A method and device for predicting the health value of AGVs

Abstract

This application provides a method and apparatus for predicting the health value of an AGV. The prediction method includes: acquiring measured values ​​of characteristic parameters of a target AGV at each sampling time; determining the operating condition of the target AGV at each sampling time based on the measured values ​​of the characteristic parameters; and inputting the measured values ​​of the characteristic parameters at each sampling time into a health value prediction model corresponding to the operating condition at that sampling time to obtain the health value of the target AGV at that sampling time. The prediction method and apparatus can improve the accuracy of the predicted AGV health value.

Description

Technical Field

[0001] This application relates to the field of safety status monitoring technology for industrial equipment, and in particular to a method and device for predicting the health value of AGVs. Background Technology

[0002] The health value of industrial equipment is a crucial aspect of predictive maintenance, reflecting both the equipment's current health status and remaining lifespan. With the continuous development of industrial intelligence, the operating environment of industrial equipment is becoming increasingly complex, and the complexity of industrial equipment itself is also increasing. Therefore, predicting equipment health values ​​through models has become a current trend in equipment health status assessment.

[0003] In existing technologies, for AGVs (Automated Guided Vehicles) in industrial equipment, the measured values ​​of the AGV's characteristic parameters are typically input directly into a health prediction model to predict the AGV's health value. However, due to the inherent characteristics of AGVs, they generally experience different operating conditions during use. For example, an AGV usually moves to a specific location and then stops to perform other operations. Therefore, operating conditions include both motion conditions and operational conditions. Under these different conditions, the measured values ​​of the AGV's characteristic parameters may exhibit different characteristics. For instance, the changes in voltage and current under motion conditions differ from those under operational conditions. If the influence of operating conditions is not considered, and all the measured values ​​of characteristic parameters are directly input into the health prediction model to predict the AGV's health value, the predicted AGV health value will be inaccurate, thus affecting subsequent predictive maintenance of the equipment. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a method and apparatus for predicting the health value of AGVs, so as to improve the accuracy of the predicted health value of AGVs.

[0005] In a first aspect, embodiments of this application provide a method for predicting the health value of an AGV, the prediction method comprising:

[0006] Obtain the measured values ​​of the characteristic parameters of the target AGV at each sampling time;

[0007] For the measured values ​​of the characteristic parameters at each sampling time, the operating condition of the target AGV at that sampling time is determined;

[0008] The measured values ​​of the characteristic parameters at the sampling time are input into the health value prediction model corresponding to the operating conditions at the sampling time to obtain the health value of the target AGV at that sampling time.

[0009] Optionally, determining the operating condition of the target AGV at each sampling time based on the measured values ​​of the characteristic parameters includes:

[0010] For each sampling time, the measured value of the characteristic parameter is input into the pre-trained classification model to determine the operating condition of the target AGV at that sampling time.

[0011] The classification model is trained using an unsupervised training method based on a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period.

[0012] Optionally, a health value prediction model corresponding to each operating condition can be trained in the following manner:

[0013] Obtain a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period; the historical sample dataset includes first historical sample data corresponding to each historical time period;

[0014] For each historical time in the historical sample dataset, the operating condition corresponding to that historical time is determined based on the first historical sample data corresponding to that historical time.

[0015] Based on the determined operating conditions, the historical sample dataset is split to obtain the first sub-historical sample dataset under each operating condition and the second sub-historical sample dataset under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset.

[0016] For each operating condition, the health value prediction model is trained based on the first sub-historical sample dataset and each second sub-historical sample dataset under that operating condition.

[0017] Optionally, the step of splitting the historical sample dataset based on the determined operating conditions to obtain a first sub-historical sample dataset under each operating condition and a second sub-historical sample dataset under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset includes:

[0018] Based on the determined operating conditions, the historical sample dataset is classified to obtain a first sub-historical sample dataset corresponding to each operating condition; the first sub-historical sample dataset includes first historical sample data corresponding to multiple first predetermined historical times.

