A digital intelligent management method and system

Abstract

The present invention relates to the field of data processing technology, and more specifically, to a digital intelligent management method and system, the method comprising: clustering the storage data of any product to obtain multiple clusters, calculating the suspected abnormality degree of each cluster, screening multiple data points from each cluster according to the characteristic strength of each data point from large to small to obtain target data points, the number of target data points in each cluster is negatively correlated with the suspected abnormality degree of the corresponding cluster, constructing an isolation forest based on the target data points in all clusters, performing abnormality detection on the storage data, and managing the product based on the results of the abnormality detection. The present invention can improve the accuracy of abnormality detection on storage data, so that the product can be accurately managed intelligently.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a digital intelligent management method and system. Background Art

[0002] In modern enterprise management, digital transformation has become an important strategy to improve operational efficiency and reduce costs. Digital management can optimize the business processes of enterprises by utilizing information technology and data analysis. Taking warehouse management as an example, many management systems currently focus on data recording, but lack in-depth analysis of data, resulting in the need for manual monitoring to prevent products from expiring due to time, which consumes a lot of manpower and material resources. Therefore, there is an urgent need for an intelligent management method to manage products.

[0003] In the related technology, for example, a patent application document with publication number CN112053119A discloses a cold chain logistics warehousing intelligent management system based on big data. The system includes: dividing the warehouse storage area, obtaining the packaging shape parameters and commodity types of the incoming products, obtaining the storage area and shelf number of the incoming products, and marking the RFID temperature tag on the surface of each product packaging to collect the temperature of the product during storage, and issuing early warnings for products with abnormal temperature changes.

[0004] However, the above scheme only compares the storage temperature of each incoming product received with the standard storage temperature range of its corresponding incoming storage area to determine the abnormal temperature for early warning, which may not fully reflect the actual storage conditions of the product. Relying solely on the standard storage temperature range may cause some products to not be identified at critical temperatures, which may reduce the accuracy of abnormality detection to a certain extent, resulting in the inability to accurately manage products intelligently. Summary of the invention

[0005] In order to solve the problem of being unable to accurately perform intelligent management on products, the present invention provides a digital intelligent management method and system.

[0006] According to a first aspect of the present invention, there is provided a digital intelligent management method, comprising:

[0007] Get the storage data of any product, including temperature, humidity and inventory;

[0008] Cluster the warehouse data to obtain multiple clusters, and calculate the suspected abnormality degree of each cluster. The suspected abnormality degree reflects the abnormality of the temperature, humidity and inventory value of the data points in the corresponding cluster;

[0009] According to the characteristic strength of each data point, multiple data points are screened from large to small in each cluster to obtain the target data point. The number of target data points in each cluster is negatively correlated with the suspected abnormality degree of the corresponding cluster.

[0010] Construct an isolation forest based on the target data points in all clusters, perform anomaly detection on warehouse data, and manage products based on the results of anomaly detection;

[0011] The method for obtaining the feature strength is as follows: calculating the average value and standard deviation of the distance between any data point and its neighboring data points in any cluster, normalizing the ratio of the average value and the standard deviation of the distance, and obtaining the feature strength of any data point.

[0012] The present invention combines temperature, humidity and inventory to perform anomaly detection on warehouse data, so that the abnormal situation of warehouse data can be evaluated from multiple aspects, and relies on the isolation forest constructed by target data points screened from each cluster to perform anomaly detection on warehouse data, which can avoid the influence of the number of abnormal data on the anomaly detection using isolation forest, thereby ensuring the accuracy of the anomaly detection results, and then based on the results of anomaly detection with higher accuracy, accurately perform intelligent management of products.

[0013] Preferably, the number of target data points in each cluster satisfies the following relationship:

[0014] ;

[0015] In the formula, For the The number of target data points in a cluster; For the The number of data points in a cluster; For the The degree of suspected abnormality of each cluster; is the ceiling function.

[0016] Preferably, the method for obtaining the degree of suspected abnormality includes:

[0017] Take temperature, humidity and inventory as a dimension respectively, calculate the difference between the value of any data point in any cluster in any dimension and the average value of all data points in any cluster in any dimension, and get the abnormality degree of any data point in any dimension; calculate the suspected abnormality degree of any cluster:

[0018] ;

[0019] In the formula, For the The degree of suspected abnormality of each cluster; For the The first The data point in The degree of abnormality of the values ​​of each dimension; is the number of dimensions; For the The number of data points in a cluster; is the normalization function.

[0020] The present invention integrates data from multiple dimensions to determine the suspected abnormality degree of any cluster, which can ensure the accuracy of the determined results.

[0021] Preferably, the degree of abnormality of any data point in any dimension satisfies the following relationship:

[0022] ;

[0023] In the formula, For the The first The data point in The degree of abnormality of the values ​​of each dimension; For the The first The data point in Dimension values; For the All data points in the cluster The average value of the dimension; is the number of dimensions; for the summation symbol; is the absolute value symbol.

