Power head gear prediction method and device and model training method and device

Abstract

The invention provides a power head gear prediction method and device and a model training method and device, and relates to the field of engineering machinery, and the prediction method comprises the steps: obtaining at least one group of data in a first group of data in a first time period and a second group of data in a second time period in a power head operation process, the first group of data comprises impact pressure data and impact flow data of the power head, and the second group of data comprises rotation pressure data and rotation speed data of the power head; and using a machine learning model to perform at least one of a first prediction to predict an impact gear of the power head based on the first set of data and a second prediction to predict a rotation gear of the power head based on the second set of data.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of engineering machinery, and in particular to a method and device for predicting a power head gear position, and a method and device for training a model. BACKGROUND

[0002] In the field of engineering machinery, a multifunctional drill is a common device for geological engineering, which can be applied to tunnel advanced geological drilling, tunnel drilling and blasting method construction drilling, pipe shed construction drilling, anchor rod drilling and other engineering. The power head is the core power component of the multifunctional drill, which is responsible for providing rotation, impact and propulsion power during the drilling construction process of the multifunctional drill. In order to adapt to the efficient drilling of rocks with different physical properties, the operator usually manually adjusts the mechanical structure on the power head to realize the adjustment of the impact gear position and the rotation gear position. The traditional power head of the multifunctional drill lacks the function of identifying the gear position of the power head, and the operator needs to manually record the gear position information and input it to the vehicle-mounted computer, which has low digitalization degree, affects the work efficiency, and the driving position is far away from the power head on the mast, which is inconvenient to operate.

[0003] In related technologies, by setting an empty gear sensor, a direct current converter, an analog-to-digital converter and a programmable logic controller (PLC), the pulse width modulation (PWM) signal output by the empty gear sensor is converted into a voltage signal by the direct current converter, and then converted into a current signal by the analog-to-digital converter, and finally the gear position of the power head is obtained according to the current signal based on the pre-acquired current signal-duty cycle relationship and the duty cycle-gear position relationship through the PLC.

[0004] In related technologies, it is also proposed to set an electromagnetic switch, and when the manual gearbox gear position switch moves, the electromagnetic switch generates a level signal change, and based on the corresponding relationship between the level signal change and the gear position, the current gear position of the gearbox is identified. SUMMARY

[0005] According to a first aspect of an embodiment of the present disclosure, a method for predicting a power head gear position is provided, comprising: obtaining at least one of a first group of data in a first time period and a second group of data in a second time period during a power head operation process, the first group of data comprising impact pressure data and impact flow data of the power head, and the second group of data comprising rotation pressure data and rotation speed data of the power head; and using a machine learning model to perform at least one of a first prediction and a second prediction, the first prediction being to predict an impact gear position of the power head based on the first group of data, and the second prediction being to predict a rotation gear position of the power head based on the second group of data.

[0006] In some embodiments, the first time period and the second time period are the same time period.

[0007] In some embodiments, the first set of data further comprises impact frequency data of the hammerhead.

[0008] In some embodiments, the impact frequency data is determined according to the impact pressure data.

[0009] In some embodiments, the machine learning model comprises a plurality of models, the first prediction comprises predicting a plurality of impact gears based on the plurality of models and the first set of data, and taking an impact gear with the most repetitions in the plurality of impact gears as the impact gear of the hammerhead, and the second prediction comprises predicting a plurality of rotation gears based on the plurality of models and the second set of data, and taking a rotation gear with the most repetitions in the plurality of rotation gears as the rotation gear of the hammerhead.

[0010] In some embodiments, each model is a simple recurrent unit model.

[0011] According to a second aspect of embodiments of the present disclosure, a method for training a model is provided, comprising: obtaining training data, the training data comprising at least one of first training data in a third time period during a hammerhead operation process and second training data in a fourth time period during the hammerhead operation process, the first training data comprising a first set of data and a corresponding impact gear of the hammerhead, the first set of data comprising impact pressure data and impact flow data of the hammerhead, the second training data comprising a second set of data and a corresponding rotation gear of the hammerhead, the second set of data comprising rotation pressure data and rotation speed data of the hammerhead; and performing at least one of first training and second training on a machine learning model using the training data, wherein: the first training takes the first set of data as input and the impact gear corresponding to the first set of data as output, and the second training takes the second set of data as input and the rotation gear corresponding to the second set of data as output.

[0012] In some embodiments, the first set of data further comprises impact frequency data of the hammerhead.

[0013] In some embodiments, the impact frequency data is determined according to the impact pressure data.

[0014] In some embodiments, the machine learning model comprises a plurality of models, the first training is performed for each model, and the second training is performed for each model.

[0015] In some embodiments, each model is a simple recurrent unit model.

