Trolley detection method and device, electronic equipment and storage medium

Abstract

This invention relates to the field of manufacturing technology, providing a trolley inspection method, apparatus, electronic device, and storage medium. The trolley inspection method includes: acquiring original feature information of the trolley to be inspected, including parameter information collected by the trolley during the inspection of target products; determining target feature information based on the original feature information, including feature information remaining after anomaly processing of the original feature information; and obtaining a target inspection result for the trolley to be inspected based on the target feature information and a preset detection model. Using this invention, the purpose of timely and intelligent detection of trolley anomalies can be achieved, saving time and effort while being highly efficient and accurate.

Description

Technical Field

[0001] This invention relates to the field of manufacturing technology, and in particular to a method, apparatus, electronic device, and storage medium for trolley inspection. Background Technology

[0002] In the air conditioner production process, there is a testing room at the end of the production line. This room contains trolleys used to inspect the air conditioners for problems. These trolleys operate continuously throughout the production process. However, due to environmental factors and equipment wear and tear, the inspection results from these trolleys are prone to significant deviations. Therefore, knowing how to detect abnormalities in these trolleys is crucial.

[0003] In related technologies, the detection method for trolleys usually involves workers installing a normal air conditioner on multiple trolleys and manually collecting parameters such as power and pressure during the inspection of the air conditioner on each trolley. Then, the average value of the corresponding parameters is calculated, and finally, the trolley with the average value deviating significantly from the preset average value is identified as an abnormal trolley.

[0004] However, when workers inspect the trolley on-site, they need to manually install the air conditioner, manually collect data during operation, and manually remove the air conditioner after operation. The entire inspection process is not only overly reliant on human factors, but also time-consuming and labor-intensive, resulting in low efficiency and low accuracy of the inspection trolley. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in related technologies. To this end, this invention proposes a trolley detection method, which achieves timely and intelligent detection of whether a trolley has abnormalities through automatic information acquisition, anomaly handling, and model detection. This method is not only time-saving and labor-saving, but also highly efficient and accurate.

[0006] The present invention also proposes a trolley detection device.

[0007] The present invention also proposes an electronic device.

[0008] The present invention also proposes a non-transitory computer-readable storage medium.

[0009] The present invention also proposes a computer program product.

[0010] A trolley detection method according to a first aspect of the present invention includes:

[0011] Obtain the original feature information of the trolley to be inspected, which includes parameter information collected by the trolley to be inspected during the inspection of the target product;

[0012] Based on the original feature information, target feature information is determined, wherein the target feature information includes the feature information remaining after anomaly processing of the original feature information;

[0013] Based on the target feature information and the preset detection model, the target detection result for the vehicle to be detected is obtained.

[0014] According to the trolley inspection method of the present invention, the original feature information of the trolley to be inspected is first obtained. Since the original feature information includes parameter information collected by the trolley during the inspection of the target product, the target feature information is determined from the original feature information by performing anomaly processing. Then, based on the target feature information and a preset detection model, the target detection result for the trolley to be inspected is obtained. In this way, by automatically acquiring information, handling anomalies, and performing model detection, the purpose of timely and intelligent detection of whether the trolley has anomalies is achieved. This method is not only time-saving and labor-saving, but also highly efficient and accurate. At the same time, the combination of model learning and detection technology can further ensure the accuracy of the detection results, thereby effectively improving the accuracy and reliability of trolley inspection.

[0015] According to an embodiment of the present invention, determining the target feature information based on the original feature information includes:

[0016] The column data of the column field containing the parameter information in the original feature information is transformed into a row to determine the feature information to be processed of the trolley to be detected;

[0017] Based on preset anomaly handling rules, the feature information to be processed is subjected to anomaly handling;

[0018] The target feature information is determined based on the feature information obtained from anomaly processing;

[0019] The preset anomaly handling rules include, for the feature information to be processed, removing column data whose column field missing rate exceeds the column missing rate threshold, filling row data whose row field missing rate is lower than the row missing rate threshold, removing column data that does not meet the preset distribution pattern, removing column data that does not meet the preset temperature correlation, removing column data that does not meet the preset temperature range, and removing column data that does not meet the preset power correlation.

[0020] According to an embodiment of the present invention, the anomaly processing of the feature information to be processed further includes:

[0021] Obtain the target quantity of target products belonging to the same preset equipment part code, where the preset equipment part code is the hardware parameter model of the same target product;

[0022] If the target quantity is determined to be less than a preset quantity threshold, the row of data containing the preset equipment part code corresponding to the target quantity in the feature information to be processed is removed.

[0023] According to an embodiment of the present invention, determining the target feature information based on the feature information obtained from anomaly processing includes:

[0024] The feature information obtained from anomaly processing is subjected to feature amplification processing to determine multiple feature information after the amplification processing;

[0025] Based on the pre-defined relevance and quantity of feature information, the multiple feature information is filtered to determine the target feature information.

[0026] According to an embodiment of the present invention, the preset detection model includes different preset detection sub-models, and obtaining the target detection result for the vehicle to be detected based on the target feature information and the preset detection model includes:

[0027] The target feature information is input into the preset detection model to obtain a preset number of detection results of the trolley to be detected under different preset equipment part codes and different preset detection sub-models.

[0028] Based on the abnormal detection results among the preset number of detection results, the target detection result for the vehicle to be inspected is obtained.

[0029] According to an embodiment of the present invention, obtaining the target detection result for the vehicle to be inspected based on the abnormal detection result among the preset number of detection results includes:

[0030] Determine the percentage of abnormal detection results among the preset number of detection results;

[0031] Once the percentage is determined to exceed a first preset percentage, a target detection result is obtained indicating that the vehicle to be detected is an abnormal vehicle.

[0032] Determine that the percentage is between the second preset percentage and the first preset percentage, and obtain the target detection result that the trolley to be detected is a risk trolley;

[0033] If the percentage is determined to be lower than the second preset percentage, the target detection result of the trolley to be detected being a normal trolley is obtained.

