Method for controlling working state of excavator

Abstract

The embodiment of the invention provides a control method for the working state of an excavator. The invention relates to the field of excavator control, and the method comprises the steps that a sensor array is arranged on a hydraulic cylinder of an excavator, a switch valve group is arranged at an oil inlet and an oil outlet of the hydraulic cylinder, and vibration signals and strain data collected by the sensor array are obtained; extracting a vibration feature vector of the vibration signal, and extracting a strain feature vector of the strain data; inputting the vibration feature vector and the strain feature vector into an excavation medium identification model to obtain a current excavation medium type of the excavator; based on the control type, determining a control operation corresponding to the current mining medium type; and a control instruction corresponding to the control operation is sent to the switch valve group, so that the switch valve group executes the control operation. The method is used for accurately recognizing the type of the excavation medium when the excavator works, and the technical effect of effectively controlling the working state of the excavator is achieved.

Description

Technical Field

[0001] This application relates to the field of excavator control, and more particularly to a method for controlling the operating state of an excavator. Background Technology

[0002] During operation, in unstructured and dynamically changing complex working conditions (such as mines, construction sites, farmland, etc.), traditional hydraulic cylinders, as actuators, need to complete precise tasks such as digging, transporting, and crushing.

[0003] In the existing technology, displacement sensors and pressure sensors are usually equipped on the hydraulic cylinders of excavators to determine the operating status of the excavator and thus control the excavator.

[0004] However, excavators encounter various complex operating conditions and scenarios during operation, which the sensors equipped on existing hydraulic cylinders cannot accurately detect. This results in the excavator's inability to dynamically adjust to the operating conditions. Therefore, existing technology suffers from the technical problem of not being able to accurately identify the type of digging medium during excavation, thus failing to effectively control the excavator's operating status. Summary of the Invention

[0005] This application provides a method for controlling the operating status of an excavator, which aims to accurately identify the type of excavating medium during excavation and thus effectively control the operating status of the excavator.

[0006] In a first aspect, embodiments of this application provide a method for controlling the operating state of an excavator, including:

[0007] The excavator's hydraulic cylinders are equipped with sensor arrays, and the inlet and outlet ports of the hydraulic cylinders are equipped with switching valve assemblies.

[0008] Acquire vibration signals and strain data collected by the sensor array;

[0009] Extract the vibration feature vector from the vibration signal and the strain feature vector from the strain data;

[0010] The vibration feature vector and strain feature vector are input into the excavation medium identification model to obtain the current excavation medium type of the excavator;

[0011] Based on the control type, determine the control operation corresponding to the current excavation medium type;

[0012] The control command corresponding to the control operation is sent to the valve assembly so that the valve assembly can perform the control operation.

[0013] One possible implementation involves extracting the vibration feature vector from the vibration signal and the strain feature vector from the strain data, including:

[0014] The vibration signal is processed using wavelet packet decomposition or fast Fourier transform algorithms to obtain vibration feature vectors.

[0015] Mechanical analysis is performed on the strain data to obtain the strain characteristic vector.

[0016] In one possible implementation, acquiring vibration signals and strain data collected by sensors mounted on the excavator includes:

[0017] At a preset sampling rate, the original vibration signal and original strain data collected by the sensor are acquired;

[0018] The original vibration is bandpass filtered to obtain the vibration signal;

[0019] Temperature compensation correction is performed on the original strain data to obtain the strain data.

[0020] In one possible implementation, based on the control type, the control operation corresponding to the current excavation medium type is determined, including:

[0021] When the control type is active protection control and the current excavation medium type is hard medium, the control operation is determined to be audible and visual alarm and hydraulic system depressurization.

[0022] In one possible implementation, based on the control type, the control operation corresponding to the current excavation medium type is determined, including:

[0023] When the control type is autonomous operation control, the excavation strategy corresponding to the current excavation medium type is determined based on the correspondence between the medium type and the excavation strategy; the excavation strategy is either an efficient excavation strategy or a demolition strategy.