[0016] According to a third aspect of the embodiments of the present disclosure, a power head gear position prediction device is provided, comprising: a device configured to perform the prediction method of any one of the above-mentioned embodiments.

[0017] According to a fourth aspect of the embodiments of the present disclosure, a model training device is provided, comprising: a device configured to perform the training method of any one of the above-mentioned embodiments.

[0018] According to a fifth aspect of the embodiments of the present disclosure, an electronic device is provided, comprising: a memory; and a processor coupled to the memory, the processor being configured to perform the method of any one of the above-mentioned embodiments based on instructions stored in the memory.

[0019] According to a sixth aspect of the embodiments of the present disclosure, a power head gear position prediction system is provided, comprising: the prediction device of any one of the above-mentioned embodiments; and at least one of a first group of sensors and a second group of sensors; wherein the first group of sensors comprises an impact pressure sensor configured to obtain the impact pressure data, and an impact flow sensor configured to obtain the impact flow data, and the second group of sensors comprises a swing pressure sensor configured to obtain the swing pressure data, and a Hall sensor configured to obtain the swing speed data.

[0020] According to a seventh aspect of the embodiments of the present disclosure, an engineering machine is provided, comprising: the prediction system of any one of the above-mentioned embodiments.

[0021] According to an eighth aspect of the embodiments of the present disclosure, a computer readable storage medium is provided, comprising computer program instructions, wherein the computer program instructions are executed by a processor to implement the method of any one of the above-mentioned embodiments.

[0022] According to a ninth aspect of the embodiments of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program is executed by a processor to implement the method of any one of the above-mentioned embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0024] Figure 1 is a flowchart of a power head gear position prediction method according to some embodiments of the present disclosure.

[0025] Figure 2An implementation of acquiring impact frequency data from impact pressure data is shown according to some embodiments of the present disclosure.

[0026] Figure 3 is a flow chart of a training method of a model according to some embodiments of the present disclosure.

[0027] Figure 4 is a structural schematic diagram of a power head gear prediction device according to some embodiments of the present disclosure.

[0028] Figure 5 is a structural schematic diagram of a training device of a model according to some embodiments of the present disclosure.

[0029] Figure 6 is a structural schematic diagram of an electronic device according to some embodiments of the present disclosure.

[0030] Figure 7 is a structural schematic diagram of a prediction system according to some embodiments of the present disclosure.

[0031] Figure 8 is a structural schematic diagram of a prediction system according to some other embodiments of the present disclosure. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, but not all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present disclosure.

[0033] Unless otherwise specifically stated, the relative arrangement of parts and steps, numerical expressions, and numerical values set forth in the various examples disclosed herein are not limiting but merely exemplary.

[0034] At the same time, it should be understood that, for the convenience of description, the size of each part shown in the drawings is not drawn according to the actual proportional relationship.

[0035] The techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail, but should be considered part of the specification where appropriate.

[0036] In all the examples shown and discussed herein, any specific value should be interpreted as merely exemplary, and not as a limitation. Thus, other examples of the example embodiments can have different values.

[0037] It should be noted that like numerals and letters refer to like items throughout the several views, and once an item is defined in one view, it should not have to be discussed further in subsequent views.

[0038] The inventor finds that in the related art, by means of setting a neutral sensor, a direct current converter, an analog-to-digital converter and a PLC, on the one hand, the cost of the whole vehicle is significantly increased, and on the other hand, the neutral sensor needs to be arranged on the power head, and the multifunctional drilling machine usually works in a harsh environment, and the power head is close to the working surface, so that the sensor is easily damaged.

[0039] In addition, in the related art, if an electromagnetic switch is used, the electromagnetic switch needs to be arranged on the handle for adjusting the gear of the power head. However, when the power head is working normally, a large impact vibration is generated, so that the electromagnetic switch is prone to induction error.

[0040] Therefore, the present disclosure proposes the following technical solutions, which can stably and accurately obtain the gear of the power head in a low-cost manner.

[0041] Figure 1 is a flowchart of a prediction method of a gear of a power head according to some embodiments of the present disclosure. In some embodiments, the power head is a power head of a multifunctional drilling machine.

[0042] As shown in Figure 1 , the prediction method comprises steps S102 and S104.

[0043] In step 102, at least one of a first group of data in a first time period and a second group of data in a second time period during the operation of the power head is obtained.

[0044] Here, the first group of data comprises impact pressure data and impact flow data of the power head, and the second group of data comprises rotation pressure data and rotation speed data of the power head.

[0045] The first time period can be a period of time from the current time, at which time the first group of data can be regarded as the impact pressure data and the impact flow data of the power head in a period of time from the current time, which can reflect the current impact gear of the power head.

