Parameter prediction method and device, model training method and device and loader

Abstract

The invention provides a parameter prediction method and device, a model training method and device and a loader, and relates to the technical field of engineering machinery, and the prediction method comprises the steps: determining the current working condition of the loader; acquiring working data of the loader under the current working condition at the current first moment; the working data are input into a machine learning model corresponding to the current working condition so as to predict the charge state of a battery of the loader at the second moment and the vehicle speed of the loader at the second moment, the second moment is after the first moment, the current working condition is one of the multiple working conditions, and the vehicle speed of the loader is one of the multiple working conditions. The machine learning model corresponding to one of the at least two working conditions is different from the machine learning model corresponding to the other working condition.

Description

Technical Field

[0001] The present disclosure relates to the technical field of engineering machinery, and in particular to a parameter prediction method and device, a model training method and device, and a loader. Background Art

[0002] Loaders are a type of engineering machinery widely used in construction. The main function of a loader is to load and unload bulk materials such as soil, sand and gravel. Summary of the invention

[0003] In the related art, it is impossible to accurately predict the state of charge (SOC) of the battery of the loader and the vehicle speed of the loader.

[0004] After analysis, it is found that there are multiple working conditions when the loader is working. In the related art, when the loader is working, it is impossible to accurately obtain the SOC of the battery of the loader and the speed of the loader under different working conditions in the future. In other words, it is impossible to accurately predict the SOC of the battery of the loader and the speed of the loader.

[0005] In order to solve the above problems, the embodiments of the present disclosure propose the following solutions.

[0006] According to one aspect of an embodiment of the present disclosure, a parameter prediction method is provided, including determining a current operating condition of a loader; obtaining working data of the loader under the current operating condition at a current first moment; and inputting the working data into a machine learning model corresponding to the current operating condition to predict a battery state of charge of the loader at a second moment and a vehicle speed at the second moment, the second moment being after the first moment, wherein the current operating condition is one of multiple operating conditions, and the machine learning model corresponding to one of at least two operating conditions is different from the machine learning model corresponding to another one.

[0007] In some embodiments, the working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power.

[0008] In some embodiments, determining the current operating condition of the loader includes: determining the current operating condition of the loader based on a pressure value of a working pump of the loader, wherein the working pump is used to drive at least one of a steering cylinder of the loader, a bucket cylinder of the loader, and a boom cylinder of the loader.

[0009] In some embodiments, determining the current operating condition of the loader based on the pressure value of the working pump of the loader includes: when the pressure value of the working pump is greater than a first preset pressure value, determining that the current operating condition is one of a first group of operating conditions, the first group of operating conditions including one or more of a first operating condition, a second operating condition, a third operating condition, a fourth operating condition and a fifth operating condition, wherein: under the first operating condition, the operating stages of the loader include, in chronological order: an empty forward stage of driving in a straight line, a shoveling operation stage, a loaded backward stage of driving in a straight line, and a loaded forward stage of driving in a curve; under the second operating condition, the operating stages of the loader include, in chronological order: an empty forward stage of driving in a straight line, a shoveling operation stage, a loaded backward stage of driving in a straight line, and a loaded forward stage of driving in a curve. The operating stages of the loader include the following: an unloaded forward stage, a shoveling operation stage, a loaded reverse stage while traveling in a straight line, and a loaded forward stage while traveling in a straight line; under the third operating condition, the operating stages of the loader include, in chronological order: an unloaded forward stage while traveling along a curve, a shoveling operation stage, a loaded reverse stage while traveling along a curve, and a loaded forward stage while traveling along a curve; under the fourth operating condition, the operating stages of the loader include, in chronological order: an unloaded forward stage while traveling in a straight line, a shoveling operation stage, a loaded reverse stage while traveling along a curve, and a loaded forward stage while traveling in a straight line; under the fifth operating condition, the operating stages of the loader include, in chronological order: an unloaded forward stage, a shoveling operation stage, and an unloading stage.

[0010] In some embodiments, when the pressure value of the working pump is greater than a first preset pressure value, determining that the current operating condition is one of the first group of operating conditions includes: when a first condition is met, determining that the current operating condition is the third operating condition, wherein the first condition includes: the loader performs shoveling operations after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is not equal to the first preset displacement.

[0011] In some embodiments, when the pressure value of the working pump is greater than a first preset pressure value, determining that the current operating condition is one of the first group of operating conditions includes: when a second condition is met, determining that the current operating condition is any one of the first group of operating conditions except the third operating condition, wherein the second condition includes: the loader performs shoveling operations after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is equal to the first preset displacement.

[0012] In some embodiments, when the second condition is met, determining that the current operating condition is any one of the first group of operating conditions except the third operating condition includes at least one of the following: when the loader performs shoveling operation and then unloads, determining that the current operating condition is the fifth operating condition; when the third condition is met, determining that the current operating condition is the fourth operating condition; when the fourth condition is met, determining that the current operating condition is the second operating condition; and when the fifth condition is met, determining that the current operating condition is the first operating condition; wherein the third condition includes: the loader reverses with load after performing shoveling operation, and in the loading When the loader is moving backward with the load, the displacement of the steering cylinder is not equal to the first preset displacement; the fourth condition includes: the loader moves backward with the load and moves forward with the load in sequence after performing the shoveling operation, wherein, during the process of the loader moving backward with the load and moving forward with the load, the displacement of the steering cylinder is equal to the first preset displacement; the fifth condition includes: the loader moves backward with the load and moves forward with the load in sequence after performing the shoveling operation, wherein, during the process of the loader moving backward with the load, the displacement of the steering cylinder is equal to the first preset displacement; and during the process of the loader moving forward with the load, the displacement of the steering cylinder is not equal to the first preset displacement.

[0013] In some embodiments, determining the current operating condition of the loader based on the pressure value of the working pump of the loader includes: when the pressure value is equal to the first preset pressure value and the vehicle speed is greater than 0, determining that the current operating condition is one of a second group of operating conditions, the second group of operating conditions including one or more of a straight-line driving condition, a non-horizontal driving condition and a turning driving condition, the straight-line driving condition including at least one of a first straight-line driving condition and a second straight-line driving condition, the non-horizontal driving condition including at least one of an uphill driving condition and a downhill driving condition, the vehicle speed of the first straight-line driving condition is greater than a first preset vehicle speed, and the vehicle speed of the second straight-line driving condition is less than or equal to the first preset vehicle speed.

[0014] In some embodiments, when the pressure value is equal to the first preset pressure value and the vehicle speed is greater than 0, determining that the current operating condition is one of the second group of operating conditions includes at least one of the following: when the displacement of the steering cylinder of the loader is not equal to the first preset displacement, determining that the current operating condition is the steering driving condition; when the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is not 0, and the accelerator pedal opening is greater than the first preset opening, determining that the current operating condition is the uphill driving condition; when the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is not 0, and the accelerator pedal opening is equal to the first preset opening, determining that the current operating condition is the downhill driving condition; when the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is 0, and the vehicle speed is greater than the first preset speed, determining that the current operating condition is the first straight-line driving condition; and when the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is 0, and the vehicle speed is less than or equal to the first preset speed, determining that the current operating condition is the second straight-line driving condition.

[0015] According to another aspect of an embodiment of the present disclosure, a model training method is provided, including: obtaining working data of a first loader under a preset working condition at a third moment; obtaining the state of charge of a battery of the first loader under the preset working condition at a fourth moment and the vehicle speed at the fourth moment, the fourth moment being after the third moment; and taking the working data at the third moment as input and the state of charge at the fourth moment and the vehicle speed at the fourth moment as output, to train a machine learning model corresponding to the preset working condition, wherein the preset working condition is one of multiple working conditions, and the machine learning model corresponding to one of at least two working conditions is different from the machine learning model corresponding to another one.

[0016] In some embodiments, the working data of a second loader deployed with the machine learning model under the current working condition at a fifth moment is obtained, and the preset working condition includes the current working condition; the battery state of charge and vehicle speed of the second loader under the preset working condition at a sixth moment are obtained, and the sixth moment is after the fifth moment; and the working data at the fifth moment is used as input, the state of charge at the sixth moment and the vehicle speed at the sixth moment are used as output, to train the machine learning model corresponding to the current working condition.

[0017] In some embodiments, the second loader is different from the first loader.