[0024] The calculation formula provided by the present invention can reduce the influence of data of various dimensions at different orders of magnitude, thereby ensuring the accuracy of the determined abnormality degree.

[0025] Preferably, the method for obtaining neighboring data points includes:

[0026] Get the distance between any data point in any cluster and the remaining data points, as well as the preset number;

[0027] A preset number of data points are selected from the remaining data points in order of distance from small to large to obtain the neighboring data points of any data point.

[0028] Preferably, the warehouse data is clustered to obtain multiple clusters, including:

[0029] The warehouse data is clustered using an iterative self-organizing clustering algorithm to obtain multiple clusters.

[0030] The present invention can cluster similar data points in the warehouse data into a cluster, so that analysis can be performed in clusters as units, thereby ensuring the accuracy of the analysis results.

[0031] Preferably, abnormality detection is performed on the warehouse data, including:

[0032] Get the preset abnormal threshold;

[0033] When the average of the anomaly scores of all target data points determined based on the isolation forest is greater than the anomaly threshold, it is determined that the warehouse data is abnormal.

[0034] The present invention can accurately detect abnormal data in warehouse data, thereby realizing intelligent management of products.

[0035] According to a second aspect of the present invention, a digital intelligent management system is provided. The digital intelligent management system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the first aspect of the present invention.

[0036] The present invention has the following effects:

[0037] The method of constructing an isolation forest in the present invention can avoid the isolation forest algorithm from segmenting the abnormal data into the lower layer when the number of abnormal data in the warehouse data is large, thereby ensuring the accuracy of anomaly detection of the warehouse data using the isolation forest algorithm; and when determining the target data points for constructing the isolation forest, it depends on the suspected abnormality degree of each cluster and the feature strength of each data point in each cluster, thereby ensuring the accuracy of the number of determined target data points and the accuracy of the selected target data points, thereby further ensuring the accuracy of anomaly detection of the warehouse data, and then based on the results of anomaly detection with higher accuracy, accurately and intelligently manage the products. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0039] Figure 1 It is a schematic diagram of the steps of a digital intelligent management method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0041] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0042] Reference Figure 1 , a digital intelligent management method, including steps S1 to S4, specifically as follows:

[0043] S1: Get the storage data of any product, including temperature, humidity and inventory.

[0044] In an exemplary embodiment of the present invention, the storage data refers to the temperature, humidity and inventory of each product in the warehouse, which is used to monitor the status of the corresponding products. The temperature and humidity in the storage data can be obtained by temperature sensors and humidity sensors arranged in the area where each product is located, and the inventory in the storage data can be determined by the inbound and outbound data of the corresponding product.

[0045] Optionally, the collection period of the temperature sensor and the humidity sensor may be set to 5 seconds. This embodiment does not specifically limit the setting of the sampling period.

[0046] S2: Cluster the warehouse data to obtain multiple clusters, and calculate the suspected abnormality degree of each cluster. The suspected abnormality degree reflects the abnormality of the temperature, humidity and inventory values ​​of the data points in the corresponding cluster.

[0047] In an exemplary embodiment of the present invention, the suspected abnormality degree refers to data that can reflect the abnormal conditions of the temperature, humidity and inventory values ​​of the data points in each cluster. For example, when the data points in any cluster have more abnormalities in the temperature, humidity or inventory indicators, it means that the cluster contains more abnormal data, and the suspected abnormality degree of the cluster is greater.

[0048] In an exemplary embodiment of the present invention, clustering of warehouse data can be achieved by following the steps below:

[0049] The warehouse data is clustered using an iterative self-organizing clustering algorithm to obtain multiple clusters.

[0050] Optionally, the warehouse data can also be clustered by combining an iterative self-organizing clustering algorithm with a K-means algorithm. This embodiment does not specifically limit the clustering algorithm used. It should be noted that the process of clustering data using an iterative self-organizing clustering algorithm, or using an iterative self-organizing clustering algorithm combined with a K-means algorithm, is a prior art and is not described in detail in this embodiment.

[0051] In an exemplary embodiment of the present invention, the determination of the suspected abnormality degree of each cluster can be achieved by the following steps:

[0052] (1) Taking temperature, humidity and inventory as a dimension respectively, calculate the difference between the value of any data point in any cluster in any dimension and the average value of all data points in any cluster in any dimension, and obtain the abnormality degree of any data point;

[0053] Specifically, the degree of abnormality of any data point in any dimension in any cluster satisfies the following relationship:

[0054] ;

[0055] In the formula, For the The first The data point in The degree of abnormality of the values ​​of each dimension; For the The first The data point in Dimension values; For the All data points in the cluster The average value of the dimension; is the number of dimensions; for the summation symbol; is the absolute value symbol.