[0046] The first time period can also be a period of time before the current time, at which time the first group of data can reflect the impact gear of the power head in a period of time before the current time.

[0047] Similarly, the second time period can be a period of time from the current time, at which time the second group of data can be regarded as the rotation pressure data and the rotation speed data of the power head in a period of time from the current time, which can reflect the current rotation gear of the power head.

[0048] The second time period can also be a period of time before the current time, and the second set of data can reflect the swing gear of the power head in the period of time before the current time.

[0049] The first time period and the second time period can be the same time period or different time periods. In the case of different time periods, the length of the first time period and the length of the second time period can be the same or different.

[0050] For example, the first set of data and / or the second set of data are obtained by sensors that are not arranged on the power head and are self-provided by the engineering machinery (e.g., a multifunctional drilling machine). For example, the sensors collect the first set of data and / or the second set of data and transmit them to the controller, and the controller sends the received first set of data and / or the second set of data (e.g., via a CAN bus protocol) to the on-board computer, so that the on-board computer obtains the first set of data and / or the second set of data.

[0051] In some embodiments, the impact pressure data in the first set of data can be obtained by an impact pressure sensor; the impact flow data in the first set of data can be obtained by an impact flow sensor; the swing pressure data in the second set of data can be obtained by a swing pressure sensor; and the swing speed data in the second set of data can be obtained by a Hall sensor.

[0052] In step S104, at least one of the first prediction and the second prediction is performed using the machine learning model. Here, the first prediction is to predict the impact gear of the power head based on the first set of data, and the second prediction is to predict the swing gear of the power head based on the second set of data.

[0053] In the first prediction, the machine learning model can output the impact gear of the power head with the first set of data as input. In the second prediction, the machine learning model can output the swing gear of the power head with the second set of data as input.

[0054] It can be understood that, in the case of the same time period for the first time period and the second time period, the impact gear and the swing gear of the power head in the same time period can be simultaneously predicted using the machine learning model. In the case of different time periods for the first time period and the second time period, the impact gear of the power head in the first time period and the swing gear of the power head in the second time period can be respectively predicted using the machine learning model.

[0055] The above power head gear position prediction method can predict at least one of the impact gear position and the rotation gear position of the power head at low cost without adding other components to the power head, by the relationship between the impact pressure data and the impact flow data of the power head and the impact gear position of the power head, and / or the relationship between the rotation pressure data and the rotation speed data of the power head and the rotation gear position of the power head. In addition, the data required for prediction is derived from real data during the operation of the power head, and the accuracy of gear position prediction is not affected by the impact vibration caused by the power head, and the accuracy is higher.

[0056] In some embodiments, at least one of normalization processing and standardization processing can be performed on the first set of data before inputting the first set of data into the machine learning model during the execution of the first prediction. After the output of the machine learning model, the output data can be decoded (for example, ONE-HOT decoding and classification decoding are performed in sequence) to convert into a text label corresponding to the impact gear position of the power head, and then the impact gear position of the power head is obtained.

[0057] In some embodiments, at least one of normalization processing and standardization processing can be performed on the second set of data before inputting the second set of data into the machine learning model during the execution of the second prediction. After the output of the machine learning model, the output data can be decoded (for example, ONE-HOT decoding and classification decoding are performed in sequence) to convert into a text label corresponding to the rotation gear position of the power head, and then the rotation gear position of the power head is obtained.

[0058] At least one of normalization processing and standardization processing can be performed on the input data by a preprocessing module. Decoding the output data can be performed by a post-processing module. That is, the preprocessing module and the post-processing module can be used in cooperation with the machine learning model to predict the gear position of the power head.

[0059] Normalization processing of input data to scale to a smaller range can speed up the convergence speed of the gradient descent-based algorithm in the machine learning model training process. Standardization processing of input data can reduce the influence of outliers on variance in input data, making the machine learning model training process more fair between different training data.

[0060] During the machine learning model training process, the text label corresponding to the gear position (impact gear position or rotation gear position) of the power head can be encoded (for example, classification encoding and ONE-HOT encoding are performed in sequence) as output, which can convert discrete classification labels into binary patterns that are easy for machine learning algorithms to process, thereby improving the processing capability of machine learning algorithms for discrete features.

[0061] For example, the text label corresponding to the gear (impact gear or rotation gear) is "Gear x", x is a positive integer, which becomes x-1 after classification coding, and becomes (0, …, 0, 1, 0, …, 0) after ONE-HOT coding, where the xth bit is 1 and the rest are 0.

[0062] Correspondingly, when using the machine learning mode to make a prediction, the output data of the machine learning model needs to be sequentially ONE-HOT decoded and classification decoded to obtain the text label corresponding to the gear (impact gear or rotation gear) of the power head.

