CN-121981646-A - Apple storage environment abnormal data detection method

CN121981646ACN 121981646 ACN121981646 ACN 121981646ACN-121981646-A

Abstract

The invention relates to the technical field of data processing, in particular to a method for detecting abnormal data of an apple warehouse environment, which comprises the steps of uniformly arranging monitoring points in an apple warehouse according to a grid shape, collecting historical monitoring data of each environmental index at each historical moment at each monitoring point, acquiring initial abnormal degrees of each environmental index at each monitoring point at the current moment according to the data change rule of the historical monitoring data of each environmental index at each monitoring point, correcting the initial abnormal degrees of each environmental index at each monitoring point according to the position of a grid where each monitoring point is located and the distribution condition of the initial abnormal degrees of each environmental index at each monitoring point, obtaining the final abnormal degrees of each environmental index at each monitoring point at the current moment, further monitoring and early warning the environment in the apple warehouse, and improving the accuracy of abnormality detection of the apple warehouse environment.

Inventors

LI SANHONG
GAO JIANWEI
GAO JIANBIN
WANG XIAOHU
YANG ZONGWEN

Assignees

陕西果业集团子洲有限公司

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (9)

1. The method for detecting the abnormal data of the apple storage environment is characterized by comprising the following steps of: uniformly arranging monitoring points in an apple warehouse according to a grid shape, collecting real-time monitoring data of each environmental index at the current time and historical monitoring data of a preset number of historical time at each monitoring point, wherein the real-time monitoring data and the historical monitoring data are normalized data; aiming at any environmental index, according to the data change similarity of the historical monitoring data of any environmental index at any monitoring point and the difference between the real-time monitoring data and the historical monitoring data of any environmental index at any monitoring point, acquiring the initial abnormality degree of any environmental index at any monitoring point at the current moment; acquiring the initial abnormality degree of any environmental index at each monitoring point at the current moment, and correcting the initial abnormality degree of any environmental index at each monitoring point according to the position of the grid of each monitoring point and the distribution condition of the initial abnormality degree of any environmental index at each monitoring point to obtain the final abnormality degree of any environmental index at each monitoring point at the current moment; Acquiring the final abnormality degree of each environmental index at each monitoring point at the current moment, and monitoring and early warning the environment in the apple warehouse according to the final abnormality degree of each environmental index at each monitoring point at the current moment.
2. The method for detecting abnormal data in an apple warehouse environment according to claim 1, wherein the obtaining the initial abnormal degree of any environmental index at any monitoring point at the current time according to the similarity of data change of the historical monitoring data of any environmental index at any monitoring point and the difference between the real-time monitoring data and the historical monitoring data of any environmental index at any monitoring point comprises: Forming a data sequence by all the historical monitoring data of any environmental index at any monitoring point according to time sequence, constructing a graph according to the data sequence, wherein the abscissa of the graph represents historical time, the ordinate represents the historical monitoring data, acquiring the tangential slope of the historical monitoring data at each historical time according to the graph, recording the tangential slope as a local change trend characteristic value of the any environmental index at each historical time, and clustering the local change trend characteristic value of the any environmental index at each historical time to obtain at least one cluster; According to the continuity of the historical moment corresponding to the local change trend characteristic value in each cluster, at least two historical subintervals of any environmental index at any monitoring point are obtained, and each cluster at least corresponds to one historical subinterval; For any history subperiod, marking a cluster in which the any history subperiod is located as a target cluster, marking the data similarity degree of the any history subperiod as a preset minimum value if the target cluster corresponds to only one history subperiod, and acquiring the data similarity degree of any environmental index at any monitoring point in any history subperiod according to the length difference between the any history subperiod corresponding to the target cluster and other history subperiods if the target cluster corresponds to at least two history subperiods; The historical monitoring data of the any environmental index at the any monitoring point at each historical time in the any historical subperiod is formed into a historical data subsequence, and the abnormal score of the real-time monitoring data of the any environmental index at the any monitoring point in the any historical subperiod is obtained according to the difference between the real-time monitoring data of the any environmental index at the any monitoring point and the historical data subsequence; The method comprises the steps of obtaining the data similarity degree corresponding to each history subperiod, obtaining the abnormal score of real-time monitoring data of any environmental index at any monitoring point in each history subperiod, and obtaining the initial abnormal degree of any environmental index at any monitoring point at the current time according to the data similarity degree corresponding to each history subperiod and the abnormal score of real-time monitoring data of any environmental index at any monitoring point in each history subperiod.
3. The method for detecting abnormal data in an apple warehouse environment according to claim 2, wherein the obtaining at least two historical subintervals of any environmental indicator at any monitoring point according to the continuity of the historical moments corresponding to the local variation trend feature values in each cluster comprises: According to any cluster, the historical moments corresponding to the local change trend characteristic values in any cluster are arranged according to time sequences to obtain a historical moment sequence, the historical moment sequence is divided according to continuity between every two adjacent historical moments in the historical moment sequence to obtain at least one historical moment sequence of any environmental index at any monitoring point, the historical moment sequence at least comprises two historical moments, every two adjacent historical moments are continuous, and a time period corresponding to each historical moment sequence is recorded as a historical subtime period.
