parcels, such as barcodes, destination addresses, and sender information, thereby improving the efficiency and accuracy of logistics processing. Such a system is particularly useful in logistics centers and warehouses, which can significantly reduce the need for manual sorting, speed up parcel processing, and at the same time reduce error rates and operating costs.

[0004] Although existing express automatic classification systems have achieved a high degree of automation, there are still limitations in recognition accuracy and processing efficiency. Especially when dealing with damaged or partially occluded labels, the error rate of existing systems is relatively high, which directly affects the accuracy of classification and logistics efficiency. When existing technologies encounter classification errors, they mostly rely on manual intervention for error correction, which not only increases operating costs but also affects processing speed. Due to the lack of an effective self-correction mechanism, the technology limits the adaptability of the system in complex environments, resulting in challenges to the stability and reliability of the system during peak periods or when processing a large number of irregular parcels. Summary of the Invention

[0005] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose an express automatic classification system based on RFID technology.

[0006] To achieve the above purpose, the present invention adopts the following technical solution: An express automatic classification system based on RFID technology includes:

[0007] The label and image reading module reads the express package information based on the RFID reader and camera input, reads the label information and captures feature points, generates calibrated information, and uses the calibrated information to perform consistency verification of label and image information to obtain the express information reading verification result.

[0008] Based on the express information reading verification result, the error verification and signal decoding module performs signal decoding and target detection, compares the consistency of RFID information and image recognition information, obtains the misclassification analysis result, and determines whether to trigger a reclassification operation based on the misclassification analysis result to obtain the express package classification verification result.

[0009] The execution adjustment and target orientation module automatically adjusts the conveyor belt direction according to the express package classification verification result, corrects the misclassified express packages, obtains the adjustment operation instruction, and updates the classification status after executing the adjustment operation instruction to obtain the execution status update result.

[0010] The performance evaluation and adjustment module uses the execution status update result to evaluate the overall performance of express automatic classification, monitors the processing speed and classification accuracy rate, obtains the express classification performance evaluation result, and optimizes the operation parameters according to the express classification performance evaluation result to obtain the express classification processing flow adjustment plan.

[0011] As a further solution of the present invention, the steps for obtaining the express information reading verification result are specifically as follows:

[0012] Read the express package information based on the RFID reader and camera input, process the captured image through image enhancement technology, identify and extract the label text, and use the formula:

[0013]

[0014] Calculate the three-dimensional distance between the label edges to generate accurate label information, where S a represents the label clarity index, and x a1 , x a2 , y a1 , y a2 and z a1 respectively represent the three-dimensional coordinate points of the label edges on the image;

[0015] Using the accurate label information, use a deep learning model to identify and analyze the feature points of the image, and use the formula:

[0016]

[0017] Synchronize the feature points and image information to obtain the calibrated feature point information, where P a represents the calibrated feature point information, and d airepresents the distance from the i-th feature point to the center, v ai is the visual clarity score of the feature point, w ai is the weight of the i-th feature point, and n is the total number of feature points;

[0018] Perform consistency verification using the calibrated feature point information, and adopt the formula:

[0019]

[0020] Calculate the standardized error between the calibrated feature points and the preset threshold. If the error is lower than the set standard, determine the information consistency and obtain the express information reading verification result. Among them, V a is the validity index of the verification result, P a is the calibrated feature point information, T a is the set feature point threshold, S a is the label clarity index.

[0021] As a further solution of the present invention, the specific steps for obtaining the misclassification analysis result are as follows:

[0022] Perform signal decoding from the express information reading verification result, extract the sequence of data points from the RFID, and adopt the formula:

[0023]

[0024] Obtain the total sum of the decoded RFID information. Among them, R b represents the adjusted total sum of the RFID information, r bi is the reading value of a single RFID data point, α b is the data reliability adjustment coefficient, and n is the total number of read data points;

[0025] Based on the total sum of the decoded RFID information, use the image processing algorithm to identify the image information, and weighted integrate the image recognition value of each data point with the confidence level, and adopt the formula:

[0026]

[0027] Obtain the weighted total sum of the image information. Among them, I b is the adjusted total sum of the image information, g bj is the recognition value of the image data point, p bj is the confidence weight of the data point, β b is the adjustment factor, and m is the total number of image data points;

[0028] Compare the results of the weighted total sum of the image information with the total sum of the decoded RFID information, and adopt the formula:

[0029]

[0030] Evaluate the consistency between RFID information and image recognition information. If C b is lower than the preset threshold, it is recorded as misclassification, and a misclassification analysis result is obtained. Among them, C b is the consistency index, R b is the sum of the adjusted RFID information, I b is the sum of the adjusted image information, and γ b is a non-zero coefficient.

[0031] As a further solution of the present invention, the steps for obtaining the classification verification result of the express package are specifically as follows:

[0032] Collect the misclassification probability of each package from the misclassification analysis result, calculate the total misclassification analysis index, and use the formula:

[0033]

[0034] Generate the total misclassification analysis index, where E c represents the total misclassification analysis index, e ci is the misclassification probability of each package, c ci is the correction coefficient, and n is the total number of packages;

[0035] Apply the total misclassification analysis index to the decision logic to determine whether a reclassification operation is required, through the formula:

[0036]

[0037] Generate the reclassification decision result, where D c is the reclassification decision result, λ c is the trade-off coefficient applied to the total misclassification analysis index, Λ c is the threshold of the misclassification index, and E c represents the total misclassification analysis index;

[0038] Based on the reclassification decision index, judge whether to trigger a reclassification operation, through the formula:

[0039] V c = if D c > θ c ‘then’ and ‘else’

[0040] Obtain the classification verification result of the express package, where V c is the classification verification result of the express package, D c is the reclassification decision index, and θ c is used to judge whether reclassification is required. Then represents reclassification, and else represents classification determination.