[0063] In some embodiments, both the first set of data and the second set of data are obtained in step S102, and the first time period and the second time period are the same time period. In this way, the first prediction and the second prediction can be performed simultaneously to obtain the impact gear and the rotation gear of the power head in the same time period.

[0064] As some implementations, the first time period and the second time period are the same period of time starting from the current time. In this way, the current impact gear and rotation gear of the power head can be predicted in real time.

[0065] In some embodiments, in addition to the impact pressure data of the power head and the impact flow data of the power head, the first set of data also includes impact frequency data of the power head. In this way, by the relationship between the impact pressure data, the impact flow data and the impact frequency data of the power head and the impact gear of the power head, the error of predicting the impact gear of the power head based on the impact pressure data and the impact flow data can be reduced, so that the impact gear of the power head can be more accurately predicted.

[0066] As some implementations, the impact frequency data of the power head is determined according to the impact pressure data of the power head. For example, the impact frequency data of the power head can be determined according to the impact pressure data of the power head by a frequency extraction program.

[0067] In this way, on the one hand, no additional sensors are needed to obtain the impact frequency data; on the other hand, the impact frequency data obtained directly from the impact pressure data is also more accurate, which is more conducive to accurately predicting the impact gear of the power head.

[0068] The following describes an implementation of obtaining the impact frequency data from the impact pressure data according to some embodiments of the present disclosure.

[0069] The impact pressure data is the data of the impact pressure changing with time in the first time period. As shown in (a) of Figure 2 The impact pressure data is sampled by a window with a certain number of sample points. As shown in (b) of Figure 2As shown in (b) of FIG. 1, a Fast Fourier Transform (FFT) is performed on the sample points in the window, and after the FFT transform, the shock pressure data is converted from time domain data to frequency domain data. As shown in (c) of FIG. 1, after removing the direct current component (the part with a frequency of 0 in the frequency domain data), the frequency corresponding to the maximum amplitude value in the frequency domain data is the shock frequency of the window, and the frequency value is recorded. Then, the sample window is moved by one sample point, and the shock frequency of the next window is calculated in the same way until the window reaches the end of the time domain data of the shock pressure. The frequency values corresponding to all the windows form the shock frequency data. Figure 2

[0070] In the above implementation, the sample window is moved by one sample point each time, and at the two positions adjacent before and after each movement, the sample window has more sample points in common, and the frequency values obtained in this way can be more accurate.

[0071] In some embodiments, the machine learning model used in step S104 includes multiple models. In this case, the first prediction and the second prediction can be realized in the following way.

[0072] As some implementations, the first prediction includes predicting multiple shock gears based on the multiple models and the first set of data, and taking the shock gear with the most repetitions in the multiple shock gears as the shock gear of the power head. Here, each model can output one shock gear with the first set of data as input.

[0073] As some implementations, the second prediction includes predicting multiple rotation gears based on the multiple models and the second set of data, and taking the rotation gear with the most repetitions in the multiple rotation gears as the rotation gear of the power head. Here, each model can output one rotation gear with the second set of data as input.

[0074] It should be understood that in the case of taking the first set of data and the second set of data as input at the same time, each model can output one shock gear and one rotation gear at the same time.

[0075] For example, the machine learning model can also include a combination module. The combination module votes for the gears predicted by each model to determine the gear with the most repetitions. At this time, the machine learning model can be an ensemble learning model.

[0076] In the above embodiments, the inaccuracy of a single model can be avoided, which leads to the inaccuracy of the shock gear or rotation gear of the power head predicted by the machine learning model, and the accuracy of the prediction of the machine learning model is further improved.

[0077] ​In some embodiments, each model included in the machine learning model is a Simple Recurrent Unit (SRU) model. By selecting the SRU model, the impact gear and / or the rotation gear of the power head can be more accurately predicted in the case of using the impact pressure data and the impact flow data, and / or the rotation pressure data and the rotation speed data.

[0078] In some embodiments, after the gear of the power head is predicted by the machine learning model, the gear can be output to a display screen for reference by an operator. In some embodiments, the predicted gear of the power head can also be called in a surrounding rock identification program interface to identify surrounding rock information (for example, strength information of the surrounding rock) according to the gear. For example, the strength information of the surrounding rock is identified according to the impact gear and the rotation gear (for example, other information can also be combined).

[0079] The present disclosure also proposes a model training method for training a machine learning model so as to use the machine learning model to predict the gear of the power head.

[0080] Figure 3 is a flowchart of the model training method according to some embodiments of the present disclosure.

[0081] As shown in Figure 3 , the training method includes steps S302 and S304.

[0082] In step S302, training data is obtained.