[0018] In some embodiments, the working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power.

[0019] In some embodiments, the preset operating condition includes a straight-line driving condition, and the straight-line driving condition includes a first straight-line driving condition and a second straight-line driving condition. The vehicle speed of the first straight-line driving condition is greater than a first preset speed, and the vehicle speed of the second straight-line driving condition is less than or equal to the first preset speed. The machine learning model corresponding to the first straight-line driving condition is different from the machine learning model corresponding to the second straight-line driving condition.

[0020] In some embodiments, the working data also includes a ramp angle, the preset operating condition includes a non-horizontal driving condition, the non-horizontal driving condition includes at least one of an uphill driving condition and a downhill driving condition, and the machine learning model corresponding to the uphill driving condition is different from the machine learning model corresponding to the downhill driving condition.

[0021] In some embodiments, the working data also includes one or more of a heading angle, a steering cylinder displacement, and a steering cylinder pressure, and the preset working condition includes a steering driving condition.

[0022] In some embodiments, the working data also includes one or more of bucket cylinder displacement, boom cylinder displacement, bucket cylinder pressure, boom cylinder pressure and working motor power, and the preset working conditions include at least one of the second working condition and the fifth working condition, wherein: under the second working condition, the working stages of the loader include in chronological order: an empty forward stage of driving in a straight line, a shoveling operation stage, a loaded backward stage of driving in a straight line, and a loaded forward stage of driving in a straight line; under the fifth working condition, the working stages of the loader include in chronological order: an empty forward stage, a shoveling operation stage, and an unloading stage.

[0023] In some embodiments, the preset working condition includes at least one of the first working condition, the third working condition and the fourth working condition, and the working data also includes one or more of the heading angle, the bucket cylinder displacement, the boom cylinder displacement, the steering cylinder displacement, the bucket cylinder pressure, the boom cylinder pressure, the steering cylinder pressure, and the working motor power, wherein: under the first working condition, the working stages of the loader include, in chronological order: an empty forward stage of driving along a straight line, a shoveling operation stage, a loaded backward stage of driving along a straight line, and a loaded forward stage of driving along a curve; under the third working condition, the working stages of the loader include, in chronological order: an empty forward stage of driving along a curve, a shoveling operation stage, a loaded backward stage of driving along a curve, and a loaded forward stage of driving along a curve; under the fourth working condition, the working stages of the loader include, in chronological order: an empty forward stage of driving along a straight line, a shoveling operation stage, a loaded backward stage of driving along a curve, and a loaded forward stage of driving along a straight line.

[0024] According to another aspect of the embodiments of the present disclosure, there is provided a parameter prediction device, comprising: a module configured to execute the prediction method described in any one of the above embodiments.

[0025] According to another aspect of an embodiment of the present disclosure, a model training device is provided, comprising: a module configured to execute the training method described in any one of the above embodiments.

[0026] According to another aspect of an embodiment of the present disclosure, an electronic device is provided, including: a memory; and a processor coupled to the memory, wherein the processor is configured to execute the method described in any one of the above embodiments based on instructions stored in the memory.

[0027] According to another aspect of the embodiments of the present disclosure, there is provided a loader, comprising at least one of the following: a parameter prediction device as described in any one of the above embodiments; a model training device as described in any one of the above embodiments; and an electronic device as described in any one of the above embodiments.

[0028] In some embodiments, the loader also includes: a first sensor installed on the steering cylinder of the loader, configured to obtain the steering cylinder displacement and steering cylinder pressure of the loader; a second sensor installed on the bucket cylinder of the loader, configured to obtain the bucket cylinder displacement and bucket cylinder pressure of the loader; a third sensor installed on the boom cylinder of the loader, configured to obtain the boom cylinder displacement and boom cylinder pressure of the loader; an inertial navigation system, configured to obtain the heading angle of the loader.

[0029] According to another aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, comprising computer program instructions, wherein when the computer program instructions are executed by a processor, the steps of the method described in any one of the above embodiments are implemented.

[0030] According to another aspect of the embodiments of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the steps of the method described in any one of the above embodiments are implemented.

[0031] In the disclosed embodiment, the current operating condition of the loader is one of multiple operating conditions, and the machine learning models corresponding to at least two different operating conditions are different. By inputting the working data of the loader at the first moment under the current operating condition into the machine learning model corresponding to the current operating condition of the loader, the SOC of the battery of the loader at the second moment after the first moment and the vehicle speed of the loader at the second moment can be predicted, thereby accurately predicting the SOC of the battery and the vehicle speed of the loader after the current moment under the current operating condition.

[0032] The technical solution of the present disclosure is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 It is a flowchart of a parameter prediction method according to some embodiments of the present disclosure.

[0035] Figure 2A is a schematic diagram of a first operating condition according to some embodiments of the present disclosure.

[0036] Figure 2B is a schematic diagram of a second operating condition according to some embodiments of the present disclosure.

[0037] Figure 2C is a schematic diagram of a third operating condition according to some embodiments of the present disclosure.

[0038] Figure 2D is a schematic diagram of a fourth operating condition according to some embodiments of the present disclosure.

[0039] Figure 3 It is a flowchart of a method for training a model according to some embodiments of the present disclosure.

[0040] Figure 4 It is a schematic diagram of the structure of a parameter prediction device according to some embodiments of the present disclosure.

[0041] Figure 5 It is a schematic diagram of the structure of a training device for a model according to some embodiments of the present disclosure.

[0042] Figure 6 is a schematic diagram of the structure of an electronic device according to some embodiments of the present disclosure. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.

[0044] Unless specifically stated otherwise, the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure.

[0045] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not necessarily drawn according to the actual proportional relationship.

[0046] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0047] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0048] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0049] In addition, in the description of the present disclosure, the terms "first", "second", "third", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance and order. Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that such operations be performed in the particular order shown or in a sequential order, or requiring that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing can be advantageous.

[0050] Figure 1It is a flowchart of a parameter prediction method according to some embodiments of the present disclosure.

[0051] In step 102, the current operating condition of the loader is determined.

[0052] In some embodiments, the current working condition of the loader may be determined based on one or more of the working route and working parameters of the loader during operation.

[0053] In step 104, the working data of the loader in the current working condition at the current first moment is obtained.

[0054] It should be understood that the working data is data generated when the loader is working, and the working data at the current moment is correlated with the SOC of the battery of the loader and the vehicle speed of the loader at a moment after the current moment.

[0055] In step 106, the working data of the loader in the current working condition at the current first moment is input into a machine learning model corresponding to the current working condition to predict the SOC of the battery of the loader at the second moment and the vehicle speed of the loader at the second moment.

[0056] Here, the second moment is after the first moment. The second moment can be any moment after the first moment.

[0057] That is to say, there are at least two operating conditions among the multiple operating conditions in which the machine learning models are different, the current operating condition is one of the multiple operating conditions of the loader, and the machine learning model corresponding to one of the at least two operating conditions is different from the machine learning model corresponding to another of the at least two operating conditions.

[0058] As some implementation methods, there is a preset time interval between the second moment and the first moment. For example, the preset time interval is a fixed time interval; for another example, the preset time interval is a dynamically changing time interval.

[0059] In the above embodiment, the current operating condition of the loader is one of multiple operating conditions, and the machine learning models corresponding to at least two different operating conditions are different. By inputting the working data of the loader at the first moment under the current operating condition into the machine learning model corresponding to the current operating condition of the loader, the SOC of the battery of the loader at the second moment after the first moment and the vehicle speed of the loader at the second moment can be predicted, thereby accurately predicting the SOC of the battery and the vehicle speed of the loader after the current moment under the current operating condition.

[0060] In some embodiments, the working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power.

[0061] As some implementation methods, the vehicle speed, gear position, accelerator pedal opening, brake pedal opening and travel motor power may be obtained by using a controller area network (CAN) bus of the loader.

[0062] For example, the working data includes vehicle speed, gear, accelerator pedal opening, brake pedal opening or travel motor power; for another example, the working data includes any two of vehicle speed, gear, accelerator pedal opening, brake pedal opening and travel motor power; for another example, the working data includes any three of vehicle speed, gear, accelerator pedal opening, brake pedal opening and travel motor power; for another example, the working data includes any four of vehicle speed, gear, accelerator pedal opening, brake pedal opening and travel motor power; for another example, the working data includes vehicle speed, gear, accelerator pedal opening, brake pedal opening and travel motor power.

