A laboratory full-process quality control management system based on paperless office

Abstract

The present invention relates to the field of laboratory management technology, and specifically to a laboratory full-process quality control management system based on paperless office, comprising an anomaly matching unit, an anomaly warning unit, an experiment monitoring unit, an experiment review unit, and a performance evaluation unit. The system also comprises an identity recognition unit, wherein the identity recognition unit is configured to capture facial images of experimenters by arranging facial image acquisition equipment outside the laboratory to obtain facial images of the experimenters, and to extract features from the facial images using a trained feature extraction model to obtain a facial feature set. The present invention uses the anomaly matching unit and the anomaly warning unit to monitor and identify abnormal flow trajectories of experimental samples in real time. Once an anomaly is detected, abnormal information is immediately generated and an early warning is issued, which helps laboratory managers take timely measures to prevent potential problems from escalating during the experiment.

Description

Technical Field

[0001] The present invention relates to the technical field of laboratory management, and in particular to a laboratory full-process quality control management system based on paperless office. Background Art

[0002] Paperless office refers to the use of information technology to minimize or eliminate the dependence on paper in traditional office processes and achieve comprehensive digital management and operations.

[0003] Traditional systems usually rely on paper records and manual management, which can easily lead to data input errors, information loss or confusion, reducing the accuracy and completeness of the data. Traditional systems often lack real-time monitoring capabilities and are unable to immediately identify and warn of abnormal situations during the experiment. They need to rely on regular manual inspections or reports, which may result in problems not being discovered until they have already occurred. In addition, paper records and traditional management methods consume a lot of paper and office resources, which is not conducive to environmental protection and sustainable resource utilization. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems existing in the background technology and propose a laboratory full-process quality control management system based on paperless office.

[0005] The technical solution of the present invention is a laboratory full-process quality control management system based on paperless office, including an identity recognition unit, an experiment arrangement unit, a sample tracking unit and an abnormal matching unit, and also includes:

[0006] The identity recognition unit is used to collect a facial image of an experimenter by setting up a facial image collection device outside the laboratory to obtain a facial image of the experimenter, and perform feature extraction on the facial image using a trained feature extraction model to obtain a facial feature set, and calculate the similarity between the facial feature set and each standard feature set in a standard feature set sequence pre-stored in a database, and compare the similarity with a preset similarity threshold. If the similarity is greater than the preset similarity threshold, basic information of the experimenter is obtained and transmitted to the experiment arrangement unit. Otherwise, it indicates that the experimenter does not have experimental authority;

[0007] The experiment arrangement unit receives the basic information of the experimenter transmitted by the identity recognition unit, and constructs an experiment sample table based on the experiment type, and maps and associates the experiment sample table with the basic information of the experimenter to obtain an association mapping table, and matches the association mapping table with the basic information of the experimenter to obtain the experiment sample information required by the experimenter, and obtains the experiment type of the experimenter based on the experiment sample information required by the experimenter, and generates the experiment location of the experimenter based on the experiment type, and sends the experiment location to the experimenter, and transmits the experiment sample information required by the experimenter to the sample tracking unit;

[0008] The sample tracking unit receives the experimental sample information required by the experimenter transmitted by the experiment arrangement unit, tracks the real-time location of the experimental sample in the laboratory through RFID radio frequency identification technology, and records the flow trajectory node set of the experimental sample in the laboratory through a reader, and transmits the flow trajectory node set to the abnormal matching unit.

[0009] Preferably, it also includes an abnormal warning unit, the experimental monitoring unit, an experimental review unit and a performance evaluation unit; the abnormal matching unit receives the flow trajectory node set transmitted by the sample tracking unit, and matches the flow trajectory node set with the standard flow trajectory node set pre-stored in the database to obtain the matching degree between the two, and compares the matching degree with a preset matching degree threshold. If it is greater than the preset matching degree threshold, it indicates that the flow trajectory of the experimental sample is normal; otherwise, it indicates that the flow trajectory of the experimental sample is abnormal, and flow abnormality information is generated, the flow abnormality information including the abnormal flow trajectory node and the level of the flow trajectory abnormality of the experimental sample, and the flow abnormality information is transmitted to the abnormal warning unit.

[0010] Preferably, the experiment monitoring unit is used to capture the video of the experiment personnel's experimental operation by setting up a video monitoring device at the experimental position corresponding to the experiment type to obtain an experimental operation video, and generate an experiment full-process monitoring table based on the experiment type according to the experimental professional knowledge, and display the experiment full-process monitoring table on the smart terminal at the experimental position corresponding to the experiment type. The experiment personnel capture images of each process of the experiment through the experimental image acquisition device according to the experiment full-process monitoring table displayed on the smart terminal to obtain an experimental process image set, and transmit the experimental operation video to the experiment review unit, and transmit the experimental process image set to the performance evaluation unit.