[0083] Here, the training data includes at least one of first training data in a third time period during the operation of the power head and second training data in a fourth time period during the operation of the power head, the first training data includes a first set of data and a corresponding impact gear of the power head, the first set of data includes impact pressure data and impact flow data of the power head, and the second training data includes a second set of data and a corresponding rotation gear of the power head, the second set of data includes rotation pressure data and rotation speed data of the power head.

[0084] It should be understood that the first training data and the second training data in the training data can each include multiple sets of data. For the first training data, each set of data includes the first set of data and the corresponding impact gear of the power head. For the second training data, each set of data includes the second set of data and the corresponding rotation gear of the power head.

[0085] In the first training data, different sets of data correspond to different third time periods, i.e. are obtained in different third time periods during the operation of the power head. In the second training data, different sets of data correspond to different fourth time periods, i.e. are obtained in different fourth time periods during the operation of the power head.

[0086] As described above, the impact pressure data, the impact flow rate data, the rotation pressure data, and the rotation speed data can be acquired by the impact pressure sensor, the impact flow rate sensor, the rotation pressure sensor, and the Hall sensor, respectively.

[0087] When the impact pressure data and the impact flow rate data are acquired, the operator can manually record the impact gear of the power head at this time. For example, the impact pressure data, the impact flow rate data, and the corresponding impact gear are stored in the memory as a set of data in the first training data for use in training.

[0088] When the rotation pressure data and the rotation speed data are acquired, the operator can manually record the rotation gear of the power head at this time. For example, the rotation pressure data, the rotation speed data, and the corresponding rotation gear are stored in the memory as a set of data in the second training data for use in training.

[0089] In step S304, at least one of the first training and the second training is performed on the machine learning model using the training data.

[0090] Here, the first training takes the first set of data as input and the impact gear corresponding to the first set of data as output. The second training takes the second set of data as input and the rotation gear corresponding to the second set of data as output.

[0091] The first training and the second training are respectively used to train the machine learning model to predict the impact gear of the power head and to predict the rotation gear of the power head.

[0092] In the training method of the above embodiment, by establishing the relationship between the impact pressure data and the impact flow rate data of the power head and the impact gear of the power head, and / or the relationship between the rotation pressure data and the rotation speed data of the power head and the rotation gear of the power head, the trained machine learning model can perform at least one of the first prediction and the second prediction described above.

[0093] In some embodiments, the training data acquired in step S302 includes both the first training data and the second training data, and the first training and the second training are simultaneously performed in step S304. In this way, when the input data includes the impact pressure data, the impact flow rate data, the rotation pressure data, and the rotation speed data at the same time, the machine learning model can simultaneously predict the impact gear and the rotation gear of the power head, which helps to understand the detailed situation of the multi-functional drilling rig during operation.

[0094] In some embodiments, the third time period and the fourth time period are the same, i.e., the first training data and the second training data can be acquired in the same time period during the power head operation. In this way, the first training data and the second training data can be simultaneously used as input, and the impact gear and the swing gear of the power head can be simultaneously used as output to train the machine learning model.

[0095] In some embodiments, the first training data in the training data acquired in step S302 further includes impact frequency data of the power head. In this way, when the trained machine learning model is used, the impact gear of the power head can be predicted according to the impact pressure data, the impact flow data and the impact frequency data, and the prediction accuracy can be improved.

[0096] As some implementation manners, the impact frequency data of the power head is determined according to the impact pressure data of the power head. In this way, the impact frequency data can be acquired without additional installation of sensors, and the impact frequency data acquired directly according to the impact pressure data is more accurate, so that the training effect of the machine learning model is better.

[0097] The implementation manner of acquiring the impact frequency data according to the impact pressure data can refer to the above, which will not be described here.

[0098] In some embodiments, the machine learning model trained by the above training method includes a plurality of models.

[0099] In this case, when the first training is performed, the first training is performed for each model, i.e., the first group of data is used as input, the impact gear corresponding to the first group of data is used as output, and each model is trained respectively; when the second training is performed, the second training is performed for each model, i.e., the second group of data is used as input, the swing gear corresponding to the second group of data is used as output, and each model is trained respectively; when the first training and the second training are performed, the first training and the second training are simultaneously performed for each model, i.e., the first group of data and the second group of data are simultaneously used as input, the impact gear corresponding to the first group of data and the swing gear corresponding to the second group of data are simultaneously used as output, and each model is trained respectively.

[0100] In some embodiments, the first training data used for training each model is a part of the training data randomly selected. The first training data used by different models can be the same or different.

[0101] In the above embodiments, when the machine learning model is used, the gear with the most repetitions in the multiple gears (impact gears, or rotation gears, or a combination of impact gears and rotation gears) predicted by the multiple models is taken as the gear of the power head predicted by the machine learning model. In this way, the inaccuracy of a single model can be avoided, and the inaccuracy of the impact gear or rotation gear of the power head predicted by the machine learning model can be avoided, and the prediction accuracy of the machine learning model can be further improved.