[0063] In the above embodiment, the working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power. By establishing a correlation between such working data and the SOC of the battery and the speed of the loader, the machine learning model can more accurately predict the SOC of the battery and the speed of the loader.

[0064] As some embodiments, the multiple working conditions include working conditions and driving conditions. It is understandable that in the working condition, the loader not only performs work (for example, shoveling, unloading, etc.), but also drives; in the driving condition, the loader only drives and does not perform other additional work.

[0065] As some embodiments, the operating conditions may include one or more of a first operating condition, a second operating condition, a third operating condition, a fourth operating condition, and a fifth operating condition.

[0066] For example, the first operating condition may be a V-type operating condition; for another example, the second operating condition may be an I-type operating condition; for another example, the third operating condition may be a T-type operating condition; for another example, the fourth operating condition may be an L-type operating condition; for another example, the fifth operating condition may be a stacking operating condition.

[0067] As some embodiments, the driving condition may include one or more of a straight-line driving condition, a non-level driving condition, and a turning driving condition. For example, the straight-line driving condition includes at least one of a high-speed straight-line driving condition and a low-speed straight-line driving condition; for another example, the non-level driving condition includes at least one of an uphill driving condition and a downhill driving condition.

[0068] Next, combine FIG. 2A to FIG. 2D The first working condition, the second working condition, the third working condition and the fourth working condition in the operation working condition are introduced. Here, the loader 202 needs to load materials from the working platform 201 and unload the materials to the dump truck 203.

[0069] Figure 2A is a schematic diagram of a first operating condition according to some embodiments of the present disclosure.

[0070] like Figure 2A As shown, in the first working condition, the operation phases of the loader 202 include, in chronological order: an unloaded forward phase 2021a of traveling along a straight line, a shovel loading operation phase (not shown), a loaded backward phase 2022a of traveling along a straight line, and a loaded forward phase 2023a of traveling along a curve. In other embodiments, in the first working condition, the operation phases of the loader 202 also include an unloaded backward phase 2024a of traveling along a curve after the loaded forward phase 2023a of traveling along a curve.

[0071] As some embodiments, loaded may refer to the presence of materials in the bucket of the loader, and unloaded may refer to the absence of materials in the bucket of the loader.

[0072] As some implementation modes, the shovel loading operation may refer to a process in which a bucket of a loader is loaded with materials.

[0073] As some embodiments, there is an angle α between the center line of the dump truck 203 and the working platform 201. For example, α is 50 degrees to 55 degrees.

[0074] As some embodiments, in the loaded reverse phase 2022a of traveling in a straight line, the loader 202 reverses and drives away from the working platform 201 .

[0075] As some embodiments, after the loaded backward phase 2022a of traveling in a straight line and before the loaded forward phase 2023a of traveling in a curve, the front frame of the loader 202 turns toward the dump truck 203. For example, the turning angle is 35 to 45 degrees.

[0076] Figure 2B is a schematic diagram of a second operating condition according to some embodiments of the present disclosure.

[0077] like Figure 2B As shown, in the second working condition, the operation phases of the loader 202 include, in chronological order: an unloaded forward phase 2021b of traveling in a straight line, a shovel loading operation phase (not shown), a loaded backward phase 2022b of traveling in a straight line, and a loaded forward phase 2023b of traveling in a straight line. In other embodiments, in the second working condition, the operation phases of the loader 202 also include an unloaded backward phase 2024b of traveling in a straight line after the loaded forward phase 2023b of traveling in a straight line.

[0078] As some embodiments, the dump truck 203 moves forward and backward reciprocatingly, and the direction of the front of the dump truck 203 intersects with the direction of the front of the loader 202. The angle of intersection is, for example, 90 degrees.

[0079] As some embodiments, in the loaded backward phase 2022b of traveling in a straight line, the loader 202 reverses and drives away from the working platform 201. As some other embodiments, in the unloaded backward phase 2024b of traveling in a straight line, the loader 202 reverses and drives away from the dump truck 203.

[0080] Figure 2C is a schematic diagram of a third operating condition according to some embodiments of the present disclosure.

[0081] like Figure 2C As shown, under the third working condition, the operation phases of the loader 202 include, in chronological order: an unloaded forward phase 2021c of traveling along a curve, a shovel loading operation phase (not shown), a loaded backward phase 2022c of traveling along a curve, and a loaded forward phase 2023c of traveling along a curve. In other embodiments, under the third working condition, the operation phases of the loader 202 also include an unloaded backward phase 2024c of traveling along a curve after the loaded forward phase 2023c of traveling along a curve.

[0082] As some embodiments, the direction of the front of the dump truck 203 is parallel to the direction of the front of the loader 202 at the beginning of the unloaded forward phase 2021c of driving along the curve, and the distance between the center of gravity of the dump truck 203 and the working platform 201 is greater than the distance between the center of gravity of the loader 202 and the working platform 201.

[0083] Figure 2D is a schematic diagram of a fourth operating condition according to some embodiments of the present disclosure.

[0084] like Figure 2D As shown, in the fourth working condition, the operation phases of the loader 202 include, in chronological order: an unloaded forward phase 2021d of traveling along a straight line, a shovel loading operation phase (not shown), a loaded backward phase 2022d of traveling along a curve, and a loaded forward phase 2023d of traveling along a straight line. In other embodiments, in the fourth working condition, the operation phases of the loader 202 also include an unloaded backward phase 2024d of traveling along a curve after the loaded forward phase 2023d of traveling along a straight line.

[0085] As some embodiments, the direction of the front of the dump truck 203 is parallel to the direction of the front of the loader 202 at the beginning of the unloaded forward phase 2021d of traveling in a straight line.

[0086] As some embodiments, the loader 202 reverses and turns after loading the material on the working platform 201, then drives forward to the dump truck 203 to unload the material, and reverses and turns after unloading the material. The turning angle is, for example, 90 degrees.

[0087] Next, the fifth working condition is introduced in combination with some embodiments of the present disclosure. Here, in the fifth working condition, the loader needs to gather the materials at the working platform into a pile.

[0088] As some implementation modes, under the fifth working condition, the operation phases of the loader include, in chronological order: an empty forward phase, a shoveling operation phase, and an unloading phase.

[0089] Next, how to determine the operation phase of a loader is introduced in conjunction with some embodiments of the present disclosure.

[0090] Table 1

[0091] In some embodiments, as shown in Table 1, P C Indicates the current bucket cylinder pressure, P C0 Indicates the second preset pressure value, L d Indicates the current boom cylinder displacement, L d0 Indicates the second preset displacement value. For example, the second preset pressure value indicates the initial pressure value of the bucket cylinder, that is, the pressure value when the bucket cylinder is in a non-working state; for another example, the second preset displacement value indicates the initial displacement of the boom cylinder, that is, the displacement of the piston of the boom cylinder when the boom cylinder is in a non-working state.

[0092] As some embodiments, when the gear of the loader is the forward gear and P C =P C0 In this case, it is determined that the loader is in the no-load forward stage.

[0093] As some embodiments, when the gear position of the loader is forward gear, P C >P C0 , and L d =L d0 In this case, it is determined that the loader is in the shovel loading operation stage.

[0094] As some embodiments, when the gear of the loader is the reverse gear and P C >P C0 In this case, it is determined that the loader is in the loaded reverse stage.

[0095] As some embodiments, when the gear position of the loader is forward gear, P C >P C0 , and L d >Ld0 In this case, it is determined that the loader is in the forward stage with load.

[0096] As some embodiments, when the gear of the loader is the reverse gear and P C =P C0 In this case, it is determined that the loader is in the no-load reverse stage.

[0097] As some embodiments, when the gear position of the loader is neutral, P C >P C0 , and L d >L d0 In this case, it is determined that the loader is in the unloading stage.

[0098] In the above embodiment, the operating stage of the loader can be accurately determined through the gear position, bucket cylinder pressure and boom cylinder displacement of the loader, which helps to accurately determine the current working condition of the loader in combination with the operating stage of the loader, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader.

[0099] In some embodiments, the machine learning models corresponding to the multiple operating conditions are different from each other. That is, each operating condition has a machine learning model unique to that operating condition. In this way, the SOC of the battery and the speed of the loader under each operating condition can be predicted more accurately.