[0034] According to one embodiment of the present invention, after obtaining the target detection result that the trolley to be detected is a risk trolley, the method further includes:

[0035] Obtain the target coding information of the risk vehicle;

[0036] Based on the target encoding information, a warning message is sent to the user terminal. The warning message is used to remind the user terminal's corresponding review personnel to conduct an on-site assessment of the risk vehicle.

[0037] According to an embodiment of the present invention, the training method of the preset detection model includes:

[0038] Obtain multiple different initial detection sub-models for anomaly detection of the trolley to be inspected;

[0039] Based on the target feature information, each initial detection sub-model is trained a preset number of times to determine multiple intermediate detection results of multiple intermediate detection sub-models after each training.

[0040] The multiple intermediate detection results are sent to the user terminal, and the verification results of the multiple intermediate detection results are received from the user terminal.

[0041] Based on the verification results and the intermediate detection sub-model after each model parameter update, the preset detection sub-model and the preset detection model corresponding to the preset detection sub-model are determined.

[0042] A trolley detection device according to a second aspect of the present invention includes:

[0043] The acquisition module is used to acquire the original feature information of the trolley to be inspected, which includes the parameter information collected by the trolley to be inspected during the inspection of the target product.

[0044] The determination module is used to determine target feature information based on the original feature information, wherein the target feature information includes the feature information remaining after anomaly processing of the original feature information;

[0045] The detection module is used to obtain the target detection result for the vehicle to be detected based on the target feature information and the preset detection model.

[0046] According to an embodiment of the present invention, the trolley inspection device first acquires the original feature information of the trolley to be inspected. Since the original feature information includes parameter information collected by the trolley during the inspection of the target product, the target feature information is determined from the original feature information by performing anomaly processing. Then, based on the target feature information and a preset detection model, the target detection result for the trolley to be inspected is obtained. In this way, by automatically acquiring information, handling anomalies, and performing model detection, the device achieves the purpose of timely and intelligent detection of whether the trolley has anomalies. This not only saves time and effort but is also highly efficient and accurate. At the same time, the combination of model learning and detection technology can further ensure the accuracy of the detection results, thereby effectively improving the accuracy and reliability of trolley inspection.

[0047] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: First, the original feature information of the trolley to be inspected is obtained. Since the original feature information includes parameter information collected by the trolley to be inspected during the inspection of the target product, the target feature information is determined from the original feature information by performing anomaly processing on the original feature information. Then, based on the target feature information and the preset detection model, the target detection result for the trolley to be inspected is obtained. In this way, by automatically acquiring information, handling anomalies and detecting models, the purpose of timely and intelligent detection of whether the trolley has anomalies is achieved. This not only saves time and effort, but is also highly efficient and accurate. At the same time, the combination of model learning and detection technology can further ensure the accuracy of the detection results, thereby effectively improving the accuracy and reliability of trolley detection.

[0048] Furthermore, the process first involves converting the column data of the parameter feature field in the original feature information into a row to determine the feature information to be processed for the trolley to be inspected. Then, based on preset anomaly handling rules, the feature information to be processed is anomaly-processed. Finally, the target feature information is determined based on the feature information obtained from the anomaly processing. By combining column conversion and anomaly data processing techniques, not only is the accuracy and relevance of anomaly data processing improved, but the foundation for the accuracy of subsequent trolley anomaly detection is also laid.

[0049] Furthermore, by determining that the number of target products with the same preset equipment part code in the feature information to be processed is less than a preset quantity threshold, the data in the row containing the relevant preset equipment part code is removed from the data to be processed, thereby improving the flexibility and comprehensiveness of processing abnormal data.

[0050] Furthermore, by first performing feature amplification on the feature information obtained from anomaly processing and then conducting correlation analysis, target feature information that meets the preset requirements for correlation and quantity of feature information is determined. This not only improves the richness of the target feature information but also provides sufficient basis for subsequent model learning.

[0051] Furthermore, by judging the relationship between the proportion of abnormal detection results among a preset number of detection results and the first and second preset proportions, the trolley to be tested is determined to be an abnormal trolley, a normal trolley, or a risk trolley. This, combined with a comprehensive voting system, effectively improves the efficiency and accuracy of abnormal detection.

[0052] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart illustrating the trolley detection method provided in an embodiment of the present invention;

[0055] Figure 2 This is a schematic diagram of the overall process of the trolley detection method provided in the embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram of the structure of the trolley detection device provided in an embodiment of the present invention;

[0057] Figure 4 This is a schematic diagram of the physical structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0059] In the air conditioner production process, there is an operation room at the end of the production line. This room is equipped with trolleys for inspecting air conditioners to check for problems. These trolleys operate continuously throughout the production process. However, due to environmental factors and equipment wear and tear, the trolley inspections are prone to significant deviations. Current trolley inspection methods only involve workers passively checking a trolley for abnormalities after multiple error reports are detected over a period of time. Furthermore, the worker's verification method involves installing a verified, working air conditioner on approximately 10 trolleys, measuring parameters such as power and pressure, calculating the average of these parameters, and determining the deviation rate for each trolley to decide whether each trolley needs to be taken out of service or repaired.

[0060] Clearly, existing trolley inspection methods are not timely in detecting trolley problems, and by the time problems are discovered, they are already quite serious. Furthermore, manual on-site inspection and verification of trolley problems are required. In other words, inspecting a trolley requires the installation, operation, and dismantling of household appliances such as air conditioners, a complete process that takes about 7 minutes. Detecting a malfunctioning trolley can take more than an hour, which is time-consuming, labor-intensive, and inefficient.

[0061] Based on this, the present invention provides a trolley detection method, apparatus, electronic device, and storage medium, wherein the trolley detection method is executed by a terminal device, which can be a personal computer (PC), portable device, laptop computer, smartphone, tablet computer, portable wearable device, or other electronic device. The present invention does not limit the specific form of the terminal device. It is understood that the trolley detection method can also be executed by a server. The following describes the process in conjunction with... Figures 1-4 This invention describes the trolley detection method, apparatus, electronic device, and storage medium provided by the present invention, and the following method embodiments are illustrated using a terminal device as the execution subject.