[0041] As a further solution of the present invention, the step of obtaining the adjustment operation instruction is specifically as follows:

[0042] Analyze the classification verification result of the express package, identify the misclassified packages, and apply the formula:

[0043]

[0044] Generate the analysis result of the misclassified package priority, where A d represents the analysis result of the misclassified package priority, x di is the misclassification status indicator, w di is the processing priority weight, and n is the total number of packages;

[0045] Based on the analysis result of the misclassified package priority, calculate the adjustment strategy of the conveyor belt, and use the formula:

[0046]

[0047] Generate the conveyor belt adjustment decision, where D d is the conveyor belt adjustment decision result, x di is the misclassification status indicator, w di is the processing priority weight, s di is the adjustment direction weight, and n is the total number of packages;

[0048] According to the conveyor belt adjustment decision, perform the inspection and preparation work before adjustment, and use the formula:

[0049] C d = if D d > θ d then 'ZTZ' else 'YZT'

[0050] Obtain the adjustment operation instruction, where C d is the adjustment operation instruction, D d is the conveyor belt adjustment decision result, θ d is the adjustment decision threshold, ZTZ represents right adjustment, and YZT represents left adjustment.

[0051] As a further solution of the present invention, the step of obtaining the execution status update result is specifically as follows:

[0052] Based on the adjustment operation instruction, obtain the current classification status and operation instruction from the database, determine the adjustment coefficient according to the operation type, and calculate the initial adjustment status, using the formula:

[0053] S h = s hi + α hi × p hi

[0054] Generate initialization state parameters, where S h represents the initialization state parameter, s hi is the current classification state, α hi is the adjustment coefficient, p hi is the priority weight of the operation instruction;

[0055] Apply the initialization state parameters, perform adjustment operations, calculate the new state according to the state type and operation rules, using the formula:

[0056] U h = S h ×(1 + β hi ×d hi )

[0057] Generate the updated classification state, where U h is the updated classification state, S h is the initialization state parameter, β hi is the adjustment multiplier, d hi is the adjustment depth;

[0058] Record the updated classification state, calculate the state update value, using the formula:

[0059] R h = U h + γ hi ×k hi

[0060] Generate the execution state update result, where R h is the execution state update result, U h is the updated classification state, γ hi is the safety factor, k hi is the protection factor.

[0061] As a further solution of the present invention, the steps for obtaining the express classification performance evaluation result are specifically as follows:

[0062] Based on the execution state update result, extract the performance indicators associated with the execution state update from the data warehouse, including processing speed and accuracy rate, assign weights according to the criticality of the indicators, and calculate the weighted performance indicators, using the formula:

[0063]

[0064] Generate the weighted performance data set, where P f represents the weighted performance data set, v fi , c fi and w fiThe processing speed, accuracy rate, and weight for each sample respectively;

[0065] According to the weighted performance dataset, calculate the overall performance index for express delivery classification using the formula:

[0066]

[0067] Obtain the performance evaluation index, where E f is the performance evaluation index, P fj is the weighted performance data of the j-th package;

[0068] According to the performance evaluation index, determine whether the performance of express delivery classification meets the preset standard using the linear conversion formula:

[0069] R f = a fi ×E f + b fi

[0070] Adjust the range and ratio of the result to generate the express delivery classification performance evaluation result, where R f is the express delivery classification performance evaluation result, E f is the performance evaluation index, a fi and b fi are the result adjustment parameters.

[0071] As a further solution of the present invention, the obtaining steps of the express delivery classification processing flow adjustment scheme are specifically as follows:

[0072] Based on the express delivery classification performance evaluation result, analyze the current express delivery classification performance data using the formula:

[0073]

[0074] Generate the performance baseline data, where B g represents the performance baseline, s gk represents the processing speed of the k-th measurement, r gk is the accuracy rate, e gk is the environmental impact coefficient, and n is the number of samples;

[0075] Based on the performance baseline data, compare it with the industry standard and the expected goal using the formula:

[0076] D g = |T g - B g |·p g

[0077] Calculate the performance gap and determine the key direction that needs to be adjusted to generate the performance adjustment target, where D gLet T be the performance gap indicator g Let B be the target performance indicator g Let p be the current performance baseline g Let α be the priority factor;

[0078] According to the performance adjustment target, design an adjustment strategy using the formula:

[0079]

[0080] Balance the effect of adjustment and the difficulty of implementation, predict the performance improvement after adjustment, and generate an adjustment plan for the express classification and processing process. Among them, g Let D be the result of the adjustment plan g Let f be the performance gap indicator g Let h be the adjustment sensitivity coefficient g Let β be the historical resistance factor

[0081] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0082] In the present invention, by introducing an information reading and error checking mechanism based on RFID and cameras, the accuracy and efficiency of the express automatic classification system are significantly improved. The system is allowed to check the consistency between the image and RFID data after initially reading the label information, effectively reducing classification errors caused by damaged labels or incorrect information. When a classification error is found, the system can automatically trigger a reclassification operation to ensure that each package is correctly classified, which reduces the need for manual reclassification and improves the automation level of the operation. Performance evaluation and adjustment are carried out by continuously monitoring the processing speed and accuracy rate, and optimizing the operating parameters according to the evaluation results, enabling the system to adapt to different operating environments and improving the overall system stability and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 is the system flow chart of the present invention;

[0084] Figure 2 is the flow chart of the express information reading and verification result in the present invention;

[0085] Figure 3 is the flow chart of the misclassification analysis result in the present invention;

[0086] Figure 4 is the flow chart of the express package classification verification result in the present invention;

[0087] Figure 5 is the flow chart of the adjustment operation instruction in the present invention;

[0088] Figure 6 is the flow chart of the execution status update result in the present invention;

[0089] Figure 7 It is a flowchart of the express classification performance evaluation result in the present invention;

[0090] Figure 8 It is a flowchart of the express classification processing flow adjustment scheme in the present invention. Specific Embodiments

[0091] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0092] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more, unless otherwise specifically defined.