[0019] For each first sub-historical sample dataset, the first sub-historical sample dataset is decomposed according to a preset time granularity to obtain second sub-historical sample datasets at various second time granularities corresponding to the first time granularity of the first sub-historical sample dataset; each second sub-historical sample dataset includes multiple second historical sample data corresponding to a second predetermined historical time.

[0020] Optionally, the step of decomposing each first sub-historical sample dataset according to a preset time granularity to obtain second sub-historical sample datasets at various second time granularities corresponding to the first time granularity of the first sub-historical sample dataset includes:

[0021] For each first sub-historical sample dataset, the range of multiple first predetermined historical times included in the first sub-historical sample dataset is divided according to a preset time granularity, and each second predetermined historical time under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset is obtained respectively.

[0022] For each second predetermined historical time, multiple target first predetermined historical times and target first historical sample data corresponding to the second predetermined historical time are selected from the first sub-historical sample dataset;

[0023] The historical measured values ​​of all characteristic parameters included in the first historical sample data of the target are statistically analyzed according to a preset statistical method to obtain statistical values, and the obtained statistical values ​​are used as reference values ​​of characteristic parameters included in the second historical sample data corresponding to the second predetermined historical time.

[0024] The second sub-historical sample dataset is determined based on the reference values ​​of the characteristic parameters included in the second historical sample data corresponding to each second predetermined historical time.

[0025] Optionally, the health value prediction model for each operating condition is trained based on the first sub-historical sample dataset and each second sub-historical sample dataset under that operating condition, including:

[0026] For each second predetermined historical time in each second sub-historical sample dataset under each operating condition, multiple target first predetermined historical times and target first historical sample data corresponding to the second predetermined historical time are selected from the first sub-historical sample dataset.

[0027] The difference between the target first historical sample data and the second historical sample data corresponding to the second predetermined historical time is compared to determine the difference data corresponding to the target first historical sample data.

[0028] The obtained difference data is used as training sample data. The health value corresponding to each first predetermined historical time included in the first sub-historical sample dataset is used as the label, and the loss function is used to train the health value prediction model corresponding to the operating condition.

[0029] Secondly, embodiments of this application provide an AGV health value prediction device, the prediction device comprising:

[0030] The acquisition module is used to acquire the measured values ​​of the characteristic parameters of the target AGV at each sampling time.

[0031] The determination module is used to determine the operating condition of the target AGV at each sampling time based on the measured values ​​of the characteristic parameters at that sampling time.

[0032] The prediction module is used to input the measured values ​​of the characteristic parameters at the sampling time into the health value prediction model corresponding to the operating conditions at the sampling time, so as to obtain the health value of the target AGV at the sampling time.

[0033] Optionally, the determining module is specifically used for:

[0034] For each sampling time, the measured value of the characteristic parameter is input into the pre-trained classification model to determine the operating condition of the target AGV at that sampling time.

[0035] The classification model is trained using an unsupervised training method based on a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period.

[0036] Thirdly, embodiments of this application provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the AGV health value prediction method described above are performed.

[0037] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the AGV health value prediction method described above.

[0038] This application provides a method and apparatus for predicting the health value of an AGV, which considers the operating conditions of the AGV at each sampling time and inputs the measured values ​​of the characteristic parameters at each sampling time into the health value prediction model corresponding to the operating conditions. Compared with the prior art, which directly inputs the measured values ​​of the characteristic parameters of the AGV into the health value prediction model to predict the health value of the AGV, the predicted health value of the AGV is more accurate.

[0039] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 A flowchart illustrating a method for predicting the health value of an AGV, provided by an exemplary embodiment of this application, is shown.

[0042] Figure 2 This invention provides a schematic diagram of the structure of an AGV health value prediction device according to an exemplary embodiment of the present application.