[0011] Preferably, the performance assessment unit receives the experimental process image set transmitted by the experimental monitoring unit, and matches the experimental process image set with a preset standard experimental process image set to obtain the comprehensive score of the experimenter corresponding to the experimental process image set, and transmits the comprehensive score of the experimenter to the experimental review unit.

[0012] Preferably, the experiment review unit receives the experiment operation video transmitted by the experiment monitoring unit and the comprehensive score of the experimenter transmitted by the performance evaluation unit, and constructs a comprehensive score fitting curve based on the historical comprehensive score of the experimenter, and determines whether the error between the comprehensive score of the experimenter and the comprehensive score fitting curve is within a preset error threshold range. If it is within the preset error threshold range, it indicates that the comprehensive score of the experimenter is valid. Otherwise, the experiment administrator reviews the experiment operation video to determine whether the comprehensive score of the experimenter is valid.

[0013] Preferably, the abnormal warning unit receives the flow abnormality information transmitted by the abnormal matching unit, and matches the flow abnormality information with the decision support table pre-stored in the database to obtain the abnormal adjustment strategy corresponding to the flow abnormality information, and sends the abnormal adjustment strategy to the laboratory administrator, who makes corresponding adjustments to the experimenter.

[0014] Preferably, the real-time location of the experimental sample in the laboratory is tracked by RFID radio frequency identification technology, and the flow trajectory node set of the experimental sample in the laboratory is recorded by a reader, including the following steps:

[0015] A1. Multiple readers and multiple target tags are set in the laboratory, and the target tags are set on the experimental samples. An RSSI ranging model is constructed. The RSSI ranging model is as follows:

[0016]

[0017] Among them, P r (d) represents the signal strength value at the receiving end, RSSI0 represents the signal strength value at d0 meters, d represents the distance between the target tag and the reader, γ represents the path loss factor, d0 represents the reference distance, X σ represents shadow fading, and X σ is a random variable subject to N(0,σ);

[0018] A2. The plurality of readers receive signals emitted by the plurality of target tags and record the signal strengths of the signals to obtain a signal strength set, and convert the signal strength set into a distance from the target tag to each of the readers based on the RSSI ranging model to obtain a distance set;

[0019] A3. Construct an objective function based on the distance set. The objective function is as follows:

[0020]

[0021] Where f(x,y,z) represents the distance error between the estimated coordinates (x,y,z) and the actual coordinates to the i-th reader, d i represents the distance from the actual coordinate to the i-th reader, that is, the i-th distance in the distance set, D i represents the distance from the estimated coordinates (x, y, z) to the i-th reader, and (X i ,Y i ,Z i ) represents the coordinates of the i-th reader;

[0022] A4. Solve the objective function based on the improved grey wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample, and obtain the flow trajectory node set of the experimental sample based on the multiple positioning coordinates.

[0023] Preferably, solving the objective function based on the improved grey wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample includes the following steps:

[0024] B1. Generate and initialize a gray wolf population, and each gray wolf corresponds to the estimated coordinates in space;

[0025] B2. Calculate the fitness of the gray wolf individuals and save the top three gray wolf individuals with the largest fitness, i.e., the three estimated coordinates that are currently closest to the positioning coordinates;

[0026] B3. Calculate the coefficient factor of the gray wolf optimization algorithm based on the positions of the three gray wolf individuals. The coefficient factor calculation formula is as follows:

[0027]

[0028] Among them, A and B represent the coefficient factors of the gray wolf optimization algorithm, a represents the linear coefficient, and r1 and r2 represent random parameters in the interval [0,1].

[0029] B4. Update the positions of the three gray wolf individuals in the gray wolf population. The position update formula is as follows:

[0030]

[0031] in, represents the speed at the k+1th iteration, represents the position of the linear coefficient at the k+1th iteration, r1, r2 and r3 represent random parameters in the interval [0,1], x1, x2 and x3 represent the positions of the three gray wolf individuals, c1, c2 and c3 represent acceleration factors, and ω represents the inertia weight;

[0032] B5. Repeat steps B2-B4 until the number of iterations reaches a preset number to obtain the location information of the leader wolf, that is, the positioning coordinates of the experimental sample.