[0100] In some embodiments, the current working condition of the loader is determined according to the pressure value of the working pump of the loader, wherein the working pump is used to drive at least one of the steering cylinder of the loader, the bucket cylinder of the loader and the boom cylinder of the loader.

[0101] For example, the working pump is used to drive the steering cylinder of the loader, the bucket cylinder of the loader or the boom cylinder of the loader; for another example, the working pump is used to drive any two of the steering cylinder of the loader, the bucket cylinder of the loader and the boom cylinder of the loader; for another example, the working pump is used to drive the steering cylinder of the loader, the bucket cylinder of the loader and the boom cylinder of the loader.

[0102] It should be understood that the steering cylinder of the loader is used to drive the steering of the loader, the bucket cylinder of the loader is used to drive the lifting and tilting of the bucket, and the boom cylinder of the loader is used to drive the movement of the mechanical arm of the loader.

[0103] In the above embodiment, based on the pressure value of the loader's working pump, it can be determined whether at least one of the loader's steering cylinder, the loader's bucket cylinder, and the loader's boom cylinder is in a driven state, which helps to determine the loader's current working condition, so that the machine learning model can accurately predict the battery's SOC and the loader's vehicle speed.

[0104] In some embodiments, when the pressure value of the working pump is greater than a first preset pressure value, the current operating condition is determined to be one of the first group of operating conditions, where the first group of operating conditions includes one or more of the first operating condition, the second operating condition, the third operating condition, the fourth operating condition and the fifth operating condition.

[0105] For example, the first group of operating conditions includes the first operating condition, the second operating condition, the third operating condition, the fourth operating condition or the fifth operating condition. For another example, the first group of operating conditions includes any two, any three, any four or all of the first operating condition, the second operating condition, the third operating condition, the fourth operating condition and the fifth operating condition.

[0106] As some embodiments, the first preset pressure value is an initial pressure value of the working pump, that is, a pressure value when the loader is in a non-working state. That is, when the pressure value of the working pump is greater than the first preset pressure value, it is determined that the working pump is in a working state, that is, the working pump is in a state of driving at least one of the steering cylinder of the loader, the bucket cylinder of the loader, and the boom cylinder of the loader.

[0107] In the above embodiment, through the relationship between the pressure value of the working pump and the first preset pressure value, it is possible to accurately determine whether at least one of the steering cylinder of the loader, the bucket cylinder of the loader and the boom cylinder of the loader is in a driven state, thereby helping to accurately determine the current working condition of the loader, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader.

[0108] Next, how to determine which operating condition in the first group of operating conditions the current operating condition belongs to is introduced in conjunction with some embodiments of the present disclosure.

[0109] In some embodiments, when the first condition is met, the current operating condition is determined to be the third operating condition, for example, the current operating condition is determined to be a T-type operating condition.

[0110] Here, the first condition includes: the loader performs shovel loading after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is not equal to the first preset displacement. For example, the first preset displacement is the initial displacement of the steering cylinder, that is, the displacement of the piston of the steering cylinder when the steering cylinder is in a non-working state.

[0111] In the above embodiment, when it is determined that the loader enters the shoveling operation stage from the no-load forward stage of driving along a curve, it can be accurately determined that the current operating condition of the loader is the third operating condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for this operating condition.

[0112] In some embodiments, when the second condition is met, the current operating condition is determined to be any operating condition in the first group of operating conditions except the third operating condition.

[0113] Here, the second condition includes: the loader performs shovel loading after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is equal to the first preset displacement.

[0114] That is to say, in the other working conditions except the third working condition in the first group of working conditions, the loader enters the shovel loading operation stage after the no-load forward stage, and in the no-load forward stage, the displacement of the steering cylinder of the loader is equal to the first preset displacement. That is, the other working conditions (in the case of including multiple working conditions) in the first group of working conditions except the third working condition have commonalities.

[0115] In the above embodiment, when it is determined that the current operating condition of the loader is not the third operating condition, the machine learning model of any other operating condition except the third operating condition can be used to quickly predict the battery SOC and the vehicle speed of the loader.

[0116] Next, some embodiments are used to explain how to further determine which operating condition the loader's current operating condition is when it is determined that the loader's current operating condition is not the third operating condition.

[0117] In some embodiments, when the second condition is met and the loader performs shovel loading and then unloading, the current working condition is determined to be the fifth working condition. For example, the current working condition is determined to be a stacking working condition.

[0118] In the above embodiment, when it is determined that the loader enters the shoveling operation stage from the unloaded forward stage of traveling in a straight line, and enters the unloading stage from the shoveling operation stage, it can be accurately determined that the current operating condition of the loader is the stacking operation condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for this operating condition.

[0119] In some embodiments, when the second condition and the third condition are met, the current operating condition is determined to be the fourth operating condition, for example, the current operating condition is determined to be the L-type operating condition.

[0120] Here, the third condition includes: the loader moves backward with the load after performing the shovel loading operation, and during the process of the loader moving backward with the load, the displacement of the steering cylinder is not equal to the first preset displacement.

[0121] In the above embodiment, when it is determined that the loader enters the shoveling operation stage from the unloaded forward stage of traveling in a straight line, and enters the loaded reverse stage of traveling along a curve from the shoveling operation stage, it can be accurately determined that the current operating condition of the loader is the fourth operating condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for this operating condition.

[0122] In some embodiments, when the second condition and the fourth condition are met, the current operating condition is determined to be the second operating condition, for example, the current operating condition is determined to be a Type I operating condition.

[0123] Here, the fourth condition includes: the loader sequentially moves backward and forward with the load after performing the shovel loading operation. Here, during the process of the loader moving backward and forward with the load, the displacement of the steering cylinder is equal to the first preset displacement.

[0124] In the above embodiment, when it is determined that the loader enters the shoveling operation stage from the unloaded forward stage of traveling in a straight line, enters the loaded reverse stage of traveling in a straight line from the shoveling operation stage, and enters the loaded forward stage of traveling in a straight line from the loaded reverse stage, it can be accurately determined that the current operating condition of the loader is the second operating condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for this operating condition.

[0125] In some embodiments, when the second condition and the fifth condition are met, the current operating condition is determined to be the first operating condition, for example, the current operating condition is determined to be a V-type operating condition.

[0126] Here, the fifth condition includes: the loader performs backward movement and forward movement with load in sequence after performing shovel loading. Here, during the backward movement of the loader with load, the displacement of the steering cylinder is equal to the first preset displacement; and during the forward movement of the loader with load, the displacement of the steering cylinder is not equal to the first preset displacement.

[0127] In the above embodiment, when it is determined that the loader enters the shoveling operation stage from the unloaded forward stage of traveling in a straight line, enters the loaded reverse stage of traveling in a straight line from the shoveling operation stage, and enters the loaded forward stage of traveling along a curve from the loaded reverse stage, it can be accurately determined that the current operating condition of the loader is the first operating condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for this operating condition.

[0128] It should be understood that, when the second condition is met, the current operating condition can be determined as an operating condition other than the third operating condition in the first group of operating conditions by combining the above embodiments.

[0129] In some embodiments, when the pressure value of the working pump is equal to the first preset pressure value and the vehicle speed of the loader is greater than 0, it is determined that the current working condition of the loader is one of the second group of working conditions. The second group of working conditions includes one or more of a straight-line driving condition, a non-level driving condition, and a turning driving condition.

[0130] Here, the straight-line driving condition includes at least one of a first straight-line driving condition and a second straight-line driving condition, the non-level driving condition includes at least one of an uphill driving condition and a downhill driving condition, the vehicle speed of the first straight-line driving condition is greater than the first preset vehicle speed, and the vehicle speed of the second straight-line driving condition is less than or equal to the first preset vehicle speed. For example, the first straight-line driving condition may be referred to as a high-speed straight-line driving condition, and the second straight-line driving condition may be referred to as a low-speed straight-line driving condition.

[0131] In the above embodiment, through the relationship between the pressure value of the working pump and the first preset pressure value, it is possible to accurately determine whether the loader is in one of the straight-line driving condition, non-horizontal driving condition and turning driving condition, which helps to accurately determine the current condition of the loader, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader.