4. The method for detecting abnormal data in an apple warehouse environment according to claim 2, wherein the obtaining the data similarity of the any environmental index at the any monitoring point in the any history sub-period according to the length difference between the any history sub-period and other history sub-periods corresponding to the target cluster comprises: Recording the history subperiods except for any history subperiod corresponding to the target cluster as other history subperiods, and calculating the absolute value of the duration difference value between the any history subperiod and any other history subperiod according to any other history subperiod to obtain the length difference degree between the any history subperiod and any other history subperiod; Calculating the length difference degree between any one history subperiod and each other history subperiod, taking the sum of the accumulated sum of all the length difference degrees and a preset constant as a denominator, taking the number of all the other history subperiods as a numerator, obtaining the length approach degree between any one history subperiod and other history subperiods, and carrying out linear normalization on the length approach degree to obtain the similarity expression degree of any one history subperiod; and calculating the ratio of the number of all the history sub-periods corresponding to the target cluster to the number of all the history sub-periods corresponding to all the clusters, and calculating the product of the ratio and the similarity expression degree of any history sub-period to obtain the data similarity degree of any environmental index at any monitoring point in any history sub-period.
5. The method for detecting abnormal data in an apple warehouse environment according to claim 2, wherein the obtaining the abnormal score of the real-time monitoring data of the any environmental index at the any monitoring point in the any history sub-period according to the difference between the real-time monitoring data of the any environmental index at the any monitoring point and the history data sub-sequence comprises: Acquiring a data fluctuation range of the historical data subsequence by utilizing a 3sigma principle according to all historical monitoring data in the historical data subsequence, and setting an abnormal score of the real-time monitoring data of any environmental index at any monitoring point in any historical subperiod to be 1 if the real-time monitoring data of any environmental index at any monitoring point is not in the data fluctuation range; And if the real-time monitoring data of any environmental index at any monitoring point is in the data fluctuation range, setting the anomaly score of the real-time monitoring data of any environmental index at any monitoring point in any history subperiod to be 0.
6. The method for detecting abnormal data in an apple warehouse according to claim 2, wherein the obtaining, according to the similarity of the data corresponding to each historical subinterval and the abnormal score of the real-time monitoring data of any environmental index at any monitoring point in each historical subinterval, the initial abnormal degree of any environmental index at any monitoring point at the current time comprises: And respectively calculating the ratio of the data similarity degree corresponding to each history sub-period in the accumulation sum of all the data similarity degrees to obtain the similarity influence weight of each history sub-period, and carrying out weighted summation on the abnormal scores of the real-time monitoring data of any environmental index at any monitoring point in each history sub-period according to the similarity influence weight of each history sub-period to obtain the initial abnormal degree of any environmental index at any monitoring point at the current moment.
7. The method for detecting abnormal data in an apple warehouse environment according to claim 1, wherein the correcting the initial abnormality degree of any environmental index at each monitoring point according to the position of the grid where each monitoring point is located and the distribution condition of the initial abnormality degree of any environmental index at each monitoring point to obtain the final abnormality degree of any environmental index at each monitoring point at the current time comprises: constructing an abnormal degree image according to grids of each monitoring point, wherein the grids of each monitoring point are pixel points in the abnormal degree image, the initial abnormal degree of any environmental index at each monitoring point at the current moment is a characteristic value of the pixel point corresponding to each monitoring point in the abnormal degree image at the current moment, and according to the characteristic value of each pixel point in the abnormal degree image, acquiring a gradient amplitude of each pixel point in the abnormal degree image at the current moment by utilizing a Sobel operator; Acquiring initial abnormality degrees of any environmental index at each monitoring point under a preset number of historical moments, acquiring characteristic values of pixel points corresponding to each monitoring point in an abnormality degree image at each historical moment according to the initial abnormality degrees of any environmental index at each monitoring point at each historical moment, recording the characteristic values as historical characteristic values, and acquiring historical gradient amplitude values of each pixel point at each historical moment according to the historical characteristic values of each pixel point at each historical moment; For any monitoring point, marking a pixel point corresponding to the any monitoring point in the abnormal degree image as a target pixel point, respectively calculating the absolute value of the difference between the gradient amplitude of the target pixel point at the current time and each historical gradient amplitude of the target pixel point, and carrying out linear normalization on the average value of all the absolute values of the difference to obtain the abnormal distribution degree of any environmental index at the any monitoring point at the current time; And correcting the initial abnormality degree of any environmental index at any monitoring point by using the abnormality distribution degree of any environmental index at any monitoring point at the current moment to obtain the final abnormality degree of any environmental index at any monitoring point at the current moment.
8. The method for detecting abnormal data in an apple warehouse environment according to claim 7, wherein the correcting the initial abnormal degree of the any environmental index at any monitoring point by using the abnormal distribution degree of the any environmental index at any monitoring point at the current time to obtain the final abnormal degree of the any environmental index at any monitoring point at the current time comprises: And calculating the product of the initial abnormality degree of any environmental index at any monitoring point and the abnormality distribution degree of any environmental index at any monitoring point at the current time to obtain the final abnormality degree of any environmental index at any monitoring point at the current time.
9. The method for detecting abnormal data of apple warehouse environment according to claim 1, wherein the step of obtaining the final degree of abnormality of each environmental indicator at each monitoring point at the current moment, and monitoring and early warning the environment in the apple warehouse according to the final degree of abnormality of each environmental indicator at each monitoring point at the current moment, comprises the steps of: For any monitoring point, acquiring final abnormal degrees of all environmental indexes at the any monitoring point at the current moment, and calculating the average value of all final abnormal degrees corresponding to the any monitoring point to obtain the environmental risk degree of the any monitoring point at the current moment; acquiring the environmental risk degree of each monitoring point at a preset number of historical moments, and constructing a box line graph according to the environmental risk degree of each monitoring point at each historical moment; and acquiring an environmental risk degree upper limit according to the box diagram, and if the environmental risk degree of any monitoring point in the apple warehouse is greater than the environmental risk degree upper limit at the current time, carrying out abnormal early warning on the environment in the apple warehouse.