[0033] Preferably, matching the flow trajectory node set with a standard flow trajectory node set pre-stored in a database to obtain a matching degree between the two comprises the following steps:

[0034] C1. Calculate the Euclidean distance between each flow trajectory node in the flow trajectory node set and each standard flow trajectory node in the standard flow trajectory node set to obtain a Euclidean distance set, and use the standard flow trajectory node corresponding to the minimum Euclidean distance as the matching node of the flow trajectory node;

[0035] C2. constructing an undirected graph, using the flow trajectory nodes in the flow trajectory node set and the standard flow trajectory nodes in the standard flow trajectory node set as nodes of the undirected graph;

[0036] C3. Connecting the matched flow trajectory nodes and the standard flow trajectory nodes based on the matching nodes to obtain connection edges of an undirected graph, and using the Euclidean distance as the weight of the connection edges to obtain a weighted undirected graph;

[0037] C4. Summing the weights of the connecting edges in the weighted undirected graph to obtain a weight sum, and comparing the weight sum with a preset comprehensive threshold to obtain a matching degree between the flow trajectory node set and a standard flow trajectory node set pre-stored in a database.

[0038] Preferably, matching the experimental process image set with a preset standard experimental process image set to obtain a comprehensive score of the experimenter corresponding to the experimental process image set comprises the following steps:

[0039] D1. Performing feature extraction on the experimental process image set and the standard experimental process image set using a SIFT algorithm to obtain a first feature point set of the experimental process image set and a second feature point set of the standard experimental process image set, and matching the first feature point set and the second feature point set using a KNN algorithm to obtain a matching pair set;

[0040] D2, using the nearest neighbor algorithm to search for the α of each matching point in the matching pair set L and α R neighboring points, where L and R represent the experimental process image and the standard experimental process image respectively, and α L and α R Take 10;

[0041] D3. If the intersection of the neighboring points of each matching point in the experimental process image and the standard experimental process image is not empty, then the average distance from the neighboring intersection point to each matching point is calculated, and the neighboring point intersection is updated based on the average distance, and the intersection of the neighboring points of each matching point is taken as a union to obtain a union set. The neighboring point intersection update formula is as follows:

[0042]

[0043] where γ represents the ratio coefficient, and Represents the average distance from the i-th intersection point to each matching point in the experimental process image, Represents the average distance from the i-th intersection point to each matching point in the standard experimental process image;

[0044] D4. Gridding the image using the diagonal length of the experimental process image and the standard experimental process image as the side length of each grid. If the average distance of matching points in the grid is less than a preset average length threshold, then dividing the grid into 3×3 grids;

[0045] D5. Obtain a set of position change vectors based on the coordinates of the matching pairs in the union set, and calculate the cosine similarity of the position change vectors of the matching points in each grid. If the number of position change vectors of the matching points in a grid is less than the required minimum number of samples, the nine surrounding grids are included. If the number is still less than the required minimum number of samples, the matching pairs in the grid are filtered as incorrect matching pairs. Otherwise, recalculate the cosine similarity, and retain the matching pairs whose mean cosine similarity is greater than a preset mean threshold as inliers, thereby obtaining a filtered set of matching pairs.

[0046] D6. The number of matching points in the filtered matching pair set is used as the comprehensive score of the experimenter.

[0047] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0048] 1. The present invention uses an abnormal matching unit and an abnormal warning unit to monitor and identify abnormal flow trajectories of experimental samples in real time. Once an abnormality is detected, abnormal information is immediately generated and an early warning is issued, which helps laboratory managers take timely measures to prevent potential problems from expanding during the experiment. Through the experiment monitoring unit and the performance evaluation unit, the system comprehensively monitors and evaluates every aspect of the experimental operation. The experimental personnel's operating behavior and the experimental process image set are recorded and scored to ensure the quality of the experiment and the consistency of the operation.

[0049] 2. The present invention utilizes an identity recognition unit and facial recognition technology to ensure that only authorized experimenters can enter the laboratory and perform corresponding experimental operations, thereby improving the safety and management efficiency of the laboratory and realizing digital management of the entire process, reducing paper documents and manual records in traditional laboratory management, improving work efficiency and data accuracy, and conducting a comprehensive evaluation based on historical comprehensive scores through the experimental review unit. By analyzing the historical performance data of the experimenters, it is possible to identify and improve individual experimental skills and performance, thereby improving the overall experimental quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of the overall system in one embodiment of the present invention.