[0132] In some embodiments, when the pressure value of the working pump is equal to the first preset pressure value, the speed of the loader is greater than 0, and the displacement of the steering cylinder of the loader is not equal to the first preset displacement, the current working condition is determined to be a steering driving condition.

[0133] In some embodiments, when the pressure value of the working pump is equal to the first preset pressure value, the speed of the loader is greater than 0, the displacement of the steering cylinder is equal to the first preset displacement, the slope angle is not 0, and the accelerator pedal opening is greater than the first preset opening, the current working condition is determined to be an uphill driving condition. For example, the first preset opening is the initial opening of the accelerator pedal, that is, the opening of the accelerator pedal when the accelerator pedal is not depressed.

[0134] In some embodiments, when the pressure value of the working pump is equal to a first preset pressure value, the vehicle speed of the loader is greater than 0, the displacement of the steering cylinder is equal to a first preset displacement, the ramp angle is not 0, and the accelerator pedal opening is equal to the first preset opening, the current operating condition is determined to be a downhill driving condition.

[0135] In some embodiments, when the pressure value of the working pump is equal to a first preset pressure value, the vehicle speed of the loader is greater than 0, the displacement of the steering cylinder is equal to a first preset displacement, the ramp angle is 0, and the vehicle speed is greater than a first preset speed, the current operating condition is determined to be a first straight-line driving condition.

[0136] In some embodiments, when the pressure value of the working pump is equal to a first preset pressure value, the vehicle speed of the loader is greater than 0, the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is 0, and the vehicle speed of the loader is less than or equal to the first preset speed, the current operating condition is determined to be the second straight-line driving condition.

[0137] It should be understood that the current operating condition may be determined as one of the second group of operating conditions by combining the above embodiments.

[0138] In the above embodiments, the current operating condition of the loader can be accurately determined as a turning driving condition, a first straight-line driving condition, a second straight-line driving condition, an uphill driving condition or a downhill driving condition, so that the machine learning model can more accurately predict the battery SOC and the vehicle speed of the loader for these conditions.

[0139] In some embodiments, the operating data and current working condition of the loader may be sent to the user, so as to help the user understand the working status of the loader.

[0140] In some embodiments, an energy management strategy for the loader can be formulated based on the predicted battery SOC and the speed of the loader. For example, the energy management strategy includes a power distribution strategy and a battery charging and discharging strategy. In this way, based on the battery SOC and the speed of the loader predicted by the prediction method of the present disclosure, an energy management strategy for the loader can be accurately formulated, thereby helping to accurately implement energy management of the loader.

[0141] The disclosed embodiment also proposes a model training method.

[0142] Figure 3 It is a flowchart of a method for training a model according to some embodiments of the present disclosure.

[0143] In step 302, the operating data of the first loader under the preset operating conditions at the third moment is obtained.

[0144] The first loader may be any loader, and the preset working condition may be any working condition of the first loader. The working data at the third moment is used as a training sample for training the machine learning model.

[0145] In step 304, the SOC of the battery and the vehicle speed of the first loader at the fourth moment under the preset working condition are obtained. Here, the fourth moment is after the third moment.

[0146] The third moment may be any moment in the process of the first loader performing historical operations under the preset working conditions, and the fourth moment may be any moment after the third moment in the process of the first loader performing historical operations under the preset working conditions. The SOC of the battery at the fourth moment and the vehicle speed at the fourth moment are used as training samples for training the machine learning model.

[0147] In step 306, the working data at the third moment is used as input, the SOC at the fourth moment and the vehicle speed at the fourth moment are used as output, and the machine learning model corresponding to the preset working condition is trained.

[0148] Here, the preset working condition is one of the multiple working conditions, and the machine learning model corresponding to one of at least two working conditions is different from the machine learning model corresponding to the other one. In other words, the machine learning models of at least two working conditions are different.

[0149] In the process of training the machine learning model, the training samples can be increased in a variety of ways, thereby training the machine learning model and improving the training effect.

[0150] For example, for the same loader, at least one of the third moment and the fourth moment can be changed to change the training samples for training the machine learning model.

[0151] For another example, for the same loader, the type of preset working conditions can be changed to change the training samples for training the machine learning model.

[0152] For another example, multiple loaders can be used as first loaders, and steps 302 to 306 can be repeatedly performed to train the machine learning model corresponding to the preset working condition. By changing the type of the preset working condition, the training samples of different loaders can be used to train the machine learning models corresponding to different working conditions.

[0153] In the above embodiment, the working data of the first loader under the preset working condition at the third moment is used as input, and the SOC of the battery of the first loader under the preset working condition at the fourth moment and the vehicle speed at the fourth moment are used as output, and the machine learning model corresponding to the preset working condition is trained, and the preset working condition is one of multiple working conditions, and the machine learning model corresponding to one of at least two working conditions is different from the machine learning model corresponding to another. In this way, the machine learning model trained for the preset working condition can more accurately predict the SOC of the battery and the vehicle speed of the loader under the working condition.

[0154] In some embodiments, working data of the first loader at a third moment under multiple random working conditions (hereinafter referred to as random working data) may be obtained, and the random working data under multiple working conditions may be reduced in dimension and clustered to obtain working data of the first loader at the third moment under each working condition. For example, principal component analysis (PCA) may be used to reduce the dimension of the random working data.

[0155] As some implementation methods, the random operating data includes operating data of the loader under various operating conditions that occur randomly during actual use or testing.

[0156] As some implementation methods, the random working data after dimensionality reduction can be clustered to obtain multiple cluster sets, and the characteristics of the multiple cluster sets can be analyzed to determine the type of working condition. After the type of working condition is determined, the working condition can be used as a preset working condition.

[0157] For example, the operating data in the cluster set can be analyzed to determine the type of operating condition.

[0158] In the above embodiments, on the one hand, the working data of the first loader at the third moment is obtained by utilizing random working data reflecting the random working conditions of the loader in daily operation, so as to accurately train the machine learning model; on the other hand, the working data under different working conditions can be obtained by performing dimensionality reduction and clustering on the random working data, and there is no need for the loader to obtain working data separately under different working conditions, which can reduce the pressure of data processing and improve the efficiency of training the machine learning model.

[0159] In some embodiments, the machine learning model corresponding to one of any two working conditions in the multiple working conditions is different from the machine learning model corresponding to the other working condition. In other words, each working condition in the multiple working conditions has a machine learning model unique to the working condition, and the machine learning models corresponding to the multiple working conditions are different from each other. In this way, the machine learning model trained for each working condition can more accurately predict the SOC and vehicle speed of the battery of the loader under the working condition.

[0160] It is understandable that after the machine learning model is trained (e.g., offline training), the trained machine learning model can be deployed on different loaders. The loader can use the trained machine learning model to predict the SOC and vehicle speed of the battery of the loader at a certain moment. Afterwards, the training data of the loader in a certain working condition can be obtained, and the training data can be used to further train the machine learning model deployed on the loader. This will be explained in conjunction with some embodiments.

[0161] In some embodiments, the working data of the second loader deployed with the machine learning model at the fifth moment under the current working condition is obtained, and the SOC of the battery and the vehicle speed of the second loader at the sixth moment after the fifth moment under the preset working condition are obtained. After obtaining the training samples, the working data at the fifth moment is used as input, and the SOC at the sixth moment and the vehicle speed at the sixth moment are used as output to train the machine learning model corresponding to the current working condition.

[0162] For example, the preset working condition includes the current working condition. In this way, the machine learning model corresponding to a certain working condition can be trained before and after being deployed on the loader. In other words, the second loader equipped with the machine learning model can update the machine learning model online when in the current working condition. For example, at least one of the weight matrix and the bias vector in the machine learning model is updated.

[0163] As some implementations, a training window of size A may be set as the duration for collecting working data, and n online training iterations may be performed. For example, n is 1.

[0164] When the cumulative running time of the loader reaches the preset threshold T, the machine learning model is updated. Here, T=A*n.

[0165] After updating the machine learning model, increment the training counter n (i.e. n=n+1).

[0166] For example, when n is 1, when the cumulative running time of the loader reaches A, the machine learning model is updated and the training times counter is incremented (n is 2 at this time); when the loader runs for another time A (that is, the cumulative running time reaches 2A), the machine learning model is updated again and the training times counter is incremented, and so on.