[0043] Figure 3 A schematic diagram of the structure of an electronic device provided by an exemplary embodiment of this application is shown. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0045] In existing technologies, for AGVs (Automated Guided Vehicles) in industrial equipment, the measured values ​​of the AGV's characteristic parameters are typically input directly into a health prediction model to predict the AGV's health value. However, due to the inherent characteristics of AGVs, they generally experience different operating conditions during use. For example, an AGV usually moves to a specific location and then stops to perform other operations. Therefore, operating conditions include both motion conditions and operational conditions. Under these different conditions, the measured values ​​of the AGV's characteristic parameters may exhibit different characteristics. For instance, the changes in voltage and current under motion conditions differ from those under operational conditions. If the influence of operating conditions is not considered, and all the measured values ​​of characteristic parameters are directly input into the health prediction model to predict the AGV's health value, the predicted AGV health value will be inaccurate, thus affecting subsequent predictive maintenance of the equipment.

[0046] Based on this, the present application provides a method and apparatus for predicting the health value of an AGV, which takes into account the operating conditions of the AGV at each sampling time and inputs the measured values ​​of the characteristic parameters at each sampling time into the health value prediction model corresponding to the operating conditions. Compared with the prior art, which directly inputs the measured values ​​of the characteristic parameters of the AGV into the health value prediction model to predict the health value of the AGV, the predicted health value of the AGV is more accurate.

[0047] Please see Figure 1 , Figure 1 A flowchart of a method for predicting the health value of an AGV provided in an embodiment of this application is shown.

[0048] like Figure 1 As shown in the figure, an AGV health value prediction method provided in this application includes the following steps:

[0049] S1. Obtain the measured values ​​of the characteristic parameters of the target AGV at each sampling time.

[0050] Here, the time granularity corresponding to the sampling time is set according to the actual situation. In specific implementation, in order to collect as many measured values ​​of characteristic parameters as possible, the time granularity corresponding to the sampling time can be set to "minutes". When the time granularity corresponding to the sampling time is "minutes", the measured values ​​of the characteristic parameters of the target AGV can be collected once every minute.

[0051] Here, the characteristic parameters include, but are not limited to, at least one of the following: battery level, navigation status, current, voltage, coordinates, speed, controller device value, total mileage, and cumulative running time. Coordinates may include X-coordinate, Y-coordinate, and angular coordinate; speed may include linear velocity and angular velocity; and controller device value may include controller temperature value, controller humidity value, and controller voltage value. It should be understood that the above characteristic parameters are merely illustrative, and the actual content and number of characteristic parameters may vary depending on the AGV's environment.

[0052] S2. For the measured values ​​of the characteristic parameters at each sampling time, determine the operating condition of the target AGV at that sampling time;

[0053] Here, operating conditions can include motion conditions and operation conditions. Motion conditions indicate that the target AGV is in a driving state, while operation conditions indicate that the target AGV is in a state of mechanical operation after driving to a specific position.

[0054] S3. Input the measured value of the characteristic parameter at the sampling time into the health value prediction model corresponding to the operating condition at the sampling time to obtain the health value of the target AGV at the sampling time.

[0055] Here, the health value prediction model corresponding to each operating condition is pre-trained. When the operating conditions include motion conditions and operational conditions, the health value prediction model includes the health value prediction model corresponding to the motion conditions and the health value prediction model corresponding to the operational conditions.

[0056] As an example, the health value prediction model could be the LightGBM model.

[0057] The present application provides a method for predicting the health value of an AGV, which considers the operating conditions of the AGV at each sampling time and inputs the measured values ​​of the characteristic parameters at each sampling time into the health value prediction model corresponding to the operating conditions. Compared with the prior art, which directly inputs the measured values ​​of the characteristic parameters of the AGV into the health value prediction model to predict the health value of the AGV, the predicted health value of the AGV is more accurate.

[0058] Specifically, as an example, in step S2, the measured values ​​of the characteristic parameters at each sampling time can be input into a pre-trained classification model to determine the operating condition of the target AGV at that sampling time.

[0059] The classification model is trained using an unsupervised training method based on a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period.

[0060] Here, the historical sample dataset can be a dataset provided by the AGV manufacturer. The predetermined historical time period can be the time span from the start of use of the same model of AGV to the end of its service life. The time granularity of the historical sample dataset is generally in minutes. The historical sample dataset includes first historical sample data corresponding to each historical time point. The first historical sample data includes historical measured values ​​and historical health values ​​of characteristic parameters. Here, characteristic parameters may include, but are not limited to, at least one of the following: battery level, navigation status, current, voltage, coordinates, speed, controller device values, total mileage, and cumulative running time. Coordinates may include X-coordinate, Y-coordinate, and angular coordinates; speed may include linear velocity and angular velocity; and controller device values ​​may include controller temperature, controller humidity, and controller voltage.