[0093] Please refer to Figure 1 , an express automatic classification system based on RFID technology includes:

[0094] The label and image reading module reads the express parcel information based on the input of the RFID reader and the camera, reads the label information and captures the feature points, generates the calibrated information, and uses the calibrated information to perform the consistency verification of the label and image information to obtain the express information reading verification result.

[0095] The error verification and signal decoding module performs signal decoding and target detection based on the express information reading verification result, compares the consistency of the RFID information and the image recognition information, obtains the misclassification analysis result, and determines whether to trigger the reclassification operation through the misclassification analysis result to obtain the express parcel classification verification result.

[0096] The execution adjustment and target orientation module automatically adjusts the conveyor belt direction according to the express parcel classification verification result, corrects the misclassified express parcels, obtains the adjustment operation instruction, and updates the classification status after executing the adjustment operation instruction to obtain the execution status update result.

[0097] The performance evaluation and adjustment module evaluates the overall performance of the automatic express classification by using the execution status update result, monitors the processing speed and classification accuracy rate, obtains the express classification performance evaluation result, and optimizes the operation parameters according to the express classification performance evaluation result to obtain the adjustment plan for the express classification processing flow.

[0098] The calibrated information includes signal strength, resolution, and recognition time. The express information reading and verification result includes error range, matching success rate, and conflict record. The misclassification analysis result includes error frequency, error category, and impact degree. The express parcel classification verification result includes reclassification, estimated error rate, and verification depth. The adjustment operation instruction includes adjustment parameters, target trajectory, and operation priority. The execution status update result includes status code, update frequency, and processing queue. The express classification performance evaluation result includes detection efficiency and response speed. The adjustment plan for the express classification processing flow includes adjustment frequency, strategy change, and implementation stage.

[0099] Please refer to Figure 2 , and the specific steps for obtaining the express information reading and verification result are as follows:

[0100] Read the express parcel information based on the input of the RFID reader and camera, process the captured image through image enhancement technology, identify and extract the label text, and use the formula:

[0101]

[0102] Calculate the three-dimensional distance between the label edges to generate accurate label information, where S a represents the label clarity index, and x a1 , x a2 , y a1 , y a2 and z a1 respectively represent the three-dimensional coordinate points of the label edges on the image;

[0103] Use the accurate label information to identify and analyze the feature points of the image by using a deep learning model, and use the formula:

[0104]

[0105] Synchronize the feature points with the image information to obtain the calibrated feature point information, where P a represents the calibrated feature point information, d ai represents the distance from the i-th feature point to the center, v ai is the visual clarity score of the feature point, w ai is the weight of the i-th feature point, and n is the total number of feature points;

[0106] Perform consistency verification using the calibrated feature point information, and use the formula:

[0107]

[0108] Calculate the standardized error between the calibrated feature points and the preset threshold. If the error is lower than the set standard, determine the information consistency and obtain the express information reading verification result, where V a is the validity index of the verification result, P a is the information of the calibrated feature points, T a is the set feature point threshold, and S a is the label clarity index.

[0109] Label clarity index formula:

[0110]

[0111] x a1 、x a2 、y a1 、y a2 : Represent the coordinate points of the label edge in the image. Assume the coordinates of the four corners of the label in the image are (x a1 ,y a1 ) = (100, 150) and (x a2 ,y a2 ) = (200, 250);

[0112] z a1 : Assume the coordinate of the vertical axis is z a1 = 50, which represents the depth of the label relative to the camera;

[0113] First, calculate the horizontal distance difference:

[0114] x a1 -x a2 = 100 - 200 = -100

[0115] y a1 -y a2 = 150 - 250 = -100

[0116] Then, calculate the label clarity index S a :

[0117]

[0118] The value S a = 150 represents the label clarity index calculated from the image. A higher value indicates good readability and recognition of the label.

[0119] Formula for the information of the calibrated feature points:

[0120]

[0121] d ai : The distance from the feature point to the center of the image. Suppose there are three feature points, and their distances are 10, 20, and 30 respectively;

[0122] v ai : The visual clarity score of each feature point, set to 0.8, 0.6, and 0.43;

[0123] w ai : The weight of each feature point, set to 1, 2, and 3;

[0124] Calculate the weighted average:

[0125]

[0126] The value P a = 23.87 represents the position of the weighted average feature point, showing the central tendency of the feature point position. A higher value reflects the tendency of the feature point to shift towards the image edge.

[0127] The formula for the validity index of the verification result:

[0128]

[0129] T a : The feature point threshold set by the system, set to 20;

[0130] S a : The label clarity index, and the previous calculation result is 150;

[0131] P a : The calibrated feature point information, which is 23.87;

[0132] Calculate V a :

[0133]

[0134] The value V a = 0.0258 represents the validity index of the verification result. A lower value indicates a smaller error between the feature point and the preset threshold, showing the high accuracy of the consistency verification.