[0167] When the cumulative running time of the loader does not reach the preset threshold T, the machine learning model is not updated, that is, the original machine learning model is used for prediction.

[0168] In the above embodiment, online updating of the machine learning model can continuously optimize the machine learning model, enhance the adaptability and flexibility of the machine learning model to different working conditions, and thus more accurately predict the SOC and vehicle speed of the loader's battery.

[0169] In some embodiments, the second loader is different from the first loader. That is, the machine learning model can be trained using the working data of the first loader, and after the machine learning model is deployed on the second loader, the machine learning model is updated with the working data of the current working condition when the second loader is actually working.

[0170] In the above embodiment, on the one hand, the machine learning model can be trained with the working data of the first loader, thereby speeding up the efficiency of training the machine learning model; on the other hand, the machine learning model trained with the working data of the first loader can be updated with the working data of the second loader, so that the updated machine learning model is more in line with the actual working conditions of the second loader, thereby more accurately predicting the SOC and vehicle speed of the battery of the second loader.

[0171] In some embodiments, the model of the second loader is the same as that of the first loader. In this way, the machine learning model trained using the working data of the first loader is more suitable for the second loader, so that the machine learning model can more accurately predict the SOC and vehicle speed of the battery of the second loader.

[0172] In some embodiments, the working data used as training samples include one or more of the vehicle speed of the loader, the gear position of the loader, the accelerator pedal opening of the loader, the brake pedal opening of the loader, and the travel motor power of the loader. Here, the travel motor of the loader is used to drive the loader to travel.

[0173] For example, the working data includes any one, any two, any three, any four or all of the loader's speed, the loader's gear position, the loader's accelerator pedal opening, the loader's brake pedal opening and the loader's travel motor power.

[0174] In the above embodiment, the working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power, which can more accurately train the machine learning model, so that the trained machine learning model can more accurately predict the SOC and vehicle speed of the loader's battery.

[0175] In some embodiments, the preset working condition includes a straight-line driving working condition. Here, the straight-line driving working condition includes a first straight-line driving working condition and a second straight-line driving working condition, the vehicle speed of the first straight-line driving working condition is greater than the first preset vehicle speed, and the vehicle speed of the second straight-line driving working condition is less than or equal to the first preset vehicle speed. In this way, the machine learning model corresponding to the straight-line driving working condition can be trained using training data under driving working conditions of different vehicle speeds, which helps to improve the training effect.

[0176] As some implementations, the machine learning model corresponding to the first straight-line driving condition is different from the machine learning model corresponding to the second straight-line driving condition. In this way, different machine learning models can be trained for the first straight-line driving condition and the second straight-line driving condition, respectively, so that the trained machine learning model can more accurately predict the SOC and vehicle speed of the battery of the loader under the two straight-line driving conditions.

[0177] In some embodiments, the preset working condition includes a non-horizontal driving condition. Here, the non-horizontal driving condition includes at least one of an uphill driving condition and a downhill driving condition. In this case, the working data includes not only one or more of the vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power, but also a ramp angle. In this way, using the working data including the ramp angle to train the machine learning model corresponding to the non-horizontal driving condition helps to improve the training effect.

[0178] As some implementations, the machine learning model corresponding to the uphill driving condition is different from the machine learning model corresponding to the downhill driving condition. In this way, different machine learning models can be trained for the uphill driving condition and the downhill driving condition, respectively, so that the trained machine learning model can more accurately predict the SOC and vehicle speed of the battery of the loader under the corresponding conditions.

[0179] In some embodiments, the preset working condition includes a steering driving condition. In this case, the working data includes one or more of the vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power, as well as one or more of the heading angle, steering cylinder displacement, and steering cylinder pressure.

[0180] For example, the working data includes any one, any two or all of the heading angle, steering cylinder displacement and steering cylinder pressure. As some implementations, the working data includes vehicle speed, gear position, accelerator pedal opening, brake pedal opening, travel motor power, heading angle, steering cylinder displacement and steering cylinder pressure.

[0181] In the above embodiment, working data including one or more of the heading angle, steering cylinder displacement and steering cylinder pressure are used as training samples to train the machine learning model corresponding to the steering driving condition. This allows the trained machine learning model to more accurately predict the SOC and vehicle speed of the loader's battery under the steering driving condition.

[0182] In some embodiments, the preset working condition includes at least one of the second working condition and the fifth working condition. In this case, the working data includes one or more of the following: the displacement of the bucket cylinder, the displacement of the boom cylinder, the pressure of the bucket cylinder, the pressure of the boom cylinder, and the power of the working motor, in addition to one or more of the vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power. For example, the preset working condition includes the second working condition or the fifth working condition; for another example, the preset working condition includes the second working condition and the fifth working condition.

[0183] For example, in addition to vehicle speed, gear position, accelerator pedal opening, brake pedal opening and travel motor power, the working data also includes any one, any two, any three, any four or all of the bucket cylinder displacement, boom cylinder displacement, bucket cylinder pressure, boom cylinder pressure and working motor power.

[0184] In the above embodiment, at least one corresponding machine learning model of the second working condition and the fifth working condition is trained using working data including one or more of bucket cylinder displacement, boom cylinder displacement, bucket cylinder pressure, boom cylinder pressure and working motor power, so that the trained machine learning model can more accurately predict the SOC and vehicle speed of the battery of the loader under the corresponding working condition.

[0185] In some embodiments, the preset working condition includes at least one of the first working condition, the third working condition and the fourth working condition. In this case, the working data includes one or more of the vehicle speed, gear position, accelerator pedal opening, brake pedal opening and travel motor power, and also includes one or more of the heading angle, bucket cylinder displacement, boom cylinder displacement, steering cylinder displacement, bucket cylinder pressure, boom cylinder pressure, steering cylinder pressure and working motor power.

[0186] For example, the preset operating conditions include the first operating condition, the third operating condition or the fourth operating condition; for another example, the preset operating conditions include any two of the first operating condition, the third operating condition and the fourth operating condition; for another example, the preset operating conditions include the first operating condition, the third operating condition and the fourth operating condition.

[0187] It should be understood that the working data includes any one, any two, any three, any four, any five, any six, any seven or all of the heading angle, bucket cylinder displacement, boom cylinder displacement, steering cylinder displacement, bucket cylinder pressure, boom cylinder pressure, steering cylinder pressure, and working motor power.

[0188] In the above embodiment, at least one corresponding machine learning model of the first working condition, the third working condition and the fourth working condition is trained using one or more working data including heading angle, bucket cylinder displacement, boom cylinder displacement, steering cylinder displacement, bucket cylinder pressure, boom cylinder pressure, steering cylinder pressure and working motor power (for example, machine learning models can be trained for V-type working condition, T-type working condition and L-type working condition respectively), so that the trained machine learning model can more accurately predict the SOC and vehicle speed of the battery of the loader under the corresponding working condition.

[0189] In some embodiments, the machine learning model includes, but is not limited to, a hierarchical recurrent neural network (HRNN) model.

[0190] As some embodiments, the working data The HRNN model is constructed and trained with the SOC of the loader's battery and the loader's vehicle speed as input and output.

[0191] For example, the hidden layer of the HRNN model has two layers. The first layer is a recurrent neural network (RNN) expressed as .

[0192] here, is the first layer of RNN at time step t The hidden state of and is the weight matrix, is the bias vector, is the activation function.

[0193] For example, the second hidden layer of the HRNN model is RNN, and the expression is:

[0194] .

[0195] In the above formula, is the activation value of the forget gate, is the activation value of the input gate, is the activation value of the output gate, is the candidate cell state, is the cell state, is the hidden state at the current time step, represents element-wise multiplication, , , is the weight matrix, , , Bias vector, is the activation function.

[0196] For example, the output layer of the HRNN model can be expressed as:

[0197] .

[0198] In the above formula, is the last hidden layer of the RNN at the last time step T The hidden state of , is the weight matrix, , is the bias vector.

[0199] In the above embodiment, HRNN can process sequence data with a complex hierarchical structure, and the hierarchical structure can more sensitively capture the hierarchical relationships and dependencies in the work data, thereby improving the accuracy of the machine learning model in predicting the SOC and vehicle speed of the loader.