[0051] Figure numerals: 1. Identity recognition unit; 2. Experiment arrangement unit; 3. Sample tracking unit; 4. Abnormal matching unit; 5. Abnormal warning unit; 6. Experiment monitoring unit; 7. Experiment review unit; 8. Performance assessment unit. DETAILED DESCRIPTION

[0052] Example 1, as Figure 1 As shown, the present invention proposes a laboratory full-process quality control management system based on paperless office, including an identity recognition unit 1, an experiment arrangement unit 2, a sample tracking unit 3, an abnormality matching unit 4, an abnormality warning unit 5, an experiment monitoring unit 6, an experiment review unit 7 and a performance evaluation unit 8, and also includes:

[0053] The identity recognition unit 1 is used to collect facial images of the experimenter by setting up a facial image collection device outside the laboratory to obtain the experimenter's facial image, and extract features from the facial image using a trained feature extraction model to obtain a facial feature set, and calculate the similarity between the facial feature set and each standard feature set in a standard feature set sequence pre-stored in the database, and compare the similarity with a preset similarity threshold. If the similarity is greater than the preset similarity threshold, the basic information of the experimenter is obtained and transmitted to the experiment arrangement unit 2. Otherwise, it indicates that the experimenter does not have experimental authority;

[0054] The experiment arrangement unit 2 receives the basic information of the experimenter transmitted by the identity recognition unit 1, and constructs an experiment sample table based on the experiment type, and maps and associates the experiment sample table with the basic information of the experimenter to obtain an association mapping table, and matches the association mapping table with the basic information of the experimenter to obtain the experiment sample information required by the experimenter, and obtains the experiment type of the experimenter based on the experiment sample information required by the experimenter, and generates the experiment location of the experimenter based on the experiment type, and sends the experiment location to the experimenter, and transmits the experiment sample information required by the experimenter to the sample tracking unit 3;

[0055] The sample tracking unit 3 receives the experimental sample information required by the experimenter transmitted by the experiment arrangement unit 2, and tracks the real-time location of the experimental sample in the laboratory through RFID radio frequency identification technology, and records the flow trajectory node set of the experimental sample in the laboratory through the reader, and transmits the flow trajectory node set to the abnormal matching unit 4.

[0056] In the present invention, the feature extraction model generally refers to a machine learning model or a deep learning model, which is used to extract representative features from facial images. These features can be geometric features of the face (such as the position and size of the eyes, nose, and mouth), surface features (such as skin texture), color features (such as skin color), texture features (such as the density and shape of eyebrows), etc. These feature sets can be compared with pre-stored standard feature sets to determine the identity and authority of the experimenter; the standard feature set includes known facial features, which are obtained through a previous registration process and are used to establish the identity profile or authority setting of each experimenter. When the experimenter is authenticated through the face recognition system The extracted facial feature set will be compared with the standard feature set to determine whether it meets the preset similarity threshold, thereby deciding whether to grant experimental permissions; the experimental sample table is used to be constructed according to the type of experiment. This table lists the specific sample needs or requirements of each experiment, including experimental materials and experimental operation steps; the role of the association mapping table is to ensure that each experimenter has a corresponding experimental sample table; the experimental location table is used to generate the experimenter's experimental location information based on the experimenter's experimental type, including the specific location of the laboratory or other relevant location information. The experimental location is sent to the experimenter to ensure that they know where to conduct the experiment.

[0057] In an optional embodiment, the abnormal matching unit 4 receives the flow trajectory node set transmitted by the sample tracking unit 3, and matches the flow trajectory node set with the standard flow trajectory node set pre-stored in the database to obtain the matching degree between the two, and compares the matching degree with a preset matching degree threshold. If it is greater than the preset matching degree threshold, it indicates that the flow trajectory of the experimental sample is normal; otherwise, it indicates that the flow trajectory of the experimental sample is abnormal, and flow abnormality information is generated. The flow abnormality information includes the abnormal flow trajectory node and the level of the flow trajectory abnormality of the experimental sample, and the flow abnormality information is transmitted to the abnormal warning unit 5.

[0058] In an optional embodiment, the experiment monitoring unit 6 is used to capture the experimental operations of the experimenter by setting up a video monitoring device at the experimental position corresponding to the experiment type to obtain the experimental operation video, and generate an experiment full process monitoring table based on the experiment type and the experimental professional knowledge, and display the experiment full process monitoring table on the smart terminal at the experimental position corresponding to the experiment type. The experimenter uses the experimental image acquisition device to capture images of each process of the experiment according to the experiment full process monitoring table displayed on the smart terminal to obtain an experimental process image set, and transmits the experimental operation video to the experiment review unit 7, and transmits the experimental process image set to the performance evaluation unit 8.

[0059] It should be noted that the full-process monitoring form for the experiment is a form or document generated based on the specific experiment type and relevant experimental expertise, which is used to record and display the complete operational process of the experiment.