[0102] In some embodiments, each model included in the machine learning model is an SRU model, so that the machine learning model trained in this way has better prediction effect.

[0103] In some embodiments, during the training process, at least one of normalization processing and standardization processing can be performed on the input data (at least one of the first group of data and the second group of data). In some embodiments, the gear of the power head corresponding to the input data (i.e., the text label corresponding to the gear) is sequentially classified and ONE-HOT encoded as the output data of the model. In this way, the performance of the machine learning model can be improved.

[0104] In some embodiments, after training the machine learning model, the prediction accuracy of the machine learning model can be determined according to the training data. For example, for a certain group of data in the training data, it is determined whether the gear predicted by the machine learning model according to the input data (e.g., the first group of data, or the second group of data, or both the first group of data and the second group of data) of the group of data is consistent with the gear included in the group of data, and then the proportion of the number of groups of data with accurate prediction results in the total number of groups of data is determined, i.e., the prediction accuracy.

[0105] When the prediction accuracy is less than a preset accuracy threshold, the parameters of the machine learning model can be adjusted, such as adjusting the parameters (e.g., learning rate, etc.) of each model in the machine learning model, retraining the machine learning model, until the prediction accuracy of the machine learning model is greater than or equal to the accuracy threshold. In this way, the machine learning model trained in this way can have a high prediction accuracy.

[0106] In some embodiments, the historical data can be filtered to obtain normal data as training data. For example, by setting a preset value, normal data greater than the preset value can be filtered out. In this way, data fluctuating around zero caused by noise interference when the multifunctional drilling rig is in standby mode can be excluded, and the training effect of the machine learning model can be improved.

[0107] The various embodiments are described in the specification by progressive stages, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be mutually referred to. For the device embodiments, since they are basically corresponding to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0108] The disclosure also proposes a power head gear prediction device, including a module configured to perform the prediction method of any one of the above embodiments.

[0109] Figure 4 FIG. 1 is a structural schematic diagram of a power head gear prediction device according to some embodiments of the disclosure.

[0110] As shown in Figure 4 , the power head gear prediction device includes a first acquisition module 401 and a prediction module 402.

[0111] The first acquisition module 401 is configured to acquire at least one of a first group of data in a first time period and a second group of data in a second time period during the power head operation. Here, the first group of data includes impact pressure data and impact flow data of the power head, and the second group of data includes rotation pressure data and rotation speed data of the power head.

[0112] The prediction module 402 is configured to use a machine learning model to perform at least one of a first prediction and a second prediction. Here, the first prediction is to predict the impact gear of the power head based on the first group of data, and the second prediction is to predict the rotation gear of the power head based on the second group of data.

[0113] The power head gear prediction device can also include other modules to perform the prediction method of any one of the above embodiments.

[0114] The disclosure also proposes a model training device, including a module configured to perform the training method of any one of the above embodiments.

[0115] Figure 5 FIG. 2 is a structural schematic diagram of a model training device according to some embodiments of the disclosure.

[0116] As shown in Figure 5 , the power head gear prediction device includes a second acquisition module 501 and a training module 502.

[0117] The second acquisition module 501 is configured to acquire training data. Here, the training data includes at least one of first training data within a third time period during the power head operation process and second training data within a fourth time period during the power head operation process, the first training data including a first set of data and a corresponding impact gear position of the power head, the first set of data including impact pressure data and impact flow data of the power head, and the second training data including a second set of data and a corresponding rotation gear position of the power head, the second set of data including rotation pressure data and rotation speed data of the power head.

[0118] The training module 502 is configured to perform at least one of a first training and a second training on the machine learning model using the training data. The first training uses the first set of data as input and outputs the impact gear position corresponding to the first set of data. The second training uses the second set of data as input and outputs the swing gear position corresponding to the second set of data.

[0119] The model training device may further include other modules to execute the training method of any of the above embodiments.

[0120] In some embodiments, the prediction device for the power head gear position may include a model training device.

[0121] The present disclosure also provides an electronic device.

[0122] The electronic device can be a device for predicting the gear position of the power head or a device for training the model.

[0123] Figure 6 is a schematic structural diagram of an electronic device according to some embodiments of the present disclosure.

[0124] like Figure 6 As shown, the electronic device 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 one of the aforementioned embodiments based on instructions stored in the memory 601 .

[0125] 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, application programs, a boot loader, and other programs.