[0056] in, For the The first The standard deviation of the values ​​of a data point in all dimensions can reduce the impact of data of different orders of magnitude and dimensions, ensuring the accuracy of the determined degree of abnormality.

[0057] (2) Calculate the degree of suspected abnormality of any cluster.

[0058] Specifically, the suspected abnormality degree of any cluster satisfies the following relationship:

[0059] ;

[0060] In the formula, For the The degree of suspected abnormality of each cluster; For the The first The data point in The degree of abnormality of the values ​​of each dimension; is the number of dimensions. =3; For the The number of data points in a cluster; is the normalization function; is the summation symbol.

[0061] S3: According to the feature strength of each data point, multiple data points are screened from large to small to obtain target data points. The number of target data points in each cluster is negatively correlated with the suspected abnormality degree of the corresponding cluster.

[0062] In an exemplary embodiment of the present invention, feature strength refers to reference data used to screen target data points in each cluster. The feature strength of any data point can be determined based on the similarity between the data point and its neighboring data points in the corresponding cluster. For example, when the similarity between the data point and its neighboring data points in the corresponding cluster is low, it means that the feature strength of the data point is high.

[0063] The target data point refers to the data point with a larger characteristic intensity selected from each cluster, which is used as the data basis for anomaly detection. It should be noted that the present invention sets the number of target data points in each cluster to be negatively correlated with the suspected anomaly degree of the corresponding cluster, which can reduce the number of abnormal data and the impact on the anomaly detection results.

[0064] Specifically, the number of target data points in each cluster satisfies the following relationship:

[0065] ;

[0066] In the formula, For the The number of target data points in a cluster; For the The number of data points in a cluster; For the The degree of suspected abnormality of each cluster; is the ceiling function.

[0067] In another embodiment, the number of target data points in each cluster may be determined by other calculation formulas, such as not using a product calculation formula but using a difference calculation formula such as Determine, among which, is a constant and can be set according to the specific situation. The value of is not negative.

[0068] In an exemplary embodiment of the present invention, the determination of the characteristic strength of each data point can be achieved by the following steps:

[0069] (1) Calculate the mean and standard deviation of the distance between any data point and its neighboring data points in any cluster;

[0070] In an exemplary embodiment of the present invention, the determination of neighboring data points of any data point can be achieved by the following steps:

[0071] Get the distance between any data point in any cluster and the remaining data points, as well as the preset number; select a preset number of data points from the remaining data points in order of distance from small to large to obtain the neighboring data points of any data point.

[0072] Optionally, you can set the preset quantity to In this embodiment =10, then calculate the distance between any data point in any cluster and the remaining data points, and select 10 data points from the remaining data points in order of distance from small to large, so as to obtain the neighboring data points of the data point.

[0073] Optionally, the distance between any data point in any cluster and the remaining data points can be Euclidean distance or Manhattan distance. Of course, according to the specific situation, a suitable distance metric can be selected to measure the similarity between each data point and the remaining data points. This embodiment does not specifically limit the type of distance.

[0074] (2) Normalize the ratio of the mean and standard deviation of the distance to obtain the feature intensity of any data point.

[0075] Specifically, the characteristic strength of any data point satisfies the following relationship:

[0076] ;

[0077] In the formula, For the The first The feature strength of each data point; For the The first Data points and The distance between the nearest neighbor data points; For the The first The average distance between a data point and all its neighboring data points; is the number of neighboring data points. =10; is the normalization function.

[0078] Optionally, other normalization functions may be used for normalization processing, such as using an exponential function for normalization processing. This embodiment does not specifically limit the specific manner of the normalization processing.

[0079] Furthermore, after determining the number of target data points in any cluster and the feature strength of each data point in the cluster, a corresponding number of data points can be screened from the cluster according to the feature strength of each data point, from large to small, to obtain the target data points in the cluster.

[0080] S4: Build an isolation forest based on the target data points in all clusters and perform anomaly detection on the warehouse data to manage products based on the results of anomaly detection.

[0081] It should be noted that the traditional isolation forest algorithm is an anomaly detection algorithm based on a random binary tree. It can use the number of sample segmentation to measure the degree of alienation of the sample to determine whether the sample is abnormal. The theoretical basis of this algorithm is that in a data set, normal data points are usually densely distributed and have a large number, while abnormal data are often isolated points with a relatively small number. The isolation forest algorithm can segment a small number of data points in the upper layer of the isolation tree, that is, close to the parent node of the isolation tree, so that the path length of the small number of data points in the isolation forest is shorter, and the shorter the path length, the higher the degree of abnormality of the data point, so that the abnormal data in the data set can be determined according to the path length of each data point.