[0200] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0201] In some embodiments, the parameter prediction device includes a module configured to execute the parameter prediction method of any one of the above embodiments.

[0202] Figure 4 It is a schematic diagram of the structure of a parameter prediction device according to some embodiments of the present disclosure.

[0203] In some embodiments, Figure 4 As shown, the parameter prediction device includes a determination module 401 , an acquisition module 402 and a prediction module 403 .

[0204] The determination module 401 is configured to determine the current working condition of the loader.

[0205] The acquisition module 402 is configured to acquire the working data of the loader in the current working condition at the current first moment.

[0206] The prediction module 403 is configured to input the working data into the machine learning model corresponding to the current working condition to predict the state of charge of the battery of the loader at the second moment and the vehicle speed at the second moment. Here, the second moment is after the first moment, the current working condition is one of the multiple working conditions, and the machine learning model corresponding to one of at least two working conditions is different from the machine learning model corresponding to the other one.

[0207] In some embodiments, the parameter prediction device may further include other modules to execute the parameter prediction method of any of the above embodiments.

[0208] In some embodiments, the model training device includes a module configured to execute the model training method of any one of the above embodiments.

[0209] Figure 5 It is a schematic diagram of the structure of a training device for a model according to some embodiments of the present disclosure.

[0210] In some embodiments, Figure 5 As shown, the model training device includes a first acquisition module 501, a second acquisition module 502 and a training module 503.

[0211] The first acquisition module 501 is configured to acquire working data of the first loader under a preset working condition at a third moment.

[0212] The second acquisition module 502 is configured to acquire the state of charge of the battery and the vehicle speed of the first loader at a fourth moment under a preset working condition. Here, the fourth moment is after the third moment.

[0213] The training module 503 is configured to take the working data at the third moment as input, the state of charge at the fourth moment and the vehicle speed at the fourth moment as output, and train the machine learning model corresponding to the preset working condition. Here, the preset working condition is one of the multiple working conditions, and the machine learning model corresponding to one of at least two working conditions is different from the machine learning model corresponding to the other one.

[0214] In some embodiments, the model training device may further include other modules to execute the model training method of any of the above embodiments.

[0215] Figure 6 is a schematic diagram of the structure of an electronic device according to some embodiments of the present disclosure.

[0216] like Figure 6 As shown, the electronic device 600 includes a memory 601 and a processor 602 coupled to the memory 601, and the processor 602 is configured to execute the method of any of the aforementioned embodiments based on the instructions stored in the memory 601. As some implementation methods, the processor 602 is configured to execute the parameter prediction method or model training method of any of the aforementioned embodiments based on the instructions stored in the memory 601.

[0217] The memory 601 may include, for example, a system memory, a fixed non-volatile storage medium, etc. The system memory may store, for example, an operating system, an application program, a boot loader, and other programs.

[0218] In some embodiments, the electronic device 600 may further include an input / output interface 603, a network interface 604, a storage interface 605, etc. The input / output interface 603, the network interface 604, the storage interface 605, and the memory 601 and the processor 602 may be connected, for example, via a bus 606. The input / output interface 603 provides a connection interface for input / output devices such as a display, a mouse, a keyboard, and a touch screen. The network interface 604 provides a connection interface for various networked devices. The storage interface 605 provides a connection interface for external storage devices such as an SD card and a USB flash drive.

[0219] The disclosed embodiments further provide a loader, comprising at least one of the following: a parameter prediction device of any one of the above embodiments; a model training device of any one of the above embodiments; and an electronic device of any one of the above embodiments.

[0220] In some embodiments, the loader further includes a first sensor, a second sensor, a third sensor, and an inertial navigation system.

[0221] Here, the first sensor is installed on the steering cylinder of the loader, and the first sensor is configured to obtain the steering cylinder displacement and steering cylinder pressure of the loader.

[0222] The second sensor is installed on the bucket cylinder of the loader, and the second sensor is configured to obtain the bucket cylinder displacement and bucket cylinder pressure of the loader.

[0223] The third sensor is installed on the boom cylinder of the loader, and the third sensor is configured to obtain the displacement and pressure of the boom cylinder of the loader.

[0224] The inertial navigation system is configured to acquire the heading angle of the loader.

[0225] For example, the first sensor includes a first displacement sensor and a first pressure sensor, the first displacement sensor is configured to obtain the displacement of the steering cylinder of the loader, and the first pressure sensor is configured to obtain the pressure of the steering cylinder of the loader.

[0226] For another example, the second sensor includes a second displacement sensor and a second pressure sensor, the second displacement sensor is configured to obtain the displacement of the bucket cylinder of the loader, and the second pressure sensor is configured to obtain the pressure of the bucket cylinder of the loader.

[0227] For another example, the third sensor includes a third displacement sensor and a third pressure sensor, the third displacement sensor is configured to obtain the displacement of the boom cylinder of the loader, and the third pressure sensor is configured to obtain the pressure of the boom cylinder of the loader.

[0228] In the above embodiment, the loader also includes a first sensor, a second sensor, a third sensor and an inertial navigation system, so that the steering cylinder displacement and steering cylinder pressure of the loader, the bucket cylinder displacement and bucket cylinder pressure of the loader, the boom cylinder displacement and boom cylinder pressure of the loader and the heading angle of the loader can be directly measured, so that the machine learning model can be trained with the actually measured working data, and the machine learning model can predict the SOC and vehicle speed of the loader using the actually measured working data.

[0229] The embodiment of the present disclosure further provides a computer-readable storage medium, including computer program instructions, which implement the steps of the method of any one of the above embodiments when executed by a processor.

[0230] The embodiments of the present disclosure further provide a computer program product, including a computer program, which implements the steps of the method of any one of the above embodiments when executed by a processor.

[0231] So far, various embodiments of the present disclosure have been described in detail. In order to avoid obscuring the concept of the present disclosure, some details known in the art are not described. Based on the above description, those skilled in the art can fully understand how to implement the technical solution disclosed here.

[0232] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, systems, or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present disclosure may take the form of a computer program product implemented on one or more computer-usable non-transient storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0233] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present disclosure. It should be understood that the functions specified in one or more processes in the flowchart and / or one or more blocks in the block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions specified in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0234] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0235] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0236] Although some specific embodiments of the present disclosure have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present disclosure. It should be understood by those skilled in the art that the above embodiments may be modified or some technical features may be replaced by equivalents without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A parameter prediction method, comprising: Determine the current operating condition of the loader; Acquire working data of the loader in the current working condition at a current first moment; as well as The working data is input into a machine learning model corresponding to the current operating condition to predict the battery state of charge of the loader at a second moment and the vehicle speed at the second moment, the second moment being after the first moment, wherein the current operating condition is one of multiple operating conditions, and the machine learning model corresponding to one of at least two operating conditions is different from the machine learning model corresponding to another one.

2. The prediction method according to claim 1, wherein: The working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power.

3. The prediction method according to claim 1 or 2, wherein: Determining the current working condition of the loader includes: The current working condition of the loader is determined according to the pressure value of the working pump of the loader, wherein the working pump is used to drive at least one of the steering cylinder of the loader, the bucket cylinder of the loader and the boom cylinder of the loader.

4. The prediction method according to claim 3, wherein: Determining the current working condition of the loader according to the pressure value of the working pump of the loader includes: When the pressure value of the working pump is greater than the first preset pressure value, the current operating condition is determined to be one of a first group of operating conditions, wherein the first group of operating conditions includes one or more of a first operating condition, a second operating condition, a third operating condition, a fourth operating condition, and a fifth operating condition, wherein: Under the first working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of driving along a straight line, a shovel loading phase, a loaded backward phase of driving along a straight line, and a loaded forward phase of driving along a curve; In the second working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of traveling in a straight line, a shovel loading phase, a loaded backward phase of traveling in a straight line, and a loaded forward phase of traveling in a straight line; Under the third working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of driving along a curve, a shovel loading phase, a loaded backward phase of driving along a curve, and a loaded forward phase of driving along a curve; Under the fourth working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of driving along a straight line, a shovel loading phase, a loaded backward phase of driving along a curve, and a loaded forward phase of driving along a straight line; In the fifth working condition, the operation phases of the loader include, in chronological order: an empty forward phase, a shoveling operation phase, and an unloading phase.