[0062] Figure 1 This is a flowchart illustrating the trolley detection method provided by the present invention, as shown below. Figure 1 As shown, the trolley testing method includes the following steps:

[0063] Step 110: Obtain the original feature information of the trolley to be inspected. The original feature information includes the parameter information collected by the trolley to be inspected during the inspection of the target product.

[0064] The inspection trolley can be a trolley used to inspect target products produced on the production line, and the number of inspection trolleys can be one or more. No specific limitation is made here. The target product can be an air conditioner, and the number of target products can be one or more. No specific limitation is made here either. Furthermore, the original feature information exists in the table in the form of row fields. Row fields can include, but are not limited to, temperature, current, voltage, power, etc. The corresponding column fields can include, but are not limited to, the target product's identity document (id), the target code information of the inspection trolley, and the preset equipment part code. The preset equipment part code is the hardware parameter model of the same target product.

[0065] Specifically, the original characteristic information of the trolley to be inspected can be obtained by the terminal equipment from the Manufacturing Execution System (MES system). The MES system records the status information such as power, pressure, current, and temperature collected by the trolley to be inspected during the inspection of the target product in the operation room, and all of them are accumulated and summarized in a table into a single field. When it is necessary to process the field read from the MES system, the field can be cleaned to obtain the original characteristic information of the trolley to be inspected. For example, this field can include four columns of data: a preset equipment part code, the ID of the target product, the target code information of the trolley to be tested, and status information. The status information includes two columns: the detection item and the corresponding value of the detection item. The two columns for the detection item and the corresponding value of the detection item can include, but are not limited to, 20 rows of data. The 20 rows of data can include the three-phase current, three-phase voltage, inlet temperature, outlet temperature, cooling power, and cooling pressure for each cooling step, as well as the three-phase current, three-phase voltage, inlet temperature, outlet temperature, heating power, and cooling pressure for each heating step. By parsing this field, the data in this field is processed into the temperature, power, pressure, voltage, and current corresponding to a preset equipment part code, a target product ID, and the target code information of the trolley to be tested in a certain cooling step or a certain heating step, thereby determining the original characteristic information of the trolley to be tested.

[0066] It is understandable that the trolley to be tested can be one, multiple, or all of the trolleys in the operating room; the operating room can be a closed room where the trolley to be tested inspects the target product.

[0067] Step 120: Based on the original feature information, determine the target feature information, which includes the feature information remaining after anomaly processing of the original feature information.

[0068] Specifically, the terminal equipment analyzes the original feature information of the trolley under test, identifying which data is usable and which is not. For example, it analyzes usable and abnormal data from the data collected in steps 1, 2, 3, 4, 5, and 6 of the cooling process. The key to analyzing abnormal data lies in identifying data errors caused by mistakes in the data acquisition process. Based on this, for the abnormal data screening in the data acquisition process, corresponding reference value ranges can be pre-set and mapping relationships can be established for the three-phase current, three-phase voltage, inlet temperature, outlet temperature, power, and air pressure collected in each step. Then, based on the mapping relationship, data in the original feature information that does not match the mapping relationship is identified as abnormal data. That is, abnormal data can include, but is not limited to, row data and / or column data in the original feature information that do not match the mapping relationship. Then, abnormal data identified in the original feature information is processed to determine the target feature information.

[0069] Step 130: Based on the target feature information and the preset detection model, obtain the target detection results for the trolley to be detected.

[0070] Specifically, considering the significant impact of abnormal data on subsequent model processing, and to improve the detection accuracy of the vehicle under inspection, the preset detection model can include at least two anomaly detection algorithms. Each anomaly detection algorithm can determine whether the vehicle under inspection is abnormal based on target feature information, thereby obtaining multiple detection results. These multiple detection results are then aggregated and analyzed to determine the target detection result. For example, if the number of detection results indicating anomalies is significantly greater than the number of detection results indicating normal operation, the vehicle under inspection can be determined to be an abnormal vehicle; conversely, if the number of detection results indicating anomalies is significantly less than the number of detection results indicating normal operation, the vehicle under inspection can be determined to be a normal vehicle.

[0071] The trolley inspection method provided in this invention first acquires the original feature information of the trolley to be inspected. Since the original feature information includes parameter information collected by the trolley during the inspection of the target product, the target feature information is determined from the original feature information by performing anomaly processing. Then, based on the target feature information and a preset detection model, the target detection result for the trolley to be inspected is obtained. In this way, by automatically acquiring information, handling anomalies, and detecting using a model, the method achieves the purpose of timely and intelligent detection of whether the trolley has anomalies. This method is not only time-saving and labor-saving, but also highly efficient and accurate. At the same time, the combination of model learning and detection technology can further ensure the accuracy of the detection results, thereby effectively improving the accuracy and reliability of trolley inspection.

[0072] It is understandable that, considering the inevitable existence of abnormal data during data collection by the MES system, which may negatively impact subsequent detection—for example, during the multiple steps of air conditioning testing in the operation room, such as cooling or heating, the heating and cooling tests are usually performed at different frequencies, leading to uneven temperature distribution within the operation room and affecting the accuracy of the collected data—abnormal data processing rules can be pre-set to handle these anomalies in the original feature information. Based on this, the specific implementation process of step 120 may include:

[0073] First, the column data of the column field containing the parameter information in the original feature information is transformed into a row to determine the feature information to be processed of the trolley to be detected; then, based on the preset anomaly handling rules, anomaly handling is performed on the feature information to be processed; finally, based on the feature information obtained from the anomaly handling, the target feature information is determined.

[0074] The preset anomaly handling rules include, for the feature information to be processed, removing column data whose column field missing rate exceeds the column missing rate threshold, filling row data whose row field missing rate is lower than the row missing rate threshold, removing column data that does not meet the preset distribution pattern, removing column data that does not meet the preset temperature correlation, removing column data that does not meet the preset temperature range, and removing column data that does not meet the preset power correlation.