[0060] In an optional embodiment, the performance assessment unit 8 receives the experimental process image set transmitted by the experimental monitoring unit 6, and matches the experimental process image set with the preset standard experimental process image set to obtain the comprehensive score of the experimenter corresponding to the experimental process image set, and transmits the comprehensive score of the experimenter to the experimental review unit 7.

[0061] In an optional embodiment, the experiment review unit 7 receives the experiment operation video transmitted by the experiment monitoring unit 6 and the comprehensive score of the experimenter transmitted by the performance evaluation unit 8, and constructs a comprehensive score fitting curve based on the historical comprehensive score of the experimenter, and determines whether the error between the comprehensive score of the experimenter and the comprehensive score fitting curve is within a preset error threshold range. If it is within the preset error threshold range, it indicates that the comprehensive score of the experimenter is valid. Otherwise, the experiment administrator reviews the experiment operation video to determine whether the comprehensive score of the experimenter is valid.

[0062] In an optional embodiment, the abnormal warning unit 5 receives the flow abnormality information transmitted by the abnormal matching unit 4, and matches the flow abnormality information with the decision support table pre-stored in the database to obtain the abnormal adjustment strategy corresponding to the flow abnormality information, and sends the abnormal adjustment strategy to the laboratory administrator, who makes corresponding adjustments to the experimental personnel.

[0063] It should be noted that a decision support table is a data table or data structure containing predefined decision information, which is designed to help the system or operator make decisions or take actions based on specific input conditions. The decision support table is used to store information related to abnormal adjustment strategies.

[0064] In the second embodiment, the present invention proposes a laboratory full-process quality control management system based on paperless office. Compared with the first embodiment, this embodiment further includes: tracking the real-time location of the experimental sample in the laboratory through RFID radio frequency identification technology, and recording the flow trajectory node set of the experimental sample in the laboratory through a reader, including the following steps:

[0065] A1. Set up multiple readers and multiple target tags in the laboratory, set the target tags on the experimental samples, and build an RSSI ranging model. The RSSI ranging model is as follows:

[0066]

[0067] Among them, P r (d) represents the signal strength value at the receiving end, RSSI0 represents the signal strength value at d0 meters, d represents the distance between the target tag and the reader, γ represents the path loss factor, d0 represents the reference distance, X σ represents shadow fading, and X σ is a random variable subject to N(0,σ);

[0068] A2. Multiple readers receive signals from multiple target tags and record the signal strengths to obtain a signal strength set. Based on the RSSI ranging model, the signal strength set is converted into the distance from the target tag to each reader to obtain a distance set.

[0069] A3. Construct an objective function based on the distance set. The objective function is as follows:

[0070]

[0071] Where f(x,y,z) represents the distance error between the estimated coordinates (x,y,z) and the actual coordinates to the i-th reader, d i Indicates the distance from the actual coordinate to the i-th reader, that is, the i-th distance in the distance set, D irepresents the distance from the estimated coordinates (x, y, z) to the i-th reader, and (X i ,Y i ,Z i ) represents the coordinates of the i-th reader;

[0072] A4. Solve the objective function based on the improved grey wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample, and obtain the flow trajectory node set of the experimental sample based on the multiple positioning coordinates.

[0073] In an optional embodiment, solving the objective function based on the improved gray wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample includes the following steps:

[0074] B1. Generate and initialize the gray wolf population, and each gray wolf corresponds to the estimated coordinates in space;

[0075] B2. Calculate the fitness of the gray wolf individuals and save the top three gray wolf individuals with the highest fitness, that is, the three estimated coordinates closest to the current positioning coordinates;

[0076] B3. Calculate the coefficient factor of the gray wolf optimization algorithm based on the positions of the three gray wolf individuals. The coefficient factor calculation formula is as follows:

[0077]

[0078] Among them, A and B represent the coefficient factors of the gray wolf optimization algorithm, a represents the linear coefficient, and r1 and r2 represent random parameters in the interval [0,1].

[0079] B4. Update the positions of the three gray wolf individuals in the gray wolf population. The position update formula is as follows:

[0080]

[0081] in, represents the speed at the k+1th iteration, represents the position of the linear coefficient at the k+1th iteration, r1, r2 and r3 represent random parameters in the interval [0,1], x1, x2 and x3 represent the positions of three gray wolf individuals, c1, c2 and c3 represent acceleration factors, and ω represents the inertia weight;

[0082] B5. Repeat steps B2-B4 until the number of iterations reaches a preset number to obtain the location information of the alpha wolf, that is, the positioning coordinates of the experimental sample.