[0024] In one possible implementation, after inputting the vibration feature vector and strain feature vector into the excavation medium identification model to obtain the current excavation medium type of the excavator, the method further includes:

[0025] Outputs the current excavation medium type in real time.

[0026] One possible implementation involves training the medium recognition model, including:

[0027] Construct a training dataset, which includes multiple sets of training data. Each set of training data includes the sample mining medium type, sample vibration signal, and sample strain data.

[0028] Extract the sample vibration features corresponding to the sample vibration signal, and extract the sample strain features corresponding to the sample strain data;

[0029] The vibration characteristics and strain characteristics of the samples are fused to obtain the sample fused characteristics;

[0030] The sample fusion features are input into the mining medium identification model to obtain the predicted mining medium type;

[0031] The loss function value between the sample mining medium type and the predicted mining medium type is calculated based on the preset loss function.

[0032] The excavation medium identification model is trained based on the loss function value until the preset loss function converges, thus obtaining the trained excavation medium identification model.

[0033] Secondly, this application provides an edge computing unit for controlling the operating state of an excavator, comprising:

[0034] The acquisition module is used to acquire vibration signals and strain data collected by the sensor array;

[0035] The extraction module is used to extract the vibration feature vector of the vibration signal and the strain feature vector of the strain data.

[0036] The processing module is used to input the vibration feature vector and strain feature vector into the excavation medium identification model to obtain the current excavation medium type of the excavator;

[0037] The determination module is used to determine the control operation corresponding to the current excavation medium type based on the control type;

[0038] The processing module is also used to send control commands corresponding to the control operation to the switching valve group so that the switching valve group can perform the control operation.

[0039] In one possible implementation, the extraction module is also used for:

[0040] The vibration signal is processed using wavelet packet decomposition or fast Fourier transform algorithms to obtain vibration feature vectors.

[0041] Mechanical analysis is performed on the strain data to obtain the strain characteristic vector.

[0042] In one possible implementation, the acquisition module is also used for:

[0043] At a preset sampling rate, the original vibration signal and original strain data collected by the sensor are acquired;

[0044] The original vibration is bandpass filtered to obtain the vibration signal;

[0045] Temperature compensation correction is performed on the original strain data to obtain the strain data.

[0046] In one possible implementation, the determining module is also used for:

[0047] When the control type is active protection control and the current excavation medium type is hard medium, the control operation is determined to be audible and visual alarm and hydraulic system depressurization.

[0048] In one possible implementation, the determining module is also used for:

[0049] When the control type is autonomous operation control, the excavation strategy corresponding to the current excavation medium type is determined based on the correspondence between the medium type and the excavation strategy; the excavation strategy is either an efficient excavation strategy or a demolition strategy.

[0050] In one possible implementation, the processing module is also used for:

[0051] Outputs the current excavation medium type in real time.

[0052] In one possible implementation, the processing module is also used for:

[0053] Construct a training dataset, which includes multiple sets of training data. Each set of training data includes the sample mining medium type, sample vibration signal, and sample strain data.

[0054] Extract the sample vibration features corresponding to the sample vibration signal, and extract the sample strain features corresponding to the sample strain data;

[0055] The vibration characteristics and strain characteristics of the samples are fused to obtain the sample fused characteristics;

[0056] The sample fusion features are input into the mining medium identification model to obtain the predicted mining medium type;

[0057] The loss function value between the sample mining medium type and the predicted mining medium type is calculated based on the preset loss function.

[0058] The excavation medium identification model is trained based on the loss function value until the preset loss function converges, thus obtaining the trained excavation medium identification model.

[0059] Thirdly, embodiments of this application provide a hydraulic cylinder, comprising:

[0060] The hydraulic cylinder body, the sensor array mounted on the hydraulic cylinder body, and the switching valve assembly mounted at the inlet and outlet of the hydraulic cylinder body;

[0061] The sensor array is connected to the edge computing unit, the switching valve assembly is connected to the edge computing unit, and the edge computing unit is used to perform the first aspect and / or various possible implementations of the first aspect;

[0062] The sensor array is used to collect vibration signals and strain data of the hydraulic cylinder body;

[0063] The valve assembly includes eight independently controlled valves. The valve assembly is used to adjust the opening degree of the corresponding valves according to the control commands of the edge computing unit, so as to realize the execution of the control operation corresponding to the control command.