[0126] In some embodiments, the electronic device can further include an input / output interface 603, a network interface 604, a storage interface 605, and the like. These interfaces 603, 604, 605, and the memory 601 and the processor 602 can be connected through a bus 606, for example. The input / output interface 603 provides a connection interface for display, mouse, keyboard, touch screen, and the like input / output devices. The network interface 604 provides a connection interface for various networking devices. The storage interface 605 provides a connection interface for an external storage device such as an SD card, a U disk, and the like.

[0127] The present disclosure also proposes a power head gear prediction system.

[0128] Figure 7 is a structural schematic diagram of a prediction system according to some embodiments of the present disclosure.

[0129] As shown in Figure 7 , the prediction system includes a power head gear prediction device 701, and at least one of a first group of sensors 702 and a second group of sensors 703, for example both the first group of sensors 702 and the second group of sensors 703.

[0130] The power head gear prediction device 701 can be the prediction device of any one of the above-mentioned embodiments.

[0131] The first group of sensors 702 includes an impact pressure sensor 704 and an impact flow sensor 705. The impact pressure sensor 704 is configured to obtain impact pressure data of the power head. The impact flow sensor 705 is configured to obtain impact flow data of the power head.

[0132] The second group of sensors 703 includes a rotation pressure sensor 706 and a Hall sensor 707. The rotation pressure sensor 706 is configured to obtain rotation pressure data of the power head. The Hall sensor 707 is configured to obtain rotation speed data of the power head.

[0133] Figure 8 is a structural schematic diagram of a prediction system according to other embodiments of the present disclosure. In Figure 8 , the power head gear prediction device 701 is schematically implemented as an on-board computer 701.

[0134] As shown in Figure 8 , the prediction system can further include a controller 708 and a memory 709. Figure 8 The impact gear 710, the rotation gear 711, the display 712, and the surrounding rock identification program interface 713 are also shown.

[0135] The controller 708 is configured to receive impact pressure data and impact flow data from the first group of sensors 702, and to receive swing pressure data and swing speed data from the second group of sensors 703, and to send the received data to the on-board computer 701 (e.g., through a CAN bus protocol).

[0136] The memory 709 can be a memory 709 internal to the on-board computer 701 or a memory 709 external to the on-board computer 701. The memory 709 is configured to store a model file of the machine learning model, so that the on-board computer 701 loads the machine learning model to perform at least one of the first prediction and the second prediction.

[0137] The impact gear 710 is configured to adjust the internal piston stroke according to the operation of the operator, to adjust the impact frequency and the impact force.

[0138] The swing gear 711 is configured to adjust the gear ratio according to the operation of the operator, to adjust the swing speed and the torque.

[0139] The display 712 is configured to receive and display the gear of the power head predicted by the on-board computer 701.

[0140] The surrounding rock identification program interface 713 is configured to receive the gear of the power head predicted by the on-board computer 701, and in response to being invoked, to identify the surrounding rock information according to the gear of the power head predicted by the on-board computer 701.

[0141] The following will be described in combination with Figure 8 , a method for predicting the gear of the power head according to some embodiments of the present disclosure.

[0142] First, data is collected from the first group of sensors 702 and / or the second group of sensors 703, and the current gear information of the power head is manually recorded; based on the impact pressure data, the impact frequency data is calculated through a frequency extraction program (not shown); normal data is screened out as training data through a preprocessing module (not shown), and the input data is normalized and standardized, the output data is classified and coded, and the classified and coded data is ONE-HOT coded; the input and output data are imported into the integrated learning model training in the vehicle-mounted computer 701, and the predicted gear information is sequentially ONE-HOT decoded and classified decoded through a post-processing module (not shown); the accuracy is calculated, and whether to adjust the parameters to retrain the integrated learning model is determined through the accuracy threshold, when the accuracy is greater than or equal to the threshold, the integrated learning model is output; the frequency extraction program, the preprocessing module, the integrated learning model, and the post-processing module are stored in the storage 709, and a running environment is built in the vehicle-mounted computer 701; subsequently, the first group of sensors 702 and / or the second group of sensors 703 collect data in real time, and the collected data is transmitted to the vehicle-mounted computer 701 through the controller 708; the vehicle-mounted computer 701 automatically outputs the gear information through the frequency extraction program, the preprocessing module, the integrated learning model, and the post-processing module; the gear information is displayed in the display 712, and is called in the surrounding rock identification program interface 713.

[0143] The present disclosure also proposes an engineering machine comprising the prediction system of any one of the above embodiments. The engineering machine may, for example, be a multifunctional drilling rig.

[0144] The embodiments of the present disclosure also provide a computer-readable storage medium comprising computer program instructions, which, when executed by a processor, implement the method of any one of the above embodiments.

[0145] The embodiments of the present disclosure also provide a computer program product comprising a computer program, which, when executed by a processor, implements the method of any one of the above embodiments.