[0075] Specifically, since the original feature information includes 20 rows of data, consisting of 5 columns containing the preset equipment part code, the target product ID, the target code information of the trolley to be inspected, the inspection item and the corresponding inspection item value, and the last two columns containing the inspection item and the corresponding inspection item value, a column-to-row operation can be performed on the last two columns of data. This means expanding the last two columns into a single row to obtain the result of the column-to-row operation. This result is then used as the feature information to be processed for the trolley to be inspected. This feature information is presented as column data. The data exists in the form of a table, and under three dimensions—preset equipment part code, target product identification number (id), and target code information of the trolley to be tested—it includes three-phase current, three-phase voltage, inlet temperature, outlet temperature, cooling power, and cooling pressure for each cooling step, as well as multiple columns of data for each heating step, such as three-phase current, three-phase voltage, inlet temperature, outlet temperature, heating power, and cooling pressure. Each column of data is a feature information; thus, it intuitively displays the values ​​of various parameters collected when different trolleys test different target products.

[0076] At this point, based on preset anomaly handling rules, for the feature information to be processed, column data with a column field missing rate exceeding the column missing rate threshold are removed, for example, column data with a column missing rate exceeding 80% are removed; row data with a row field missing rate below the row missing rate threshold are filled, for example, if 5 rows of data are empty and 15 rows are not empty out of 20 rows of data corresponding to a preset equipment part code, the average value of each column in the 15 rows of data can be used to fill the corresponding column fields of the 5 rows of data; column data that does not meet the preset distribution rules are removed, for example, based on the collected three-phase current, three-phase voltage, temperature, power and air pressure, violin plots are created, and the upper and lower boundaries of the concentrated distribution area in the violin plot are selected as the upper and lower boundaries of the screening value, and data distributed outside the upper and lower boundaries of the screening value are removed; column data that does not meet the preset temperature correlation are removed, for example, based on the fact that the temperature at the air inlet is higher than the temperature at the air outlet when the air conditioner is cooling and lower than the temperature at the air outlet when the air conditioner is heating, the covariance matrix can be determined for the temperature at the air inlet and the temperature at the air outlet collected in the cooling and heating steps, and based on... The correlation coefficient of the inlet and outlet air temperature difference is calculated using the covariance matrix. If the difference between the current C, one of the three-phase currents in the cooling process, and the corresponding correlation coefficient of the inlet and outlet air temperature difference in the cooling process is small enough, the column data corresponding to the current C can be removed. Column data that does not meet the preset temperature range is also removed. For example, taking the temperature of the air conditioner in the last step of cooling as the stable cooling temperature, the average of the multiple stable cooling temperatures corresponding to multiple air conditioners in the operating room is taken as the average cooling temperature. Similarly, taking the temperature of the air conditioner in the last step of heating as the stable heating temperature, the average of the multiple stable heating temperatures corresponding to multiple air conditioners in the operating room is taken as the average heating temperature. Then, column data whose temperature values ​​are outside the range of ±16% of the average heating temperature or average cooling temperature of the air conditioner with the same preset equipment part code are removed. Column data that does not meet the preset power correlation is also removed. For example, the ratio of power to temperature is calculated for a row of data to be filtered. If the ratio is outside the range of ±8% of the preset average ratio of the air conditioner with the same preset equipment part code, the column data corresponding to the power of that ratio is removed.

[0077] At this point, for the feature information obtained from the anomaly processing, it is determined that the number of its feature information meets the preset feature quantity requirements and relevance requirements. At this point, the feature information obtained from the anomaly processing can be determined as the target feature information. That is, the target feature information can be multiple features that meet the preset feature quantity requirements and relevance requirements under the preset equipment part code, the target product identification number id, and the target code information of the trolley to be tested. For example, it can be about 30 features.

[0078] It is understandable that, for multiple air conditioners in the operating room, in addition to the uneven temperature distribution affecting the various parameters during the testing process, the following situations can also cause temperature changes: when the test trolleys are changed across sheet metal sections or from small to large machines, the heating power and pressure may increase significantly; when the air conditioner production is unstable, the temperature may drop; there are changes in power, pressure, and temperature when the trolley is powered off and then powered on again; and the temperature in the operating room increases when all trolleys are in cooling mode for 15 minutes. Therefore, it is necessary to perform temperature compensation on the original feature information, that is, to remove column data that does not meet the preset temperature range and column data that does not meet the preset power correlation.

[0079] The trolley detection method provided in this invention first determines the feature information to be processed of the trolley to be detected by performing a column-to-row operation on the column data of the column field containing the parameter features in the original feature information; then, it performs anomaly processing on the feature information to be processed based on preset anomaly processing rules; finally, it determines the target feature information based on the feature information obtained from the anomaly processing. By combining column-to-row and anomaly data processing technologies, it not only improves the accuracy and pertinence of anomaly data processing, but also lays the foundation for the accuracy of subsequent trolley anomaly detection.

[0080] Understandably, considering that when the number of trolleys in the operation room is less than the number of air conditioners produced, a small number of air conditioners may not be enough to cover multiple trolleys, leading to detection problems. In this case, it is advisable to delete the row data containing the preset equipment part code corresponding to the production of these small number of air conditioners. Based on this, anomaly handling for the feature information to be processed also includes:

[0081] First, obtain the target quantity of target products belonging to the same preset equipment part code; then, if the target quantity is less than the preset quantity threshold, remove the data in the row containing the preset equipment part code corresponding to the target quantity in the feature information to be processed.

[0082] Specifically, to ensure the accuracy and fairness of trolley inspection, the target quantity of target products belonging to the same preset equipment part code can be obtained based on the feature information to be processed, and it can be determined whether the number of trolleys to be inspected in the operation room is sufficient to cover the target quantity. Here, three times the number of trolleys on the production line for the air conditioner production target quantity can be set as the minimum number of air conditioners for the preset equipment part code for judgment. If there are 20 trolleys to be inspected and the number of air conditioners belonging to the same preset equipment part code is less than 60, then the data in the row of the preset equipment part code in the feature information to be processed will be removed.