[0061] For example, the two first historical sample data included in the historical sample dataset can be: "2019 / 4 / 13 14:01" "Battery: 56" "Navigation status: 2" "Current: -7.3221" ... "Health value; 75%" and "2019 / 4 / 13 14:02" "Battery: 55" "Navigation status: 2" "Current: -7.3221" ... "Health value; 75%".

[0062] When training a classification model using unsupervised training, the K-means clustering method can be used.

[0063] Specifically, as an example, a health value prediction model corresponding to each operating condition can be trained in the following way:

[0064] S11. Obtain a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period; the historical sample dataset includes first historical sample data corresponding to each historical time period;

[0065] Here, the historical sample dataset used to train the health value prediction model corresponding to each operating condition can be the same historical sample dataset used when training the classification model. Since the historical sample dataset has already been introduced, it will not be repeated here.

[0066] S12. For each historical time in the historical sample dataset, determine the operating condition corresponding to that historical time based on the first historical sample data corresponding to that historical time.

[0067] As an example, the operating conditions corresponding to a given historical time can be determined using the previously trained classification model.

[0068] As another example, the operating condition corresponding to each historical time can be determined based on the linear velocity and angular velocity values ​​included in the historical measured values ​​of the first historical sample data corresponding to that historical time. For example, if either the linear velocity or the angular velocity is not 0, it means that the AGV is in a traveling state, and the operating condition corresponding to that historical time is determined to be a motion condition. If both the linear velocity and the angular velocity are 0, it means that the AGV is in a state of mechanical operation after traveling to a specific position, and the operating condition corresponding to that historical time is determined to be an operation condition.

[0069] S13. Based on the determined operating conditions, the historical sample dataset is split to obtain the first sub-historical sample dataset under each operating condition and the second sub-historical sample dataset under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset.

[0070] As an example, step S13 may include the following steps:

[0071] S131. Based on the determined operating conditions, the historical sample dataset is classified to obtain a first sub-historical sample dataset corresponding to each operating condition; the first sub-historical sample dataset includes multiple first historical sample data corresponding to a first predetermined historical time.

[0072] It is understood that for each first sub-historical sample dataset, the first historical sample data corresponding to the multiple first predetermined historical times included in the first sub-historical sample dataset are extracted from the historical sample dataset. Therefore, the number of first predetermined historical times included in the first sub-historical sample dataset is less than the number of historical times in the historical sample dataset.

[0073] S132. For each first sub-historical sample dataset, decompose the first sub-historical sample dataset according to a preset time granularity to obtain second sub-historical sample datasets at various second time granularities corresponding to the first time granularity of the first sub-historical sample dataset. Each second sub-historical sample dataset includes multiple second historical sample data corresponding to a second predetermined historical time.

[0074] It is understandable that, since the first sub-historical sample dataset is classified based on the historical sample dataset, the time granularity of the first sub-historical sample dataset is the same as that of the historical sample dataset. For example, if the time granularity of the historical sample dataset is "minutes", the corresponding time granularity of the first sub-historical sample dataset is also "minutes".

[0075] Here, the preset time granularity includes multiple second time granularities. The correspondence between the second and first time granularities means that the second time granularity is larger than the first time granularity. For example, when the first time granularity is "minute," the second time granularity can include "hour" and "day."

[0076] As an example, when the second time granularity includes "hour" and "day", the second sub-historical sample dataset corresponding to the granularity of "hour" and the second sub-historical sample dataset corresponding to the granularity of "day" can be obtained in step S132 respectively.

[0077] Specifically, regarding step S132, in a practical implementation, it may include the following steps:

[0078] S1321. For each first sub-historical sample dataset, divide the range of multiple first predetermined historical times included in the first sub-historical sample dataset according to the preset time granularity, and obtain each second predetermined historical time under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset.