[0146] So far, the 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. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein according to the above description.

[0147] Those skilled in the art will appreciate that embodiments of the disclosure can be supplied as a method, a system, or a computer program product. Accordingly, the disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the disclosure can take the form of a computer program product on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, and the like) having computer-usable program code embodied in the medium.

[0148] The disclosure is described in reference to the flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that the functions identified in one or more flows and / or one or more blocks of the flow diagrams can be implemented as 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, such that the instructions, which execute via the processor of the computer or other programmable data processing device, create means for implementing the functions specified in the flow diagrams and / or specified in the block or blocks of the flow diagrams. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure.

[0149] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions means which implement the function specified in the flow diagrams and / or specified in the block or blocks of the flow diagrams. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure.

[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer implemented process such that the instructions which execute on the computer or other programmable device provide steps for implementing the functions specified in the flow diagrams and / or specified in the block or blocks of the flow diagrams. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure. Figure 1 The flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the disclosure.

[0151] Although some specific embodiments of the disclosure have been described in detail, those skilled in the art should understand that the above examples are only for illustration, and are not intended to limit the scope of the disclosure. Those skilled in the art should understand that modifications can be made to the above embodiments or equivalent substitutions can be made to some technical features without departing from the scope and spirit of the disclosure. The scope of the disclosure is defined by the appended claims.

Claims

1. A method for predicting a power head gear position, comprising: Acquire at least one of a first set of data within a first time period and a second set of data within a second time period during operation of the power head, wherein the first set of data includes impact pressure data and impact flow data of the power head, and the second set of data includes rotation pressure data and rotation speed data of the power head; and Using a machine learning model, at least one of a first prediction and a second prediction is performed, wherein the first prediction is to predict the impact gear of the power head based on the first set of data, and the second prediction is to predict the rotation gear of the power head based on the second set of data.

2. The method according to claim 1, wherein The first time period and the second time period are the same time period.

3. The method according to claim 1, wherein The first set of data also includes impact frequency data of the power head.

4. The method according to claim 3, wherein: The impact frequency data is determined based on the impact pressure data.

5. The method according to any one of claims 1 to 4, wherein: The machine learning model includes multiple models; The first prediction includes: predicting a plurality of impact gears based on the plurality of models and the first set of data; and The impact gear with the largest number of repetitions among the multiple impact gears is used as the impact gear of the power head; The second prediction includes: predicting a plurality of swing gears based on the plurality of models and the second set of data; and The rotary gear position with the greatest number of repetitions among the plurality of rotary gear positions is used as the rotary gear position of the power head.

6. The method according to claim 5, wherein: Each model is a simple recurrent unit model.

7. A model training method comprising: Acquiring training data, the training data including at least one of first training data within a third time period during the operation of the power head and second training data within a fourth time period during the operation of the power head, the first training data including a first set of data and a corresponding impact gear position of the power head, the first set of data including impact pressure data and impact flow data of the power head, the second training data including a second set of data and a corresponding rotation gear position of the power head, the second set of data including rotation pressure data and rotation speed data of the power head; Perform at least one of a first training and a second training on the machine learning model using the training data, wherein: The first training takes the first set of data as input and the impact gear corresponding to the first set of data as output. The second training takes the second set of data as input and takes the rotation gear position corresponding to the second set of data as output.

8. The method according to claim 7, wherein: The first set of data also includes impact frequency data of the power head.

9. The method according to claim 8, wherein The impact frequency data is determined based on the impact pressure data.

10. The method according to any one of claims 7 to 9, wherein: The machine learning model includes multiple models, The first training is performed for each model, and the second training is performed for each model.

11. The method according to claim 10, wherein: Each model is a simple recurrent unit model.

12. A device for predicting the gear position of a power head, comprising: A module configured to perform the method according to any one of claims 1 to 6.

13. A model training device comprising: A module configured to perform the method according to any one of claims 7 to 11.

14. An electronic device comprising: Memory; as well as A processor coupled to the memory is configured to execute the method according to any one of claims 1 to 11 based on instructions stored in the memory.

15. A power head gear prediction system, comprising: The prediction device according to claim 12; as well as at least one of a first set of sensors and a second set of sensors; Among them, the first group of sensors includes an impact pressure sensor configured to obtain the impact pressure data and an impact flow sensor configured to obtain the impact flow data, and the second group of sensors includes a rotational pressure sensor configured to obtain the rotational pressure data and a Hall sensor configured to obtain the rotational speed data.

16. An engineering machine comprising: The prediction system according to claim 15.

17. A computer-readable storage medium comprising computer program instructions, wherein: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 11 is implemented.

18. A computer program product comprising a computer program, wherein When the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images