[0083] The trolley detection method provided in this embodiment of the invention improves the flexibility and comprehensiveness of processing abnormal data by removing the row of data containing the relevant preset equipment part codes from the data to be processed by determining that the number of target products with the same preset equipment part code in the feature information to be processed is less than a preset number threshold.

[0084] It is understandable that when the number of features and / or the correlation of features obtained after anomaly processing of the original feature information do not meet the preset requirements, the target feature information can be determined by first amplifying features and then performing correlation analysis. Based on this, the process of determining the target feature information based on the feature information obtained from anomaly processing can include:

[0085] First, feature amplification processing is performed on the feature information obtained from anomaly processing to determine multiple feature information after amplification; then, based on the preset feature information correlation and preset feature information quantity, multiple feature information are filtered to determine the target feature information.

[0086] Specifically, when the number of features obtained from anomaly processing does not meet the preset number of features and / or the preset correlation of features, feature amplification processing can be performed on the feature information obtained from anomaly processing. For example, for the cooling power of each cooling step, multiple features can be determined by calculating the mean, median, range, and standard deviation. Further correlation analysis can be performed on these multiple features. Here, features with low correlation can be filtered out from multiple features by determining the covariance matrix, and sorted by the magnitude of the correlation coefficient. Finally, the features that meet the preset feature correlation and preset number of features under the preset equipment part code, the identification number id of the target product, and the target code information of the trolley to be tested are determined, which is to say, the target feature information is determined.

[0087] The trolley detection method provided in this invention determines target feature information that meets the preset requirements for correlation and quantity of feature information by first performing feature amplification processing on the feature information obtained from anomaly processing and then performing correlation analysis. This not only improves the richness of the target feature information but also provides sufficient basis for subsequent model learning.

[0088] Understandably, to improve the accuracy and efficiency of trolley inspection, inspection can be performed on trolleys of air conditioners belonging to the same preset equipment part code, and the final inspection result can be determined by using different inspection results from multiple algorithms. Based on this, when the preset inspection model includes different preset inspection sub-models, the specific implementation process of step 130 may include:

[0089] First, the target feature information is input into the preset detection model to obtain a preset number of detection results of the trolley to be inspected under different preset equipment part codes and different preset detection sub-models. Then, based on the abnormal detection results in the preset number of detection results, the target detection results for the trolley to be inspected are obtained.

[0090] Specifically, for cases where the preset detection model includes different preset detection sub-models, target feature information can be input into the preset detection model. The different preset detection sub-models then perform the detection. That is, each preset detection sub-model, for air conditioners belonging to the same preset equipment part code, determines whether the vehicle under test is abnormal based on corresponding data such as power, current, voltage, inlet temperature, and inlet / outlet temperature. This yields the detection results for each preset detection sub-model, which determines whether the vehicle under test is abnormal for each preset equipment part code. A preset number of detection results is then determined, based on the total number of preset equipment part codes and the total number of preset detection sub-models. Then, based on the abnormal detection results among the preset number of detection results, the target detection result for the vehicle under test is obtained, thereby determining whether the vehicle under test is currently in a normal, abnormal, or risky state.

[0091] It is understandable that each preset detection sub-model can contain an anomaly detection algorithm, and each anomaly detection algorithm can be determined by training a large number of existing anomaly detection algorithms.

[0092] The trolley detection method provided in this invention obtains a preset number of detection results for the trolley under different preset equipment component codes and different preset detection sub-models, and obtains the target detection result for the trolley based on the anomaly detection results among the preset number of detection results. By obtaining detection results through different preset equipment component codes and different anomaly detection algorithms, the accuracy and reliability of trolley anomaly detection are significantly improved.

[0093] Understandably, considering that the comprehensive voting mechanism of the model can improve the stability and fairness of the final detection results, the target detection result can be obtained by setting a comprehensive voting mechanism. Based on this, the target detection result for the vehicle to be detected is obtained based on the abnormal detection results among a preset number of detection results. The implementation process includes:

[0094] First, determine the percentage of abnormal detection results among a preset number of detection results; then, if the percentage exceeds the first preset percentage, obtain the target detection result that the vehicle to be detected is an abnormal vehicle; if the percentage is between the second preset percentage and the first preset percentage, obtain the target detection result that the vehicle to be detected is a risk vehicle; if the percentage is lower than the second preset percentage, obtain the target detection result that the vehicle to be detected is a normal vehicle.

[0095] Specifically, since each preset detection sub-model can determine whether the air conditioner to be tested, which has been inspected for each preset equipment part code, is abnormal, the corresponding detection results are the voting results for abnormal, risky, and normal. Therefore, it can be assumed that the air conditioner to be tested will receive K votes, where K can be the product of the total number of preset equipment part codes and the total number of preset detection sub-models. At this time, the proportion of abnormal detection results is determined from the preset number of detection results obtained from all preset detection sub-models, which is the proportion of abnormal detection results in the K votes. The proportion reflects the abnormal vote rate of the air conditioner to be tested under all preset equipment part codes and all preset detection sub-models. If the proportion exceeds the first preset proportion, the air conditioner to be tested is determined to be an abnormal air conditioner and needs to be taken out of service and repaired; if the proportion is between the second preset proportion and the first preset proportion, the air conditioner to be tested is determined to be a risky air conditioner and needs to be monitored; if the proportion is lower than the second preset proportion, the air conditioner to be tested is determined to be a normal air conditioner and can be used normally in the future. Furthermore, the first preset percentage can be 20%, and the second preset percentage can be 10%. If there are 20 preset equipment part codes and 16 preset detection sub-models, the vehicle to be inspected will receive 320 votes. If the abnormal vote rate exceeds 320*20%, that is, when the abnormal vote rate exceeds 64 votes, the vehicle to be inspected can be determined to be an abnormal vehicle.