[0079] For example, suppose the first sub-historical sample dataset is: “2019 / 4 / 13 14:01”, “Battery: 56”, “Navigation Status: 2”, “Current: -7.3221”, etc.; “2019 / 4 / 13 14:02”, “Battery: 55”, “Navigation Status: 2”, “Current: -7.3221”, etc.; “2019 / 4 / 13 15:09”, “Battery: 50”, “Navigation Status: 1”, “Current: -6.7281”, etc.; then the second predetermined historical time under each second time granularity obtained after range division can include: the second predetermined historical time “2019 / 4 / 13 14” with a time granularity of “hour”; the second predetermined historical time “2019 / 4 / 13 15” with a time granularity of “hour”; and the second predetermined historical time “2019 / 4 / 13” with a time granularity of “day”.

[0080] S1322. For each second predetermined historical time, select multiple target first predetermined historical times and target first historical sample data corresponding to the second predetermined historical time from the first sub-historical sample dataset;

[0081] Following the example above, for instance, for the second predetermined historical time "2019 / 4 / 13 14", the multiple target first predetermined historical times corresponding to the second predetermined historical time selected from the first sub-historical sample dataset are "2019 / 4 / 13 14:01" and "2019 / 4 / 13 14:01", and the target first historical sample data corresponding to the multiple target first predetermined historical times are: ["Battery: 56", "Navigation Status: 2", "Current: -7.3221"...] and ["Battery: 55", "Navigation Status: 2", "Current: -7.3221"...].

[0082] S1323. The historical measured values ​​of all characteristic parameters included in the first historical sample data of the target are statistically analyzed according to a preset statistical method to obtain statistical values, and the obtained statistical values ​​are used as reference values ​​of characteristic parameters included in the second historical sample data corresponding to the second predetermined historical time.

[0083] Here, the predetermined statistical method may include, but is not limited to, any one of the following: calculating the average, calculating the arithmetic mean.

[0084] For example, when the predetermined statistical method is to calculate the arithmetic mean, in step S1323, for each characteristic parameter included in the target first historical sample data, the arithmetic mean of all historical measured values ​​of the characteristic parameter is determined, and the obtained arithmetic mean is used as the reference value of the characteristic parameter included in the second historical sample data corresponding to the second predetermined historical time.

[0085] For example, regarding the characteristic parameter "power consumption", for the power consumption included in the first historical sample data of the target, the arithmetic mean of all historical measured values ​​of the power consumption is determined, and the obtained arithmetic mean is used as a reference value for the power consumption included in the second historical sample data corresponding to the second predetermined historical time.

[0086] S1324. Based on the reference values ​​of the characteristic parameters included in the second historical sample data corresponding to each second predetermined historical time, determine the second sub-historical sample dataset.

[0087] S14. For each operating condition, the health value prediction model is trained based on the first sub-historical sample dataset and each second sub-historical sample dataset under that operating condition.

[0088] As an example, step S14 includes the following steps:

[0089] S141. For each second predetermined historical time in each second sub-historical sample dataset under each operating condition, select multiple target first predetermined historical times corresponding to the second predetermined historical time and target first historical sample data corresponding to the target first predetermined historical time from the first sub-historical sample dataset;

[0090] Here, the screening steps in step S141 and step S1322 are similar, so they will not be described again.

[0091] S142. Compare the target first historical sample data with the second historical sample data corresponding to the second predetermined historical time to determine the difference data corresponding to the target first historical sample data.

[0092] In step S142, firstly, for each characteristic parameter in the target first historical sample data, the historical measured value of the characteristic parameter is compared with the reference value of the corresponding characteristic parameter in the second historical sample data corresponding to the second predetermined historical time to determine the difference between the historical measured value of each characteristic parameter in the target first historical sample data and the reference value of the corresponding characteristic parameter. Then, all the obtained difference values ​​are determined as the difference data corresponding to the target first historical sample data.