[0135] Please refer to Figure 3 , and the specific steps to obtain the misclassification analysis result are as follows:

[0136] Execute signal decoding from the verification result read from the express information, extract the sequence of data points from the RFID, and use the formula:

[0137]

[0138] Obtain the total decoded RFID information, where R b represents the adjusted total RFID information, r bi is the reading value of a single RFID data point, α b is the data reliability adjustment coefficient, and n is the total number of data points read;

[0139] Based on the total decoded RFID information, use an image processing algorithm to identify the image information, and weighted integrate the image recognition value and confidence of each data point. Use the formula:

[0140]

[0141] Obtain the total weighted image information, where I b is the adjusted total image information, g bj is the recognition value of the image data point, p bj is the confidence weight of the data point, β b is the adjustment factor, and m is the total number of image data points;

[0142] Compare the results of the total weighted image information with the total decoded RFID information. Use the formula:

[0143]

[0144] Evaluate the consistency between the RFID information and the image recognition information. If C b is lower than the preset threshold, record it as misclassified to obtain the misclassification analysis result, where C b is the consistency index, R b is the adjusted total RFID information, I b is the adjusted total image information, γ b is a non-zero coefficient.

[0145] Formula for the adjusted total RFID information:

[0146]

[0147] r bi : RFID reading value of the i-th data point. Assume there are 5 data points with reading values of 10, 15, 20, 25, and 30 respectively;

[0148] α b : Data reliability adjustment coefficient, assumed to be 0.5;

[0149] n: Total number of data points, which is 5 here.

[0150] Calculate the total RFID information:

[0151]

[0152] The adjustment coefficient is multiplied by the total number of data points:

[0153] α b ·n = 0.5·5 = 2.5

[0154] The final total RFID information:

[0155] R b = 100 + 2.5 = 102.5

[0156] This value (R b = 102.5) represents the total adjusted RFID information, which includes the sum of the original data plus a small adjustment value based on the number of data points to improve the reliability assessment of the data.

[0157] Formula for the total adjusted image information:

[0158]

[0159] g bj : The recognition value of the j-th image data point. Assuming there are 3 data points with recognition values of 20, 25, and 30 respectively;

[0160] p bj : The confidence level of the data point. Assuming they are 0.8, 0.85, and 0.9 respectively;

[0161] β b : The adjustment factor. Assuming it is 1;

[0162] m: The total number of image data points. Here it is 3;

[0163] The weighted total image information:

[0164]

[0165] The adjustment factor is multiplied by the total number of data points:

[0166] β b ·m = 1·3 = 3

[0167] The final total image information:

[0168] I b = 64.25 + 3 = 67.25

[0169] This value (I b = 67.25) represents the total adjusted weighted image information, which includes the sum of the products of the original recognition values and confidence levels and the adjustment value based on the total number of data points.

[0170] Consistency index formula:

[0171]

[0172] R b : The sum of the adjusted RFID information, assumed to be 102.5;

[0173] I b : The sum of the adjusted image information, assumed to be 67.25;

[0174] γ b : A non - zero coefficient, assumed to be 0.1;

[0175] Calculate the absolute difference between the two results:

[0176] |R b - I b | = |102.5 - 67.25| = 35.25

[0177] Calculate the denominator:

[0178] max(R b , I b ) + γ b = max(102.5, 67.25) + 0.1 = 102.5 + 0.1 = 102.6

[0179] Calculate the consistency index:

[0180]

[0181] This value (C b = 0.6564) represents the consistency ratio between the RFID information and the image information. A ratio close to 0.66 indicates a relatively high consistency, but there are still certain differences, pointing to specific misclassification cases.

[0182] Please refer to Figure 4 , the specific steps for obtaining the classification verification result of the express package are as follows:

[0183] Collect the misclassification probability of each package from the misclassification analysis result, calculate the total misclassification analysis index, using the formula:

[0184]

[0185] Generate the total misclassification analysis index, where E c represents the total misclassification analysis index, e ci is the misclassification probability of each package, c ci is the correction coefficient, and n is the total number of packages;

[0186] Apply the total misclassification analysis index to the decision logic to determine whether a re - classification operation is required, through the formula:

[0187]

[0188] Generate a reclassification decision result, where D c is the reclassification decision result, and λ c is the trade-off coefficient applied to the total misclassification analysis index, and Λ c is the threshold of the misclassification index, and E c represents the total misclassification analysis index;

[0189] Based on the reclassification decision index, determine whether to trigger a reclassification operation through the formula:

[0190] V c = if D c > θ c ‘then’ and ‘else’

[0191] Obtain the classification verification result of the express package, where V c is the classification verification result of the express package, D c is the reclassification decision index, and θ c is used to determine whether reclassification is required. ‘then’ indicates reclassification, and ‘else’ indicates that the classification is determined.

[0192] The formula for the total misclassification analysis index:

[0193]

[0194] e ci : The misclassification probability of the i-th package. Assume there are 3 packages, and the misclassification probabilities are 0.1, 0.2, and 0.15 respectively;

[0195] c ci : The correction coefficient of each package, reflecting the package processing priority. Assume the coefficients are 1, 1.5, and 1.2 respectively;

[0196] n: The total number of packages, which is 3 here;

[0197] Calculate the total misclassification analysis index:

[0198] E c = (0.1×1) + (0.2×1.5) + (0.15×1.2)

[0199] E c = 0.1 + 0.3 + 0.18 = 0.58

[0200] This value E c = 0.58 represents the weighted sum of the misclassification probabilities of the packages considering the priority, and is used to evaluate the severity of the overall misclassification situation.