[0083] In an optional embodiment, matching the flow trajectory node set with a standard flow trajectory node set pre-stored in a database to obtain a matching degree between the two includes the following steps:

[0084] C1. Calculate the Euclidean distance between each flow trajectory node in the flow trajectory node set and each standard flow trajectory node in the standard flow trajectory node set to obtain a Euclidean distance set, and use the standard flow trajectory node corresponding to the minimum Euclidean distance as the matching node of the flow trajectory node;

[0085] C2. construct an undirected graph, using the flow trajectory nodes in the flow trajectory node set and the standard flow trajectory nodes in the standard flow trajectory node set as nodes of the undirected graph;

[0086] C3. Connect the matched flow trajectory nodes and the standard flow trajectory nodes based on the matching nodes to obtain the connection edges of the undirected graph, and use the Euclidean distance as the weight of the connection edge to obtain a weighted undirected graph;

[0087] C4. Sum the weights of the connecting edges in the weighted undirected graph to obtain a weight sum, and compare the weight sum with a preset comprehensive threshold to obtain a matching degree between the flow trajectory node set and the standard flow trajectory node set pre-stored in the database.

[0088] In an optional embodiment, matching the experimental process image set with a preset standard experimental process image set to obtain a comprehensive score of the experimenter corresponding to the experimental process image set includes the following steps:

[0089] D1. Perform feature extraction on the experimental process image set and the standard experimental process image set using the SIFT algorithm to obtain a first feature point set of the experimental process image set and a second feature point set of the standard experimental process image set, and match the first feature point set and the second feature point set using the KNN algorithm to obtain a set of matching pairs;

[0090] D2, using the nearest neighbor algorithm to search for the α of each matching point in the matching pair set L and α R neighboring points, where L and R represent the experimental process image and the standard experimental process image respectively, and α L and α R Take 10;

[0091] D3. If the intersection of the neighboring points of each matching point in the experimental process image and the standard experimental process image is not empty, the average distance from the neighboring intersection point to each matching point is calculated, and the neighboring point intersection is updated based on the average distance. The intersection of the neighboring points of each matching point is taken as the union to obtain the union set. The neighboring point intersection update formula is as follows:

[0092]

[0093] where γ represents the ratio coefficient, and Represents the average distance from the i-th intersection point to each matching point in the experimental process image, Represents the average distance from the i-th intersection point to each matching point in the standard experimental process image;

[0094] D4. Grid the image using the diagonal length of the experimental process image and the standard experimental process image as the side length of each grid. If the average distance of the matching points in the grid is less than the preset average length threshold, the grid is divided into 3×3 grids.

[0095] D5. Obtain a set of position change vectors based on the coordinates of the matching pairs in the union set, and calculate the cosine similarity of the position change vectors of the matching points in each grid. If the number of position change vectors of the matching points in a grid is less than the required minimum number of samples, the nine surrounding grids are included. If it is still less than the required minimum number of samples, the matching pairs in the grid are filtered as incorrect matching pairs. Otherwise, recalculate the cosine similarity, and retain the matching pairs whose mean cosine similarity is greater than the preset mean threshold as inliers, thereby obtaining a filtered set of matching pairs.

[0096] D6. The number of matching points in the filtered matching pair set is used as the comprehensive score of the experimenter.

[0097] It should be noted that the SIFT algorithm is an algorithm for detecting and describing local features in an image; the nearest neighbor algorithm is a simple and effective algorithm for prediction or classification based on data similarity measurement.