[0096] The trolley detection method provided in this invention determines whether a trolley to be detected is an abnormal trolley, a normal trolley, or a risk trolley by judging the relationship between the proportion of abnormal detection results among a preset number of detection results and the first preset proportion and the second preset proportion. This, combined with a comprehensive voting system, effectively improves the efficiency and accuracy of abnormal detection.

[0097] It is understandable that for risky trolleys with potential abnormalities, the trolley's abnormality can be reassessed based on the verification results of on-site personnel. Therefore, after obtaining the target detection result that the trolley to be inspected is a risky trolley, the trolley inspection method provided in this embodiment of the invention further includes:

[0098] First, obtain the target code information of the risk vehicle; then, based on the target code information of the risk vehicle, send a warning message to the user terminal. The warning message is used to remind the corresponding review personnel of the user terminal to conduct an on-site assessment of the risk vehicle.

[0099] Specifically, for the target inspection result of obtaining the inspection result of the trolley to be inspected as a risk trolley, the target code information of the risk trolley can be obtained again. The target code information is used to locate the risk trolley in the operation room. Based on the target code information of the risk trolley, an early warning information is sent to the user terminal to remind the corresponding review personnel of the target terminal to conduct an on-site assessment of the risk trolley in a timely manner. If the on-site assessment result is unusable, it is determined that the risk trolley should be stopped and repaired. If the on-site assessment result is usable, it is determined that the risk trolley can continue to be used for subsequent air conditioning inspection.

[0100] The trolley detection method provided in this invention further improves the flexibility and reliability of trolley anomaly detection by combining the risk assessment of risky trolleys with the review of personnel.

[0101] Understandably, to improve the precision and accuracy of the pre-defined detection model, a pre-defined detection sub-model can be determined by training a large number of anomaly detection algorithms using target feature information. Based on this, the training methods for the pre-defined detection model include:

[0102] First, multiple different initial detection sub-models are obtained for anomaly detection of the trolley to be inspected. Then, each initial detection sub-model is trained a preset number of times based on target feature information, and multiple intermediate detection results of multiple intermediate detection sub-models are determined after each training. Next, the multiple intermediate detection results are sent to the user terminal, and the user terminal provides feedback on the verification results of the multiple intermediate detection results. Finally, based on the verification results and the intermediate detection sub-models with updated model parameters, a preset detection sub-model and the preset detection model corresponding to the preset detection sub-model are determined.

[0103] Specifically, obtaining multiple different initial detection sub-models for anomaly detection on the vehicle to be inspected can be based on existing anomaly detection algorithms. These existing anomaly detection algorithms can include, but are not limited to, statistical 3Seg code anomaly detection algorithm, Local Outlier Factor (LOF) algorithm, k-Nearest Neighbor (kNN) algorithm, Average k-Nearest Neighbor (AvgkNN) algorithm, Clustering-Based Local Outlier Factor (CBLOF) algorithm, One-Class Support Vector Machines (OCSVM) algorithm, Local Correlation Integration (LOCI) algorithm, Principal Component Analysis (PCA) algorithm, Minimum Covariance Determinant (MCD) algorithm, Feature Bagging algorithm, and Angle-Based Outlier Detection. Algorithms include Outlier Detection (ABOD), Isolation Forest, Histogram-based Outlier Score (HBOS), Stochastic Outlier Selection (SOS), Fully Connected Autoencoder, Average Maximum (AOM), Mean Maximization (MOA), Single-Objective Generative Adversarial Active Learning (SO-GAAL), Multiple-Objective Generative Adversarial Active Learning (MO-GAAL), XGBOD (Extremum Boosting Outlier Detection), and Locally Selective Combination in Parallel Outlier Ensembles (LSCP). No specific limitations are specified here.

[0104] Further, based on the target feature information, each initial detection sub-model is trained a preset number of times. After each training, multiple intermediate detection results of multiple intermediate detection sub-models are determined, and the intermediate detection results obtained by each intermediate detection sub-model are sent to the user terminal. The user terminal corresponds to the reviewer to review the intermediate detection results, and the user terminal receives the review results of the intermediate detection results. The purpose of the review is to determine the accuracy and stability of each anomaly detection algorithm, and the model parameters of each intermediate detection sub-model are automatically updated after each training. Then, from the multiple intermediate detection sub-models after multiple trainings, a subset of detection models that meet preset conditions are selected to determine the preset detection sub-models. The preset conditions are that the accuracy of the multiple intermediate detection results after multiple trainings meets both the preset accuracy requirement and the stability requirement. Each preset detection sub-model is an intermediate detection sub-model that meets the preset conditions and has undergone model parameter updates. Finally, a preset detection model including different preset detection sub-models is determined. The preset detection model may include preset detection sub-models that have been trained multiple times and updated with model parameters for 16 anomaly detection algorithms, including the statistical 3Seg code anomaly detection algorithm, IsolationForest algorithm, kNN algorithm, LOF algorithm, CBLOF algorithm, OCSVM algorithm, LOCI algorithm, PCA algorithm, MCD algorithm, Feature Bagging algorithm, ABOD algorithm, Iforest algorithm, HBOS algorithm, SOS algorithm, LSCP algorithm, and COPOD algorithm.

[0105] Understandably, since the raw feature information acquired from the trolleys in the operation room is only valid for the current week—that is, raw feature information is acquired once a week from Monday to Saturday for detection, and abnormal trolleys are repaired on Sunday—the raw feature information acquired the following week cannot reflect the situation of the previous week, resulting in a relatively small amount of data for model training. Based on this, the current preset detection model can be optimized by combining the raw feature information acquired each time with the previous target detection results, thereby making the target detection results more accurate.

[0106] The trolley detection method provided in this invention improves the accuracy and reliability of determining the preset detection model by using target feature information to train a large number of anomaly detection algorithms multiple times.