[0201] Reclassification decision result formula:

[0202]

[0203] λ c : Trade-off coefficient, assumed to be 0.8;

[0204] E c : Total misclassification analysis index, which is 0.58;

[0205] Λ c : Maximum allowable misclassification index, assumed to be 1.0;

[0206] Apply reclassification decision logic:

[0207]

[0208] This value D c = 0.464 is used to evaluate whether reclassification is needed. A lower D c value indicates that the current misclassification level is within an acceptable range.

[0209] Express parcel classification verification result formula:

[0210] V c = if D c > θ c ‘then’ and ‘else’

[0211] D c : The decision index obtained in the previous step is 0.464;

[0212] θ c : Decision threshold, assumed to be 0.5;

[0213] Determine the final classification verification result:

[0214] V c = if 0.464 > 0.5 ‘then’ and ‘else’

[0215] V c = ‘else’, for classification determination, the decision index D c does not exceed the threshold θ c , so the reclassification operation is not triggered and the current classification of the parcel is determined.

[0216] Please refer to Figure 5 , and the specific steps for obtaining the adjustment operation instruction are as follows:

[0217] Analyze the express parcel classification verification result, identify the misclassified parcels, and apply the formula:

[0218]

[0219] Generate the analysis result of the misclassified package priority, where A d represents the analysis result of the misclassified package priority, x di is the misclassification status indicator, w di is the processing priority weight, and n is the total number of packages;

[0220] Based on the analysis result of the misclassified package priority, calculate the adjustment strategy of the conveyor belt using the formula:

[0221]

[0222] Generate the conveyor belt adjustment decision, where D d is the conveyor belt adjustment decision result, x di is the misclassification status indicator, w di is the processing priority weight, s di is the adjustment direction weight, and n is the total number of packages;

[0223] According to the conveyor belt adjustment decision, perform the inspection and preparation work before adjustment through the formula:

[0224] C d = if D d > θ d then 'ZTZ' else 'YZT'

[0225] Obtain the adjustment operation instruction, where C d is the adjustment operation instruction, D d is the conveyor belt adjustment decision result, θ d is the adjustment decision threshold, ZTZ indicates right adjustment, and YZT indicates left adjustment.

[0226] Formula for the analysis result of the misclassified package priority:

[0227]

[0228] x di : represents the misclassification status of the i-th package (1 for misclassified, 0 for normal). Assume there are 5 packages, among which 3 are misclassified (1, 1, 1, 0, 0);

[0229] w di : the processing priority of the misclassified package. Assume the priorities are (5, 3, 4, 2, 1);

[0230] n: the total number of packages, which is 5 here;

[0231] Product calculation:

[0232]

[0233] Calculate the final result A d :

[0234] A d = 12

[0235] The obtained value A d = 12 represents the total priority weight sum of misclassified packages and is used to evaluate the urgency of the misclassification situation. Conveyor belt adjustment decision result formula:

[0236]

[0237] x di 、w di 、n: The same as the previous step;

[0238] s di : Adjustment direction weight, assumed to be (2, 1, -1, 0, 0);

[0239] Weighted sum calculation:

[0240]

[0241] Denominator total weight:

[0242]

[0243] Calculate the final result D d :

[0244]

[0245] The obtained value D d = 0.6 represents a comprehensive indicator of the direction and degree of conveyor belt adjustment and is used to decide the final adjustment direction of the conveyor belt.

[0246] Adjustment operation instruction formula:

[0247] C d = if D d > θ d then 'ZTZ' else 'YZT'

[0248] D d = 0.6: Obtained from the previous step;

[0249] θ d = 0.5: Assumed decision threshold;

[0250] Compare D d with θ d : 0.6 > 0.5, so the final result is 'ZTZ' (right adjustment);

[0251] Since D d is greater than θd , the final operation instruction is "right adjustment", indicating that the conveyor belt is adjusted to the right to correct the misclassified packages.

[0252] Please refer to Figure 6 , the specific steps for obtaining the execution status update result are as follows:

[0253] Based on the adjustment operation instruction, obtain the current classification status and operation instruction from the database, determine the adjustment coefficient according to the operation type, calculate the initial adjustment status, and use the formula:

[0254] S h = s hi + α hi × p hi

[0255] Generate the initialization status parameter, where S h represents the initialization status parameter, s hi is the current classification status, α hi is the adjustment coefficient, p hi is the priority weight of the operation instruction;

[0256] Apply the initialization status parameter, perform the adjustment operation, and calculate the new status according to the status type and operation rules, using the formula:

[0257] U h = S h × (1 + β hi × d hi )

[0258] Generate the updated classification status, where U h is the updated classification status, S h is the initialization status parameter, β hi is the adjustment multiplier, d hi is the adjustment depth;

[0259] Record the updated classification status, calculate the status update value, using the formula:

[0260] R h = U h + γ hi × k hi

[0261] Generate the execution status update result, where R h is the execution status update result, U h is the updated classification status, γ hi is the safety factor, k hi is the protection factor.

[0262] Initialization status parameter formula:

[0263] S h = s hi + α hi × p hi

[0264] s hi : The initial classification status of the current package, assumed to be 2 (indicating medium-priority classification);

[0265] α hi : The adjustment coefficient, determined according to the operation instruction, assumed to be 0.5;

[0266] p hi : The priority weight of the operation instruction, assumed to be 3 (high-priority operation);

[0267] Calculate the initialization status parameter:

[0268] S h = 2 + 0.5 × 3 = 2 + 1.5 = 3.5

[0269] The initialization status parameter S h = 3.5 indicates the new status after being affected by the adjustment coefficient and the priority weight, and is used for subsequent adjustment operations.