[0098] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A laboratory full-process quality control management system based on paperless office, comprising an identity recognition unit (1), an experiment arrangement unit (2), a sample tracking unit (3) and an abnormality matching unit (4), characterized in that: It also includes an abnormal warning unit (5), an experimental monitoring unit (6), an experimental review unit (7) and a performance evaluation unit (8); The identity recognition unit (1) is used to collect facial images of the experimenter by arranging a facial image collection device outside the laboratory to obtain the facial image of the experimenter, and to extract features from the facial image using a trained feature extraction model to obtain a facial feature set, and to calculate the similarity between the facial feature set and each standard feature set in a standard feature set sequence pre-stored in a database, and to compare the similarity with a preset similarity threshold. If the similarity is greater than the preset similarity threshold, the basic information of the experimenter is obtained and transmitted to the experiment arrangement unit (2). Otherwise, it indicates that the experimenter does not have experimental authority. The experiment arrangement unit (2) receives the basic information of the experimenter transmitted by the identification unit (1), and constructs an experiment sample table based on the experiment type, and maps and associates the experiment sample table with the basic information of the experimenter to obtain an association mapping table, and matches the association mapping table with the basic information of the experimenter to obtain the experiment sample information required by the experimenter, and obtains the experiment type of the experimenter based on the experiment sample information required by the experimenter, and generates the experiment location of the experimenter based on the experiment type, and sends the experiment location to the experimenter, and transmits the experiment sample information required by the experimenter to the sample tracking unit (3); The sample tracking unit (3) receives the experimental sample information required by the experimenter transmitted by the experiment arrangement unit (2), tracks the real-time location of the experimental sample in the laboratory through RFID radio frequency identification technology, records the flow trajectory node set of the experimental sample in the laboratory through a reader, and transmits the flow trajectory node set to the abnormal matching unit (4); The abnormal matching unit (4) receives the flow trajectory node set transmitted by the sample tracking unit (3), matches the flow trajectory node set with the standard flow trajectory node set pre-stored in the database to obtain a matching degree between the two, and compares the matching degree with a preset matching degree threshold. If the matching degree is greater than the preset matching degree threshold, it indicates that the flow trajectory of the experimental sample is normal; otherwise, it indicates that the flow trajectory of the experimental sample is abnormal, and generates flow abnormality information, the flow abnormality information including the abnormal flow trajectory node and the abnormal level of the flow trajectory of the experimental sample, and transmits the flow abnormality information to the abnormality warning unit (5); The experiment monitoring unit (6) is used to capture the experimental operation of the experimenter by setting up a video monitoring device at the experimental position corresponding to the experiment type, so as to obtain an experimental operation video, and generate an experiment full process monitoring table based on the experimental professional knowledge according to the experiment type, and display the experiment full process monitoring table on the smart terminal at the experimental position corresponding to the experiment type. The experimenter uses the experiment image acquisition device to capture images of each process of the experiment based on the experiment full process monitoring table displayed on the smart terminal to obtain an experimental process image set, and transmits the experimental operation video to the experiment review unit (7), and transmits the experimental process image set to the performance evaluation unit (8).

2. A laboratory full-process quality control management system based on paperless office according to claim 1, characterized in that: The performance evaluation unit (8) receives the experimental process image set transmitted by the experimental monitoring unit (6), matches the experimental process image set with a preset standard experimental process image set to obtain a comprehensive score of the experimenter corresponding to the experimental process image set, and transmits the comprehensive score of the experimenter to the experimental review unit (7); The experiment review unit (7) receives the experiment operation video transmitted by the experiment monitoring unit (6) and the comprehensive score of the experimenter transmitted by the performance evaluation unit (8), and constructs a comprehensive score fitting curve based on the historical comprehensive score of the experimenter, and determines whether the error between the comprehensive score of the experimenter and the comprehensive score fitting curve is within a preset error threshold range. If it is within the preset error threshold range, it indicates that the comprehensive score of the experimenter is valid. Otherwise, the experiment administrator checks the experiment operation video to determine whether the comprehensive score of the experimenter is valid.

3. A laboratory full-process quality control management system based on paperless office according to claim 1, characterized in that: The abnormality warning unit (5) receives the flow abnormality information transmitted by the abnormality matching unit (4), and matches the flow abnormality information with a decision support table pre-stored in a database to obtain an abnormality adjustment strategy corresponding to the flow abnormality information, and sends the abnormality adjustment strategy to the laboratory administrator, who makes corresponding adjustments to the laboratory personnel.

4. A laboratory full-process quality control management system based on paperless office according to claim 1, characterized in that: Tracking the real-time location of an experimental sample in a laboratory by using RFID radio frequency identification technology and recording the flow trajectory node set of the experimental sample in the laboratory by using a reader includes the following steps: A1. Multiple readers and multiple target tags are set in the laboratory, and the target tags are set on the experimental samples. An RSSI ranging model is constructed. The RSSI ranging model is as follows: ； in, Indicates the signal strength value of the receiving end. express Signal strength value at meters, Indicates the distance between the target tag and the reader, represents the path loss factor, represents the reference distance, represents shadow fading, and To obey A random variable; A2. The plurality of readers receive signals emitted by the plurality of target tags and record the signal strengths of the signals to obtain a signal strength set, and convert the signal strength set into a distance from the target tag to each of the readers based on the RSSI ranging model to obtain a distance set; A3. Construct an objective function based on the distance set. The objective function is as follows: ； in, Represents estimated coordinates The distance error between the actual coordinates and the i-th reader, Represents the distance from the actual coordinate to the i-th reader, that is, the i-th distance in the distance set, Represents estimated coordinates The distance to the i-th reader, and , represents the coordinates of the i-th reader; A4. Solve the objective function based on the improved grey wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample, and obtain the flow trajectory node set of the experimental sample based on the multiple positioning coordinates.