Description

Apple storage environment abnormal data detection method Technical Field The invention relates to the technical field of data processing, in particular to a method for detecting abnormal data of an apple warehouse environment. Background Apple storage is an important link for guaranteeing the quality of apples and prolonging the supply period. In the storage process, apples still keep vital activities such as respiration, transpiration and the like, namely, the quality of apples is highly sensitive to the conditions such as temperature, humidity, gas concentration and the like of the storage environment, so that the stability and the proper degree of the storage environment directly determine the fresh-keeping effect of apples in the storage process. Therefore, the monitoring of the apple storage environment in real time and the timely discovery and adjustment of the abnormal environment are key tasks in apple storage management. The existing apple storage environment abnormality detection method mainly sets a fixed threshold value for each environmental factor, and carries out abnormality alarm when the sensor monitors that the environmental data exceeds the threshold value. However, as the local environment is also changed due to the life activities of apples, and the warehouse is opened and closed, and the fluctuation and change of environment monitoring data can be caused by different stacking modes, the normal data fluctuation is similar to the real environment abnormality which does not accord with apple storage, such as ventilation failure of the warehouse, environment regulating system failure, apple spoilage process and the like, the conventional fixed threshold detection method is difficult to distinguish the normal fluctuation and the real environment abnormality, so that the false alarm rate of the detection system is high, and the management efficiency of the apple warehouse environment is affected. Therefore, how to improve the accuracy of abnormality detection for the apple storage environment is a problem to be solved. Disclosure of Invention In view of the above, the embodiment of the invention provides a method for detecting abnormal data of an apple storage environment, so as to solve the problem of how to improve the accuracy of detecting the abnormal data of the apple storage environment. The embodiment of the invention provides a method for detecting abnormal data of an apple warehouse environment, which comprises the following steps: uniformly arranging monitoring points in an apple warehouse according to a grid shape, collecting real-time monitoring data of each environmental index at the current time and historical monitoring data of a preset number of historical time at each monitoring point, wherein the real-time monitoring data and the historical monitoring data are normalized data; aiming at any environmental index, according to the data change similarity of the historical monitoring data of any environmental index at any monitoring point and the difference between the real-time monitoring data and the historical monitoring data of any environmental index at any monitoring point, acquiring the initial abnormality degree of any environmental index at any monitoring point at the current moment; acquiring the initial abnormality degree of any environmental index at each monitoring point at the current moment, and correcting the initial abnormality degree of any environmental index at each monitoring point according to the position of the grid of each monitoring point and the distribution condition of the initial abnormality degree of any environmental index at each monitoring point to obtain the final abnormality degree of any environmental index at each monitoring point at the current moment; Acquiring the final abnormality degree of each environmental index at each monitoring point at the current moment, and monitoring and early warning the environment in the apple warehouse according to the final abnormality degree of each environmental index at each monitoring point at the current moment. Preferably, the obtaining the initial abnormality degree of the any environmental index at the any monitoring point at the current time according to the similarity of the data change of the historical monitoring data of the any environmental index at any monitoring point and the difference between the real-time monitoring data and the historical monitoring data of the any environmental index at the any monitoring point comprises: Forming a data sequence by all the historical monitoring data of any environmental index at any monitoring point according to time sequence, constructing a graph according to the data sequence, wherein the abscissa of the graph represents historical time, the ordinate represents the historical monitoring data, acquiring the tangential slope of the historical monitoring data at each historical time according to the graph, recording the tangential slope as a local chang