[0082] However, if the storage data of any product is abnormal, such as the temperature in the area where the product is located is abnormal, the temperature in the storage data may be continuously abnormal, resulting in an increasing number of abnormal data points in the storage data. When the storage data of the product is segmented using the isolation forest algorithm, the abnormal data points may be segmented in the lower layer of the isolated tree, making the path length of the abnormal data points in the isolated tree longer. Since the data points with shorter path lengths have higher abnormality, the abnormal data in the storage data cannot be accurately detected. Therefore, the present invention constructs an isolation forest based on the target data points screened from all clusters, which can avoid the influence of the number of abnormal data, so that the abnormal data in the storage data can be accurately segmented.

[0083] In an exemplary embodiment of the present invention, abnormality detection of warehouse data can be achieved by following the steps below:

[0084] Get the preset anomaly threshold; when the average of the anomaly scores of all target data points determined based on the isolation forest is greater than the anomaly threshold, it is determined that the warehouse data is abnormal.

[0085] Among them, the anomaly score is a professional term in the isolation forest algorithm, and the value range is 0-1, which is not described in detail in this embodiment. It should be noted that the anomaly score of any target data point can be calculated based on the average path length of all isolated trees in the isolation forest for the data point, and the shorter the average path length, the higher the anomaly score of the data point.

[0086] Optionally, the anomaly threshold may be set to 0.7. If the average value of the anomaly scores of all target data points is greater than 0.7, it indicates that there are abnormal data in the warehouse data. If the average value is less than or equal to 0.7, it indicates that there are no abnormal data in the warehouse data.

[0087] Furthermore, when it is determined that there is abnormal data in the warehouse data, an early warning can be issued to remind operators that there are abnormalities in the temperature, humidity, or inventory level in the area where the product is located, so that the status of the product can be monitored in a timely manner and appropriate measures can be taken in a timely manner to reduce product losses.

[0088] The present invention also provides a digital intelligent management system, the system includes a memory and a processor, and a computer program is stored in the memory, the computer program integrates the function of a digital intelligent management method, when the computer program is executed, a digital intelligent management method can be used to improve the accuracy of abnormality detection of warehouse data, so that products can be accurately managed digitally and intelligently.

[0089] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0090] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A digital intelligent management method, characterized in that: include: Get the storage data of any product, including temperature, humidity and inventory; Cluster the warehouse data to obtain multiple clusters, and calculate the suspected abnormality degree of each cluster. The suspected abnormality degree reflects the abnormality of the temperature, humidity and inventory value of the data points in the corresponding cluster; According to the feature strength of each data point, multiple data points are screened from large to small in each cluster to obtain the target data point. The number of target data points in each cluster satisfies the following relationship: ; In the formula, For the The number of target data points in a cluster; For the The number of data points in a cluster; For the The degree of suspected abnormality of each cluster; It is an upward rounding function, which is negatively correlated with the suspected abnormality degree of the corresponding cluster; The method for obtaining the suspected abnormality degree includes: taking temperature, humidity and inventory as a dimension respectively, calculating the difference between the value of any data point in any cluster in any dimension and the average value of all data points in any cluster in any dimension, and obtaining the abnormality degree of any data point in any dimension; calculating the suspected abnormality degree of any cluster: ； In the formula, For the The first The data point in The degree of abnormality of the values ​​of each dimension; is the number of dimensions; is the normalization function; The degree of abnormality of any data point in any dimension satisfies the following relationship: ； In the formula, For the The first The data point in Dimension values; For the All data points in the cluster The average value of the dimension; Construct an isolation forest based on the target data points in all clusters, perform anomaly detection on warehouse data, and manage products based on the results of anomaly detection; The method for obtaining the feature strength is as follows: calculating the average value and standard deviation of the distance between any data point and its neighboring data points in any cluster, normalizing the ratio of the average value and the standard deviation of the distance, and obtaining the feature strength of any data point.

2. A digital intelligent management method according to claim 1, characterized in that: The method for obtaining the neighboring data points includes: Get the distance between any data point in any cluster and the remaining data points, as well as the preset number; The preset number of data points are selected from the remaining data points in order of distance from small to large to obtain neighboring data points of any data point.

3. A digital intelligent management method according to claim 1, characterized in that: The clustering of the warehouse data to obtain a plurality of clusters includes: The warehouse data is clustered using an iterative self-organizing clustering algorithm to obtain a plurality of clusters.

4. A digital intelligent management method according to claim 1, characterized in that: The performing abnormality detection on the warehouse data includes: Get the preset abnormal threshold; When the average value of the abnormal scores of all target data points determined based on the isolation forest is greater than the abnormal threshold, it is determined that the warehouse data is abnormal.

5. A digital intelligent management system, characterized in that: The digital intelligent management system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the method as described in any one of claims 1 to 4.

Images