[0107] Reference Figure 2The flowchart shown illustrates the overall process of trolley detection. The process can be understood as follows: Data input values ​​from the MES system are transferred to a large data warehouse using Extract-Transform-Load (ETL) technology to obtain the required raw feature information. Then, through data cleaning, column-to-row transformation, anomaly handling, and feature amplification, the target feature information is determined. Anomaly detection is then performed based on 16 pre-defined detection models and a comprehensive voting mechanism to determine the target detection result. Subsequently, the pre-defined detection models are iteratively optimized in their application, facilitating subsequent scheduling system calls and display of the pre-defined result table. The predicted result table can be displayed on a dashboard page. The results are verified and feedback is provided through on-site business operations. Finally, the model optimization is achieved by combining the verification results evaluated by reviewers. Specific anomaly detection processes are detailed in the aforementioned method description and will not be repeated here.

[0108] By using the trolley detection method, an automatic early warning can be issued weekly for trolleys that show abnormalities. This will then activate the EAM fault repair module and prompt relevant personnel to perform predictive maintenance, thereby ensuring the reliability, effectiveness, accuracy, and stability of the air conditioning testing in the operation room.

[0109] The trolley detection device provided by the present invention is described below. The trolley detection device described below can be referred to in correspondence with the trolley detection method described above.

[0110] Reference Figure 3 As shown, this is a trolley detection device provided in an embodiment of the present invention. Figure 3 The trolley testing device 300 includes:

[0111] The acquisition module 310 is used to acquire the original feature information of the trolley to be inspected, which includes the parameter information collected by the trolley to be inspected during the inspection of the target product.

[0112] The determination module 320 is used to determine the target feature information based on the original feature information. The target feature information includes the feature information remaining after anomaly processing of the original feature information.

[0113] The detection module 330 is used to obtain the target detection result for the trolley to be detected based on the target feature information and the preset detection model.

[0114] It is understandable that the determination module 320 can be used to perform column-to-row conversion on the column data of the column field containing the parameter information in the original feature information to determine the feature information to be processed of the trolley to be detected; perform anomaly processing on the feature information to be processed based on preset anomaly processing rules; and determine the target feature information based on the feature information obtained from the anomaly processing. Among them, the preset anomaly processing rules include, for the feature information to be processed, removing column data with column field missing rate exceeding the column missing rate threshold, filling row data with row field missing rate lower than the row missing rate threshold, removing column data that does not meet the preset distribution pattern, removing column data that does not meet the preset temperature correlation, removing column data that does not meet the preset temperature range, and removing column data that does not meet the preset power correlation.

[0115] It is understandable that module 320 can also be used to obtain the target quantity of target products belonging to the same preset equipment part code, where the preset equipment part code is the hardware parameter model of the same target product; if the target quantity is less than a preset quantity threshold, the data in the row where the preset equipment part code corresponding to the target quantity is located in the feature information to be processed is removed.

[0116] It is understandable that the determination module 320 can also be used to perform feature amplification processing on the feature information obtained from anomaly processing, and determine multiple feature information after amplification processing; based on the preset feature information correlation and preset feature information quantity, multiple feature information are filtered to determine the target feature information.

[0117] It is understandable that the detection module 330 can be used to input target feature information into a preset detection model to obtain a preset number of detection results of the trolley to be inspected under different preset equipment part codes and different preset detection sub-models; based on the abnormal detection results in the preset number of detection results, the target detection results for the trolley to be inspected are obtained.

[0118] It is understandable that the detection module 330 can also be used to determine the proportion of abnormal detection results among a preset number of detection results; if the proportion exceeds the first preset proportion, obtain the target detection result that the trolley to be detected is an abnormal trolley; if the proportion is between the second preset proportion and the first preset proportion, obtain the target detection result that the trolley to be detected is a risky trolley; if the proportion is lower than the second preset proportion, obtain the target detection result that the trolley to be detected is a normal trolley.

[0119] It is understandable that the detection module 330 can also be used to obtain the target coding information of the risk vehicle; based on the target coding information, it can send early warning information to the user terminal, which is used to remind the corresponding review personnel of the user terminal to conduct on-site assessment of the risk vehicle.

[0120] It is understood that the trolley detection method provided in this embodiment of the invention also includes a training module for training a model. The training method of the preset detection model includes: acquiring multiple different initial detection sub-models for anomaly detection of the trolley to be detected; training each initial detection sub-model a preset number of times based on target feature information, and determining multiple intermediate detection results of multiple intermediate detection sub-models after each training; sending the multiple intermediate detection results to a user terminal and receiving the review results of the multiple intermediate detection results from the user terminal; and determining the preset detection sub-model and the preset detection model corresponding to the preset detection sub-model based on the review results and the intermediate detection sub-model with updated model parameters.

[0121] The trolley inspection device provided in this embodiment of the invention first acquires the original feature information of the trolley to be inspected. Since the original feature information includes parameter information collected by the trolley during the inspection of the target product, the target feature information is determined from the original feature information by performing anomaly processing. Then, based on the target feature information and a preset detection model, the target detection result for the trolley to be inspected is obtained. In this way, by automatically acquiring information, handling anomalies, and performing model detection, the device achieves the purpose of timely and intelligent detection of whether the trolley has anomalies. This not only saves time and effort but is also highly efficient and accurate. At the same time, the combination of model learning and detection technology can further ensure the accuracy of the detection results, thereby effectively improving the accuracy and reliability of trolley inspection.

[0122] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device 400 may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute the following methods:

[0123] Obtain the original feature information of the trolley to be inspected, which includes the parameter information collected by the trolley to be inspected during the inspection of the target product;

[0124] Based on the original feature information, the target feature information is determined. The target feature information includes the feature information remaining after anomaly processing of the original feature information.

[0125] Based on target feature information and a preset detection model, the target detection results for the trolley to be detected are obtained.

[0126] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0127] On the other hand, embodiments of the present invention disclose a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer can perform the methods provided in the above-described method embodiments, such as including:

[0128] Obtain the original feature information of the trolley to be inspected, which includes the parameter information collected by the trolley to be inspected during the inspection of the target product;

[0129] Based on the original feature information, the target feature information is determined. The target feature information includes the feature information remaining after anomaly processing of the original feature information.

[0130] Based on target feature information and a preset detection model, the target detection results for the trolley to be detected are obtained.