5. A laboratory full-process quality control management system based on paperless office according to claim 4, characterized in that: Solving the objective function based on the improved gray wolf optimization algorithm to obtain multiple positioning coordinates of the experimental sample includes the following steps: B1. Generate and initialize a gray wolf population, and each gray wolf corresponds to the estimated coordinates in space; B2. Calculate the fitness of the gray wolf individuals and save the top three gray wolf individuals with the largest fitness, i.e., the three estimated coordinates that are currently closest to the positioning coordinates; B3. Calculate the coefficient factor of the gray wolf optimization algorithm based on the positions of the three gray wolf individuals. The coefficient factor calculation formula is as follows: ； in, and represents the coefficient factor of the gray wolf optimization algorithm, represents the linear coefficient, and Representation interval Random parameters of ; B4. Update the positions of the three gray wolf individuals in the gray wolf population. The position update formula is as follows: ； in, represents the speed at the k+1th iteration, Indicates the position of the linear coefficient at the k+1th iteration, 、 and Representation interval Random parameters within, 、 and represents the positions of the three gray wolf individuals, 、 and represents the acceleration factor, represents the inertia weight; B5. Repeat steps B2-B4 until the number of iterations reaches a preset number to obtain the location information of the leader wolf, that is, the positioning coordinates of the experimental sample.

6. A laboratory full-process quality control management system based on paperless office according to claim 1, characterized in that: Matching the flow trajectory node set with a standard flow trajectory node set pre-stored in a database to obtain a matching degree between the two includes the following steps: C1. Calculate the Euclidean distance between each flow trajectory node in the flow trajectory node set and each standard flow trajectory node in the standard flow trajectory node set to obtain a Euclidean distance set, and use the standard flow trajectory node corresponding to the minimum Euclidean distance as the matching node of the flow trajectory node; C2. constructing an undirected graph, using the flow trajectory nodes in the flow trajectory node set and the standard flow trajectory nodes in the standard flow trajectory node set as nodes of the undirected graph; C3. Connecting the matched flow trajectory nodes and the standard flow trajectory nodes based on the matching nodes to obtain connection edges of an undirected graph, and using the Euclidean distance as the weight of the connection edges to obtain a weighted undirected graph; C4. Summing the weights of the connecting edges in the weighted undirected graph to obtain a weight sum, and comparing the weight sum with a preset comprehensive threshold to obtain a matching degree between the flow trajectory node set and a standard flow trajectory node set pre-stored in a database.

7. A laboratory full-process quality control management system based on paperless office according to claim 2, characterized in that: Matching the experimental process image set with a preset standard experimental process image set to obtain a comprehensive score of the experimenter corresponding to the experimental process image set includes the following steps: D1. Performing feature extraction on the experimental process image set and the standard experimental process image set using a SIFT algorithm to obtain a first feature point set of the experimental process image set and a second feature point set of the standard experimental process image set, and matching the first feature point set and the second feature point set using a KNN algorithm to obtain a matching pair set; D2, using the nearest neighbor algorithm to search for each matching point in the matching pair set and neighboring points, among which and Respectively represent the experimental process image and the standard experimental process image, and and Take 10; D3. If the intersection of the neighboring points of each matching point in the experimental process image and the standard experimental process image is not empty, then the average distance from the neighboring intersection point to each matching point is calculated, and the neighboring point intersection is updated based on the average distance, and the intersection of the neighboring points of each matching point is taken as a union to obtain a union set. The neighboring point intersection update formula is as follows: ； in, represents the ratio coefficient, and , Represents the average distance from the i-th intersection point to each matching point in the experimental process image, Represents the average distance from the i-th intersection point to each matching point in the standard experimental process image; D4. Gridding the image using the diagonal length of the experimental process image and the standard experimental process image as the side length of each grid. If the average distance of matching points in the grid is less than a preset average length threshold, then dividing the grid into 3×3 grids; D5. Obtain a set of position change vectors based on the coordinates of the matching pairs in the union set, and calculate the cosine similarity of the position change vectors of the matching points in each grid. If the number of position change vectors of the matching points in a grid is less than the required minimum number of samples, the nine surrounding grids are included. If the number is still less than the required minimum number of samples, the matching pairs in the grid are filtered as incorrect matching pairs. Otherwise, recalculate the cosine similarity, and retain the matching pairs whose mean cosine similarity is greater than a preset mean threshold as inliers, thereby obtaining a filtered set of matching pairs. D6. The number of matching points in the filtered matching pair set is used as the comprehensive score of the experimenter.

Images