[0093] For example, for a characteristic parameter of power consumption, for each historical measured value of power consumption in the target first historical sample data, the historical measured value of power consumption is compared with the reference value of power consumption in the second historical sample data corresponding to the second predetermined historical time to determine the difference between the historical measured value of power consumption and the reference value of power consumption in the target first historical sample data. Then, all the obtained difference values ​​are determined as the difference data corresponding to the target first historical sample data.

[0094] S143. Using the obtained difference data as training sample data, and labeling each first predetermined historical time included in the first sub-historical sample dataset with the health value, the loss function is used to train the health value prediction model corresponding to the operating condition.

[0095] Based on the same inventive concept, this application also provides a device for predicting the health value of an AGV.

[0096] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an AGV health value prediction device provided as an exemplary embodiment of this application.

[0097] like Figure 2 As shown, the prediction device 200 includes:

[0098] The acquisition module 210 is used to acquire the measured values ​​of the characteristic parameters of the target AGV at each sampling time.

[0099] The determination module 220 is used to determine the operating condition of the target AGV at each sampling time based on the measured values ​​of the characteristic parameters at each sampling time.

[0100] The prediction module 230 is used to input the measured value of the characteristic parameter at the sampling time into the health value prediction model corresponding to the operating condition at the sampling time, so as to obtain the health value of the target AGV at the sampling time.

[0101] Optionally, the determining module 220 is specifically used for:

[0102] For each sampling time, the measured value of the characteristic parameter is input into the pre-trained classification model to determine the operating condition of the target AGV at that sampling time.

[0103] The classification model is trained using an unsupervised training method based on a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period.

[0104] Optionally, the prediction device 200 further includes: a training module 240 (not shown in the figure);

[0105] The training module 240 is specifically used to train a health value prediction model corresponding to each operating condition in the following manner:

[0106] Obtain a historical sample dataset of AGVs of the same model as the target AGV within a predetermined historical time period; the historical sample dataset includes first historical sample data corresponding to each historical time period;

[0107] For each historical time in the historical sample dataset, the operating condition corresponding to that historical time is determined based on the first historical sample data corresponding to that historical time.

[0108] Based on the determined operating conditions, the historical sample dataset is split to obtain the first sub-historical sample dataset under each operating condition and the second sub-historical sample dataset under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset.

[0109] For each operating condition, the health value prediction model is trained based on the first sub-historical sample dataset and each second sub-historical sample dataset under that operating condition.

[0110] Optionally, the training module 240 is specifically used for:

[0111] Based on the determined operating conditions, the historical sample dataset is classified to obtain a first sub-historical sample dataset corresponding to each operating condition; the first sub-historical sample dataset includes first historical sample data corresponding to multiple first predetermined historical times.

[0112] For each first sub-historical sample dataset, the first sub-historical sample dataset is decomposed according to a preset time granularity to obtain second sub-historical sample datasets at various second time granularities corresponding to the first time granularity of the first sub-historical sample dataset; each second sub-historical sample dataset includes multiple second historical sample data corresponding to a second predetermined historical time.

[0113] Optionally, the training module 240 is specifically used for:

[0114] For each first sub-historical sample dataset, the range of multiple first predetermined historical times included in the first sub-historical sample dataset is divided according to a preset time granularity, and each second predetermined historical time under each second time granularity corresponding to the first time granularity of the first sub-historical sample dataset is obtained respectively.

[0115] For each second predetermined historical time, multiple target first predetermined historical times and target first historical sample data corresponding to the second predetermined historical time are selected from the first sub-historical sample dataset;

[0116] The historical measured values ​​of all characteristic parameters included in the first historical sample data of the target are statistically analyzed according to a preset statistical method to obtain statistical values, and the obtained statistical values ​​are used as reference values ​​of characteristic parameters included in the second historical sample data corresponding to the second predetermined historical time.

[0117] The second sub-historical sample dataset is determined based on the reference values ​​of the characteristic parameters included in the second historical sample data corresponding to each second predetermined historical time.

[0118] Optionally, the training module 240 is specifically used for:

[0119] For each second predetermined historical time in each second sub-historical sample dataset under each operating condition, multiple target first predetermined historical times and target first historical sample data corresponding to the second predetermined historical time are selected from the first sub-historical sample dataset.