[0270] The updated classification status formula:

[0271] U h = S h × (1 + β hi × d hi )

[0272] S h : The initialization status parameter calculated in the previous step, which is 3.5;

[0273] β hi : The status-dependent adjustment multiplier, assumed to be 0.1;

[0274] d hi : The adjustment depth, parsed from the operation instruction, assumed to be 2;

[0275] Apply the adjustment operation:

[0276] U h = 3.5 × (1 + 0.1 × 2) = 3.5 × (1 + 0.2) = 3.5 × 1.2 = 4.2

[0277] The updated classification status U h = 4.2 indicates the new status after being affected by the adjustment coefficient and the adjustment depth, and is used to output the final status update result.

[0278] Execute the status update result formula:

[0279] R h = U h + γ hi × k hi

[0280] U h : The updated classification status obtained in the previous step is 4.2;

[0281] γ hi : Safety factor, assumed to be 0.05;

[0282] k hi : Protection factor, assumed to be 10;

[0283] Confirm and record the final status update value:

[0284] R h = 4.2 + 0.05 × 10 = 4.2 + 0.5 = 4.7

[0285] Execute the status update result R h = 4.7 represents the final status after considering the safety factor and protection factor, ensuring that the status update is within the safe range.

[0286] Please refer to Figure 7 , and the specific steps for obtaining the express classification performance evaluation results are as follows:

[0287] Based on the execution status update result, extract the performance metrics associated with the execution status update from the data warehouse, including processing speed and accuracy rate, assign weights according to the criticality of the metrics, calculate the weighted performance metrics, and use the formula:

[0288]

[0289] Generate a weighted performance data set, where P f represents the weighted performance data set, v fi , c fi and w fi are the processing speed, accuracy rate, and weight of each sample respectively;

[0290] According to the weighted performance data set, calculate the overall performance metric of the express classification, and use the formula:

[0291]

[0292] Obtain the performance evaluation metric, where E f is the performance evaluation metric, and P fj is the weighted performance data of the jth package;

[0293] According to the performance evaluation metric, judge whether the performance of the express classification meets the preset standard, and use the linear conversion formula:

[0294] R f = a fi × E f + b fi

[0295] Adjust the range and ratio of the results to generate the express classification performance evaluation results. Among them, R f is the express classification performance evaluation result, E f is the performance evaluation index, a fi and b fi are the result adjustment parameters.

[0296] Weighted performance dataset formula:

[0297]

[0298] v fi : Processing speed. Assume the values of 3 samples are 100, 120, and 110 units / hour;

[0299] c fi : Classification accuracy rate. Assume it is 0.98, 0.95, and 0.99.

[0300] w fi : Weight coefficient. Assume it is 1.0, 1.2, and 1.1;

[0301] n: Total number of samples. Here it is 3;

[0302] Calculate the weighted performance dataset:

[0303] P f = (100 × 0.98 × 1.0) + (120 × 0.95 × 1.2) + (110 × 0.99 × 1.1)

[0304] P f = 98 + 136.8 + 119.79 = 354.59

[0305] P f = 354.59 represents the total weighted performance considering processing speed, accuracy rate, and weight, and is used for subsequent comprehensive performance evaluation.

[0306] Performance evaluation index formula:

[0307]

[0308] Assume the value of P fj is 98, 136.8, and 119.79 obtained from the previous step;

[0309] m: Total number of packages. Here it is 3;

[0310] Calculate the comprehensive performance evaluation index:

[0311] E f =(98 2 +136.8 2 +119.79 2 ) 1 / 2

[0312] E f =(9604 + 18726.24 + 14350.5041) 1 / 2 =(42680.7441) 1 / 2 ≈206.59

[0313] E f = 206.59 represents the performance index after integrating all evaluation data points. A higher value indicates better system performance.

[0314] Formula for the express classification performance evaluation result:

[0315] R f = a fi ×E f + b fi

[0316] E f : The performance evaluation index calculated in the previous step, which is 206.59;

[0317] a fi : Adjustment parameter, assumed to be 0.5;

[0318] b fi : Bias parameter, assumed to be 50;

[0319] Output the final performance evaluation result:

[0320] R f = 0.5 × 206.59 + 50 = 103.295 + 50 = 153.295

[0321] R f = 153.295 represents the performance evaluation result after parameter adjustment, and this value reflects the performance of the system under the current configuration.

[0322] Please refer to Figure 8 , and the specific steps for obtaining the adjustment plan for the express classification processing flow are as follows:

[0323] Based on the express classification performance evaluation result, analyze the current express classification performance data and use the formula:

[0324]

[0325] Generate performance baseline data, where B g represents the performance baseline, s gk represents the processing speed of the k-th measurement, r gk is the correct rate, e gk is the environmental impact coefficient, and n is the number of samples;

[0326] Based on the performance baseline data, compare it with industry standards and expected goals, and use the formula:

[0327] D g = |T g - B g |·p g

[0328] Calculate the performance gap and determine the key direction for adjustment, generate the performance adjustment target, where D g is the performance gap index, T g is the target performance index, B g is the current performance baseline, p g is the priority factor;

[0329] According to the performance adjustment target, design an adjustment strategy and use the formula:

[0330]

[0331] Balance the effect of the adjustment and the difficulty of implementation, predict the performance improvement after adjustment, and generate an adjustment plan for the express classification and processing process, where A g is the result of the adjustment plan, D g is the performance gap index, f g is the adjustment sensitivity coefficient, h g is the historical resistance factor.