5. The prediction method according to claim 4, wherein: When the pressure value of the working pump is greater than the first preset pressure value, determining that the current working condition is one of the first group of working conditions includes: When the first condition is met, the current operating condition is determined to be the third operating condition, wherein the first condition includes: the loader performs shoveling operations after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is not equal to the first preset displacement.

6. The prediction method according to claim 4, wherein: When the pressure value of the working pump is greater than the first preset pressure value, determining that the current working condition is one of the first group of working conditions includes: When the second condition is met, determining that the current operating condition is any operating condition in the first group of operating conditions except the third operating condition, wherein the second condition includes: The loader performs shoveling operation after moving forward without load, and during the process of the loader moving forward without load, the displacement of the steering cylinder of the loader is equal to the first preset displacement.

7. The prediction method according to claim 6, wherein: When the second condition is met, determining that the current operating condition is any operating condition in the first group of operating conditions except the third operating condition includes at least one of the following: In the case where the loader performs shovel loading and then unloading, determining that the current working condition is the fifth working condition; When the third condition is met, determining that the current operating condition is the fourth operating condition; When the fourth condition is met, determining that the current operating condition is the second operating condition; as well as When the fifth condition is met, determining that the current operating condition is the first operating condition; The third condition includes: the loader moves backward with the load after performing the shovel loading operation, and during the process of the loader moving backward with the load, the displacement of the steering cylinder is not equal to the first preset displacement; The fourth condition includes: the loader sequentially moves backward and forward with the load after performing the shovel loading operation, wherein during the process of the loader moving backward and forward with the load, the displacement of the steering cylinder is equal to the first preset displacement; The fifth condition includes: the loader sequentially moves backward and forward with the load after performing the shoveling operation, wherein, during the process of the loader moving backward with the load, the displacement of the steering cylinder is equal to the first preset displacement; and during the process of the loader moving forward with the load, the displacement of the steering cylinder is not equal to the first preset displacement.

8. The prediction method according to claim 3, wherein: Determining the current working condition of the loader according to the pressure value of the working pump of the loader includes: When the pressure value is equal to the first preset pressure value and the vehicle speed is greater than 0, the current operating condition is determined to be one of the second group of operating conditions, the second group of operating conditions including one or more of a straight-line driving condition, a non-horizontal driving condition and a turning driving condition, the straight-line driving condition including at least one of a first straight-line driving condition and a second straight-line driving condition, the non-horizontal driving condition including at least one of an uphill driving condition and a downhill driving condition, the vehicle speed of the first straight-line driving condition is greater than a first preset vehicle speed, and the vehicle speed of the second straight-line driving condition is less than or equal to the first preset vehicle speed.

9. The prediction method according to claim 8, wherein: When the pressure value is equal to the first preset pressure value and the vehicle speed is greater than 0, determining that the current operating condition is one of the second group of operating conditions includes at least one of the following: When the displacement of the steering cylinder of the loader is not equal to the first preset displacement, determining that the current working condition is the steering driving working condition; When the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is not 0, and the accelerator pedal opening is greater than the first preset opening, determining that the current working condition is the uphill driving working condition; When the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is not 0, and the accelerator pedal opening is equal to the first preset opening, determining that the current working condition is the downhill driving working condition; When the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is 0, and the vehicle speed is greater than the first preset speed, determining that the current working condition is the first straight-line driving working condition; as well as When the displacement of the steering cylinder is equal to the first preset displacement, the ramp angle is 0, and the vehicle speed is less than or equal to the first preset speed, the current operating condition is determined to be the second straight-line driving condition.

10. A model training method comprising: Acquiring working data of the first loader under a preset working condition at a third moment; Acquire the state of charge of the battery of the first loader under the preset working condition at a fourth moment and the vehicle speed at the fourth moment, wherein the fourth moment is after the third moment; as well as The working data at the third moment is taken as input, the state of charge at the fourth moment and the vehicle speed at the fourth moment are taken as output, and the machine learning model corresponding to the preset operating condition is trained, wherein the preset operating condition is one of multiple operating conditions, and the machine learning model corresponding to one of at least two operating conditions is different from the machine learning model corresponding to the other one.

11. The training method according to claim 10, further comprising: Obtaining working data of a second loader deployed with the machine learning model under a current working condition at a fifth moment, wherein the preset working condition includes the current working condition; Acquire the state of charge of the battery and the vehicle speed of the second loader under the preset working condition at a sixth moment, the sixth moment being after the fifth moment; as well as The working data at the fifth moment is taken as input, the state of charge at the sixth moment and the vehicle speed at the sixth moment are taken as output, and the machine learning model corresponding to the current working condition is trained.

12. The training method according to claim 11, wherein: The second loader is different from the first loader.

13. The training method according to any one of claims 10 to 12, wherein: The working data includes one or more of vehicle speed, gear position, accelerator pedal opening, brake pedal opening, and travel motor power.

14. The training method according to claim 13, wherein: The preset operating conditions include a straight-line driving condition, which includes a first straight-line driving condition and a second straight-line driving condition. The vehicle speed of the first straight-line driving condition is greater than a first preset vehicle speed, and the vehicle speed of the second straight-line driving condition is less than or equal to the first preset vehicle speed. The machine learning model corresponding to the first straight-line driving condition is different from the machine learning model corresponding to the second straight-line driving condition.

15. The training method according to claim 13, wherein: The working data also includes a ramp angle, the preset working condition includes a non-horizontal driving condition, the non-horizontal driving condition includes at least one of an uphill driving condition and a downhill driving condition, and the machine learning model corresponding to the uphill driving condition is different from the machine learning model corresponding to the downhill driving condition.

16. The training method according to claim 13, wherein: The working data also includes one or more of a heading angle, a steering cylinder displacement and a steering cylinder pressure, and the preset working condition includes a steering driving working condition.

17. The training method according to claim 13, wherein: The working data also includes one or more of bucket cylinder displacement, boom cylinder displacement, bucket cylinder pressure, boom cylinder pressure and working motor power, and the preset working condition includes at least one of the second working condition and the fifth working condition, wherein: In the second working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of traveling in a straight line, a shovel loading phase, a loaded backward phase of traveling in a straight line, and a loaded forward phase of traveling in a straight line; In the fifth working condition, the operation phases of the loader include, in chronological order: an empty forward phase, a shoveling operation phase, and an unloading phase.

18. The training method according to claim 13, wherein: The preset working condition includes at least one of the first working condition, the third working condition and the fourth working condition, and the working data also includes one or more of the heading angle, the displacement of the bucket cylinder, the displacement of the boom cylinder, the displacement of the steering cylinder, the pressure of the bucket cylinder, the pressure of the boom cylinder, the pressure of the steering cylinder, and the power of the working motor, wherein: Under the first working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of driving along a straight line, a shovel loading phase, a loaded backward phase of driving along a straight line, and a loaded forward phase of driving along a curve; Under the third working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of driving along a curve, a shovel loading phase, a loaded backward phase of driving along a curve, and a loaded forward phase of driving along a curve; In the fourth working condition, the operation phases of the loader include, in chronological order: an unloaded forward phase of traveling in a straight line, a shoveling operation phase, a loaded backward phase of traveling in a curve, and a loaded forward phase of traveling in a straight line.

19. A parameter prediction device, comprising: A module configured to execute the prediction method according to any one of claims 1 to 9.

20. A model training device, comprising: A module configured to execute the training method described in any one of claims 10-18.

21. An electronic device comprising: Memory; as well as A processor coupled to the memory, configured to execute the method of any one of claims 1-18 based on instructions stored in the memory.

22. A loader, comprising at least one of the following: The parameter prediction device of claim 19; A training device for the model of claim 20; and The electronic device of claim 21.

23. The loader according to claim 22, further comprising: A first sensor, mounted on the steering cylinder of the loader, configured to obtain a displacement and a pressure of the steering cylinder of the loader; A second sensor, mounted on the bucket cylinder of the loader, is configured to obtain a displacement and a pressure of the bucket cylinder of the loader; a third sensor, mounted on the boom cylinder of the loader, configured to obtain a displacement and a pressure of the boom cylinder of the loader; The inertial navigation system is configured to obtain the heading angle of the loader.

24. A computer-readable storage medium comprising a computer program, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 18 are implemented.

25. A computer program product comprising a computer program, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 18 are implemented.

Images