[0120] The difference between the target first historical sample data and the second historical sample data corresponding to the second predetermined historical time is compared to determine the difference data corresponding to the target first historical sample data.

[0121] The obtained difference data is used as training sample data. The health value corresponding to each first predetermined historical time included in the first sub-historical sample dataset is used as the label, and the loss function is used to train the health value prediction model corresponding to the operating condition.

[0122] This application provides an AGV health value prediction device that considers the AGV's operating conditions at each sampling time and inputs the measured values ​​of the characteristic parameters at each sampling time into the health value prediction model corresponding to the operating conditions. Compared with the prior art, which directly inputs the measured values ​​of the AGV's characteristic parameters into the health value prediction model to predict the AGV's health value, this device makes the predicted AGV health value more accurate.

[0123] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 300 includes a processor 310, a memory 320, and a bus 330.

[0124] The memory 320 stores machine-readable instructions that can be executed by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate via the bus 330. When the machine-readable instructions are executed by the processor 310, the steps of the AGV health value prediction method in the above method embodiment can be performed. For specific implementation, please refer to the method embodiment, which will not be repeated here.

[0125] This application also provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it can execute the steps of the AGV health value prediction method as described in the above method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0126] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0127] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0128] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0129] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0130] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, 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 this application. 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.

[0131] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting an AGV health value, characterized by, The prediction method comprises: obtaining a measured value of a characteristic parameter of the target AGV at each sampling time; determining the running condition of the target AGV at each sampling time based on the measured value of the characteristic parameter at the sampling time; inputting the measured value of the characteristic parameter at the sampling time into the health value prediction model corresponding to the running condition at the sampling time to obtain the health value of the target AGV at the sampling time; training the health value prediction model corresponding to each running condition by the following method: obtaining a historical sample data set of a same type AGV belonging to the same type as the target AGV within a predetermined historical time period; the historical sample data set comprises first historical sample data corresponding to each historical time; determining the running condition corresponding to each historical time based on the first historical sample data corresponding to the historical time in the historical sample data set; based on the determined running condition, splitting the historical sample data set to obtain a first sub-historical sample data set under each running condition and a second sub-historical sample data set under each second time granularity corresponding to the first time granularity of the first sub-historical sample data set; based on the determined running condition, splitting the historical sample data set to obtain a first sub-historical sample data set under each running condition and a second sub-historical sample data set under each second time granularity corresponding to the first time granularity of the first sub-historical sample data set, comprising: based on the determined running condition, classifying the historical sample data set to obtain a first sub-historical sample data set corresponding to each running condition; the first sub-historical sample data set comprises first historical sample data corresponding to a plurality of first predetermined historical times; for each first sub-historical sample data set, decomposing the first sub-historical sample data set according to a preset time granularity to obtain a second sub-historical sample data set under each second time granularity corresponding to the first time granularity of the first sub-historical sample data set; each second sub-historical sample data set comprises second historical sample data corresponding to a plurality of second predetermined historical times; training the health value prediction model corresponding to each running condition based on the first sub-historical sample data set and each second sub-historical sample data set under the running condition; training the health value prediction model corresponding to each running condition based on the first sub-historical sample data set and each second sub-historical sample data set under the running condition, comprising: for each second predetermined historical time in each second sub-historical sample data set under each running condition, filtering a plurality of target first predetermined historical times corresponding to the second predetermined historical time and target first historical sample data corresponding to the target first predetermined historical time from the first sub-historical sample data set; determine difference data corresponding to the target first historical sample data by differentiating the target first historical sample data from second historical sample data corresponding to the second predetermined historical time; use the obtained difference data as sample data for training, use a health value corresponding to each first predetermined historical time included in the first sub-historical sample data set as a label, and train the health value prediction model corresponding to the operating condition by using a loss function.