[0332] Performance baseline formula:

[0333]

[0334] s gk : The processing speed of the k-th measurement, with assumed values of 150, 160, 170 units / hour;

[0335] r gk : The corresponding correct rate, with assumed values of 0.90, 0.92, 0.88;

[0336] e gk : The environmental impact coefficient, with assumed values of 0.95, 0.97, 0.93;

[0337] n: The number of samples, which is 3 here;

[0338] Calculate the comprehensive performance contribution of each sample:

[0339] 150 × 0.90 × 0.95 = 128.25

[0340] 160 × 0.92 × 0.97 = 143.6672

[0341] 170 × 0.88 × 0.93 = 138.1848

[0342] Sum and average:

[0343]

[0344] B g = 136.7 represents the average performance baseline of the system under given environmental and operating conditions, which is a comprehensive performance indicator considering speed, accuracy rate, and environmental factors.

[0345] Performance gap metric formula:

[0346] D g = |T g - B g |·p g

[0347] T g : Target performance metric, assumed to be 150;

[0348] B g : The current performance baseline obtained from the previous step, which is 136.7;

[0349] p g : Priority factor, assumed to be 1.1;

[0350] Calculate the product of the performance gap and the adjustment factor:

[0351] D g = |150 - 136.7|·1.1 = 13.3 × 1.1 = 14.63

[0352] D g = 14.63 represents the performance gap adjusted according to the priority factor, indicating the required amount of performance improvement to achieve the target performance.

[0353] Adjustment plan result formula:

[0354]

[0355] D g : The performance gap calculated from the previous step, which is 14.63;

[0356] f g : Adjustment sensitivity coefficient, assumed to be 0.8;

[0357] h g : Historical adjustment resistance factor, assumed to be 0.2;

[0358] Calculation adjustment plan:

[0359]

[0360] A g = 9.753 represents the effect of the adjustment plan predicted based on the performance gap, adjustment sensitivity, and historical resistance factor.

[0361] Above are only the preferred embodiments of the present invention, and it is not limited to other forms of the present invention. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An automatic express sorting system based on RFID technology, characterized in that: The system comprises: The label and image reading module reads the express package information based on the RFID reader and camera input, reads the label information and captures the feature points, generates calibrated information, uses the calibrated information to verify the consistency of the label and image information, and obtains the express information reading verification result. The error check and signal decoding module reads the verification result based on the express information, performs signal decoding and target detection, compares the consistency of RFID information and image recognition information, obtains the error analysis result, determines whether it is necessary to trigger the reclassification operation through the error analysis result, and obtains the express package classification verification result. The execution adjustment and target orientation module automatically adjusts the direction of the conveyor belt according to the express parcel classification verification result, corrects the misclassified express parcels, obtains an adjustment operation instruction, updates the classification status after executing the adjustment operation instruction, and obtains an execution status update result. The performance evaluation and adjustment module uses the execution status update result to evaluate the overall performance of express automatic classification, monitor the processing speed and classification accuracy, obtain the express classification performance evaluation result, and optimize the operating parameters according to the express classification performance evaluation result to obtain the express classification processing flow adjustment plan.

2. The express automatic sorting system based on RFID technology according to claim 1 is characterized in that: The steps for obtaining the express information reading and verification results are specifically as follows: Based on the RFID reader and camera input, the express package information is read, the captured image is processed by image enhancement technology, and the label text is recognized and extracted using the formula: Calculate the three-dimensional distance between label edges and generate precision label information, where S a represents the label clarity index, x a1 、x a2 ,y a1 ,y a2 and z a1 Respectively represent the three-dimensional coordinate points of the label edge on the image; Using the precision label information, a deep learning model is used to identify and analyze feature points of the image using the formula: Synchronize the feature points and image information to obtain the calibrated feature point information, where P a Represents the feature point information after calibration, d ai Represents the distance from the i-th feature point to the center, v ai Score the visual clarity of feature points, w ai is the weight of the i-th feature point, and n is the total number of feature points; Use the calibrated feature point information to perform consistency check using the formula: Calculate the standardized error between the calibrated feature point and the preset threshold. If the error is lower than the set standard, the information consistency is determined and the express information reading verification result is obtained. a is the validity index of the verification result, P a is the feature point information after calibration, T a is the set feature point threshold, S a is the label clarity index.

3. The express delivery automatic sorting system based on RFID technology according to claim 2 is characterized in that: The steps for obtaining the error analysis result are specifically as follows: Signal decoding is performed from the express information reading verification result, and a sequence of data points is extracted from the RFID using the formula: Get the sum of decoded RFID information, where R b represents the total adjusted RFID information, r bi is the reading value of a single RFID data point, α b is the data reliability adjustment coefficient, n is the total number of data points read; Based on the sum of the decoded RFID information, the image information is identified using an image processing algorithm, and the image identification value of each data point is weighted and integrated with the confidence level, using the formula: Get the weighted sum of image information, where I b is the sum of the adjusted image information, g bj is the identification value of the image data point, p bj is the confidence weight of the data point, β b is the adjustment factor, m is the total number of image data points; The result of the decoded RFID information sum weighted by the image information sum is compared using the formula: Evaluate the consistency between RFID information and image recognition information. If C b If it is lower than the preset threshold, it is recorded as a misclassification and the misclassification analysis result is obtained, where C b is the consistency index, R b is the sum of adjusted RFID information, I b is the sum of the adjusted image information, γ b is a non-zero coefficient.