[0131] In another aspect, embodiments of the present invention also provide a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the transmission methods provided in the above embodiments, including, for example:

[0132] Obtain the original feature information of the trolley to be inspected, which includes the parameter information collected by the trolley to be inspected during the inspection of the target product;

[0133] Based on the original feature information, the target feature information is determined. The target feature information includes the feature information remaining after anomaly processing of the original feature information.

[0134] Based on target feature information and a preset detection model, the target detection results for the trolley to be detected are obtained.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0136] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0137] Finally, it should be noted that the above embodiments are only for illustrating the present invention and not for limiting the present invention. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that various combinations, modifications, or equivalent substitutions of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention and should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting trolleys, characterized in that, include: Obtain the original feature information of the trolley to be inspected, which includes parameter information collected by the trolley to be inspected during the inspection of the target product; The original feature information includes the preset equipment part code, the target product ID, the target code information of the trolley to be inspected, and dynamic parameters during the inspection process, including at least one of temperature, current, voltage, and power. Based on the original feature information, target feature information is determined, wherein the target feature information includes the feature information remaining after anomaly processing of the original feature information; Based on the target feature information and the preset detection model, the target detection result for the vehicle to be detected is obtained; The step of determining the target feature information based on the original feature information includes: The column data of the column field containing the parameter information in the original feature information is transformed into a row to determine the feature information to be processed of the trolley to be detected; Based on preset anomaly handling rules, the feature information to be processed is subjected to anomaly handling; The target feature information is determined based on the feature information obtained from anomaly processing; The preset anomaly handling rules include, for the feature information to be processed, removing column data whose column field missing rate exceeds the column missing rate threshold, filling row data whose row field missing rate is lower than the row missing rate threshold, removing column data that does not meet the preset distribution pattern, removing column data that does not meet the preset temperature correlation, removing column data that does not meet the preset temperature range, and removing column data that does not meet the preset power correlation.

2. The trolley detection method according to claim 1, characterized in that, The anomaly handling of the feature information to be processed also includes: Obtain the target quantity of target products belonging to the same preset equipment part code, where the preset equipment part code is the hardware parameter model of the same target product; If the target quantity is determined to be less than a preset quantity threshold, the row of data containing the preset equipment part code corresponding to the target quantity in the feature information to be processed is removed.

3. The trolley detection method according to claim 1, characterized in that, The determination of the target feature information based on the feature information obtained from anomaly processing includes: The feature information obtained from anomaly processing is subjected to feature amplification processing to determine multiple feature information after amplification processing; Based on the pre-defined relevance and quantity of feature information, the multiple feature information is filtered to determine the target feature information.

4. The trolley detection method according to claim 1, characterized in that, The preset detection model includes different preset detection sub-models. The process of obtaining the target detection result for the vehicle to be inspected based on the target feature information and the preset detection model includes: The target feature information is input into the preset detection model to obtain a preset number of detection results of the trolley to be detected under different preset equipment part codes and different preset detection sub-models. Based on the abnormal detection results among the preset number of detection results, the target detection result for the vehicle to be inspected is obtained.

5. The trolley detection method according to claim 4, characterized in that, The step of obtaining the target detection result for the vehicle to be inspected based on the abnormal detection results among the preset number of detection results includes: Determine the percentage of abnormal detection results among the preset number of detection results; Once the percentage is determined to exceed a first preset percentage, a target detection result is obtained indicating that the vehicle to be detected is an abnormal vehicle. Determine that the percentage is between the second preset percentage and the first preset percentage, and obtain the target detection result that the trolley to be detected is a risk trolley; If the percentage is determined to be lower than the second preset percentage, the target detection result of the trolley to be detected being a normal trolley is obtained.

6. The trolley inspection method according to claim 5, characterized in that, After obtaining the target detection result that the trolley to be detected is a risk trolley, the method further includes: Obtain the target coding information of the risk trolley; Based on the target encoding information, a warning message is sent to the user terminal. The warning message is used to remind the user terminal's corresponding review personnel to conduct an on-site assessment of the risk vehicle.

7. The trolley detection method according to any one of claims 1 to 6, characterized in that, The training method for the preset detection model includes: Obtain multiple different initial detection sub-models for anomaly detection of the trolley to be inspected; Based on the target feature information, each initial detection sub-model is trained a preset number of times to determine multiple intermediate detection results of multiple intermediate detection sub-models after each training. The multiple intermediate detection results are sent to the user terminal, and the verification results of the multiple intermediate detection results are received from the user terminal. Based on the verification results and the intermediate detection sub-model after updating each model parameter, the preset detection sub-model and the preset detection model corresponding to the preset detection sub-model are determined.

8. A trolley detection device, characterized in that, include: The acquisition module is used to acquire the original feature information of the trolley to be inspected. The original feature information includes parameter information collected by the trolley to be inspected during the inspection of the target product. The original feature information includes a preset equipment part code, the ID of the target product, the target code information of the trolley to be inspected, and dynamic parameters during the inspection process, including at least one of temperature, current, voltage, and power. The determination module is used to determine target feature information based on the original feature information, wherein the target feature information includes the feature information remaining after anomaly processing of the original feature information; The detection module is used to obtain the target detection result for the vehicle to be detected based on the target feature information and the preset detection model; The determining module is further configured to perform a column-to-row operation on the column data of the column field containing the parameter information in the original feature information to determine the feature information to be processed of the trolley to be detected; Based on preset anomaly handling rules, the feature information to be processed is subjected to anomaly handling; Based on the feature information obtained from anomaly processing, the target feature information is determined; wherein, the preset anomaly processing rules include, for the feature information to be processed, removing column data with column field missing rates exceeding the column missing rate threshold, filling row data with row field missing rates lower than the row missing rate threshold, removing column data that does not meet the preset distribution pattern, removing column data that does not meet the preset temperature correlation, removing column data that does not meet the preset temperature range, and removing column data that does not meet the preset power correlation.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the trolley detection method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the trolley detection method as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the trolley detection method as described in any one of claims 1 to 7.

Images