2. The prediction method of claim 1, wherein, The measured value of the characteristic parameter at each sampling time is used to determine the operating condition of the target AGV at the sampling time, including: The measured value of the characteristic parameter at each sampling time is used to determine the operating condition of the target AGV at the sampling time, including: The classification model is trained based on historical sample data sets of AGVs of the same type as the target AGV at a predetermined historical time period, and is trained by using an unsupervised training method.

3. The prediction method of claim 1, wherein, The first sub-historical sample data set is divided into second sub-historical sample data sets corresponding to each second time granularity according to a preset time granularity, including: The first sub-historical sample data set is divided into second sub-historical sample data sets corresponding to each second time granularity according to a preset time granularity, including: For each second predetermined historical time, a plurality of target first predetermined historical times corresponding to the second predetermined historical time and target first historical sample data corresponding to the target first predetermined historical times are filtered from the first sub-historical sample data set; The historical measured values of all characteristic parameters included in the target first historical sample data are statistically processed according to a preset statistical method to obtain statistical values, and the obtained statistical values are used as reference values of the characteristic parameters included in the second historical sample data corresponding to the second predetermined historical time. Based on the reference values of the characteristic parameters included in the second historical sample data corresponding to each second predetermined historical time, a second sub-historical sample data set is determined.

4. An AGV health value prediction device characterized by comprising: The prediction device includes: An acquisition module is configured to acquire a measured value of a characteristic parameter of a target AGV at each sampling time; A determination module is configured to determine an operating condition of the target AGV at each sampling time based on the measured value of the characteristic parameter at the sampling time; A prediction module is configured to input the measured value of the characteristic parameter at the sampling time into a health value prediction model corresponding to the operating condition at the sampling time to obtain a health value corresponding to the target AGV at the sampling time. The prediction device further includes a training module, and the training module is specifically configured to train a health value prediction model corresponding to each operating condition by: obtain a historical sample data set of a same type AGV in a predetermined historical time period, the historical sample data set including first historical sample data corresponding to each historical time; for each historical time in the historical sample data set, determine a running condition corresponding to the historical time based on the first historical sample data corresponding to the historical time; based on the determined running conditions, split the historical sample data set to obtain first sub-historical sample data sets under each running condition and second sub-historical sample data sets under each second time granularity corresponding to a first time granularity of the first sub-historical sample data set; for each running condition corresponding health value prediction model, train the health value prediction model corresponding to the running condition based on the first sub-historical sample data set and each second sub-historical sample data set under the running condition; the training module is specifically configured to: based on the determined running conditions, classify the historical sample data set to obtain a first sub-historical sample data set corresponding to each running condition, the first sub-historical sample data set including first historical sample data corresponding to a plurality of first predetermined historical times; for each first sub-historical sample data set, decompose the first sub-historical sample data set according to a preset time granularity to obtain a second sub-historical sample data set under each second time granularity corresponding to a first time granularity of the first sub-historical sample data set, each second sub-historical sample data set including second historical sample data corresponding to a plurality of second predetermined historical times; the training module is specifically configured to: for each second predetermined historical time in each second sub-historical sample data set under each running condition, filter a plurality of target first predetermined historical times corresponding to the second predetermined historical time and target first historical sample data corresponding to the target first predetermined historical time from the first sub-historical sample data set; differentially compare the target first historical sample data with the second historical sample data corresponding to the second predetermined historical time to determine difference data corresponding to the target first historical sample data; use the obtained difference data as training sample data, use the health value corresponding to each first predetermined historical time included in the first sub-historical sample data set as a label, and train the health value prediction model corresponding to the running condition using a loss function.

5. The prediction device of claim 4, wherein, the determination module is specifically configured to: for the measured value of the characteristic parameter at each sampling time, input the measured value of the characteristic parameter at the sampling time into the pre-trained classification model to determine the running condition of the target AGV at the sampling time. The classification model is trained based on the historical sample data set of the same type AGV in the predetermined historical time period using an unsupervised training method.

6. An electronic device, comprising: including: A processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, the machine readable instructions are executed by the processor to perform the steps of the AGV health value prediction method according to any one of claims 1 to 3.

7. A computer readable storage medium characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to perform the steps of the AGV health value prediction method according to any one of claims 1 to 3.

Images