4. The express delivery automatic sorting system based on RFID technology according to claim 3 is characterized in that: The steps for obtaining the express parcel classification verification results are specifically as follows: The misclassification probability of each package is collected from the misclassification analysis results, and the total misclassification analysis index is calculated using the formula: Generate the total index of misclassification analysis, where E c represents the total index of misclassification analysis, e ci is the probability of misclassification of each package, c ci is the correction factor, n is the total number of packages; The total misclassification analysis index is applied to the decision logic to determine whether a reclassification operation is required, through the formula: Generate reclassification decision results, where D c is the reclassification decision result, λ c is the trade-off coefficient applied to the total index of misclassification analysis, Λ c is the threshold of the misclassification index, E c represents the total index of misclassification analysis; Based on the reclassification decision indicator, determine whether to trigger the reclassification operation, through the formula: V c =ifD c >θ c ‘then’and‘else’ Get the express parcel classification verification results, where V c is the result of express parcel classification verification, D c is the reclassification decision indicator, θ c It is used to determine whether reclassification is needed. Then means reclassification, and else means classification is confirmed.

5. The express delivery automatic sorting system based on RFID technology according to claim 4 is characterized in that: The steps of obtaining the adjustment operation instruction are specifically as follows: Analyze the express parcel classification verification results, identify the wrongly classified parcels, and apply the formula: Generate the priority analysis results of the misclassified packages, where A d Indicates the result of the priority analysis of the misclassified package, x di is the misclassification status indication, w di is the processing priority weight, n is the total number of packages; Based on the analysis results of the priority of the misclassified packages, the adjustment strategy of the conveyor belt is calculated using the formula: Generate conveyor belt adjustment decisions, where D d is the conveyor belt adjustment decision result, x di is the misclassification status indication, w di To process the priority weight, s di To adjust the direction weight, n is the total number of packages; According to the conveyor belt adjustment decision, perform the inspection and preparation work before adjustment, through the formula: C d =ifD d >θ d then‘ZTZ’else‘YZT’ Get the adjustment operation instruction, where C d To adjust the operating instructions, D d is the conveyor belt adjustment decision result, θ d It is the adjustment decision threshold, ZTZ means right adjustment, and YZT means left adjustment.

6. The express delivery automatic sorting system based on RFID technology according to claim 5 is characterized in that: The steps for obtaining the execution status update result are specifically as follows: Based on the adjustment operation instruction, the current classification state and operation instruction are obtained from the database, the adjustment coefficient is determined according to the operation type, and the initial adjustment state is calculated using the formula: S h =s hi +a hi ×p hi Generate initialization state parameters, where S h Represents the initialization state parameter, s hi is the current classification status, α hi is the adjustment coefficient, p hi is the priority weight of the operation instruction; Apply the initialization state parameters, perform the adjustment operation, and calculate the new state according to the state type and operation rules, using the formula: U h =S h ×(1+β hi ×d hi ) Generate an updated classification state, where U h is the updated classification status, S h is the initialization state parameter, β hi To adjust the multiplier, d hi To adjust the depth; The updated classification status is recorded and the status update value is calculated using the formula: R h =U h +g hi ×k hi Generate execution status update results, where R h To update the execution status, U h is the updated classification state, γ hi is the safety factor, k hi For protection factor.

7. The express delivery automatic sorting system based on RFID technology according to claim 6 is characterized in that: The steps for obtaining the express classification performance evaluation results are specifically as follows: Based on the execution status update result, the performance indicators associated with the execution status update are extracted from the data warehouse, including processing speed and accuracy, weights are assigned according to the criticality of the indicators, and weighted performance indicators are calculated using the formula: Generate a weighted performance dataset, where P f represents the weighted performance dataset, v fi 、c fi and w fi are the processing speed, accuracy and weight of each sample respectively; Based on the weighted performance data set, the overall performance index of the express delivery category is calculated using the formula: Get the performance evaluation index, where E f is the performance evaluation index, P fj is the weighted performance data of the jth package; According to the performance evaluation indicators, it is judged whether the performance of the express classification meets the preset standards, using the linear conversion formula: R f =a fi ×E f +b fi Adjust the range and proportion of the results to generate the express classification performance evaluation results, where R f is the express delivery classification performance evaluation result, E f is the performance evaluation index, a fi and b fi is the result adjustment parameter.

8. The express delivery automatic sorting system based on RFID technology according to claim 7 is characterized in that: The steps for obtaining the express classification processing flow adjustment plan are specifically as follows: Based on the express classification performance evaluation results, the current express classification performance data is analyzed using the formula: Generate performance baseline data, where B g represents the performance baseline, s gk represents the processing speed of the kth measurement, r gk is the accuracy rate, e gk is the environmental impact coefficient, n is the number of samples; Based on the performance baseline data, compared with industry standards and expected goals, the formula is used: D g =|T g -B g |·p g Calculate the performance gap, identify the key areas that need to be adjusted, and generate performance adjustment targets, where D g is the performance gap indicator, T g is the target performance indicator, B g is the current performance baseline, p g is the priority factor; According to the performance adjustment goals, design an adjustment strategy using the formula: Balance the effect of adjustment and the difficulty of implementation, predict the performance improvement after adjustment, and generate an adjustment plan for express classification processing flow, among which A g To adjust the results of the program, D g is the performance gap indicator, f g To adjust the sensitivity coefficient, h g is the historical resistance factor.

Images