[0064] In one possible implementation, the rated flow rate of the switching valve is configured according to a binary sequence.

[0065] Fourthly, this application provides an excavator equipped with a hydraulic cylinder.

[0066] Fifthly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0067] The memory stores computer-executed instructions;

[0068] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0069] In a sixth aspect, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0070] In a seventh aspect, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0071] The excavator operation state control method provided in this application embodiment achieves refined control of the excavator operation process by arranging a sensor array on the hydraulic cylinder of the excavator and precisely setting a switching valve group at the inlet and outlet of the hydraulic cylinder. It acquires vibration signals and strain data collected by the sensor array during the excavator operation process in real time, extracts vibration feature vectors characterizing the operation features from the vibration signals, and extracts strain feature vectors reflecting the stress state of the hydraulic cylinder from the strain data. The vibration feature vectors and strain feature vectors are input to a pre-trained excavation medium recognition model. Through intelligent analysis of the model, the type of medium currently being excavated by the excavator is accurately identified, such as soil, sand, stone, and other media with different physical properties. Based on the determined control type, a control operation adapted to the current excavation medium type is further determined, and the control command corresponding to the optimal control operation is precisely sent to the switching valve group. Through the precise execution of the switching valve group, precise control of the hydraulic cylinder inlet and outlet is achieved, thereby effectively adjusting the excavator's operation state so that it can adaptively adjust according to the characteristics of the current excavation medium. This method can significantly improve the operating efficiency and accuracy of excavators in different working environments, reduce equipment wear and energy waste caused by differences in media characteristics, and enhance the safety and stability of the operation process. It has important practical significance and far-reaching application value for promoting the development of intelligent and precise excavator operation technology. Attached Figure Description

[0072] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0073] Figure 1 Flowchart of the excavator operation status control method provided in this application Figure 1 ;

[0074] Figure 2 Flowchart of the excavator operation status control method provided in this application Figure 2 ;

[0075] Figure 3 This is a schematic diagram of the hydraulic cylinder structure provided in this application;

[0076] Figure 4 A schematic diagram of the edge computing unit for controlling the operating status of the excavator provided in this application;

[0077] Figure 5 A hardware schematic diagram of the edge computing unit for controlling the operating status of the excavator provided in this application.

[0078] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0079] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and approaches consistent with some aspects of this application as detailed in the appended claims.

[0080] In the current field of construction machinery, excavators often need to perform diverse tasks in unstructured scenarios such as mining, construction, and farmland reclamation. These complex working conditions are characterized by dynamic changes and uncertain geological conditions, posing severe challenges to the environmental adaptability and control precision of the equipment.

[0081] Existing technologies typically acquire basic operational parameters by configuring displacement and pressure sensors on hydraulic cylinders, attempting to infer the equipment's operating status by monitoring the cylinder's extension and retraction displacement and system pressure changes. However, this sensing method has significant limitations: firstly, displacement and pressure parameters only reflect the mechanical state of the actuator and cannot effectively capture the dynamic characteristics generated by the interaction between the excavating medium and the bucket; secondly, when facing complex conditions such as varying soil looseness, uneven rock layer distribution, or mixed materials, existing sensing systems struggle to accurately distinguish the differences in mechanical properties between different media, preventing the control system from dynamically adjusting operational parameters based on the actual working medium. This technical deficiency directly leads to insufficient adaptability and fluctuating operational efficiency when dealing with complex conditions, affecting not only construction accuracy and quality but also potentially causing a chain reaction of problems such as equipment overload and abnormal energy consumption due to mismatched operating conditions, thus hindering the further development of intelligent excavation technology.

[0082] In traditional operating modes, equipment cannot automatically sense changes in the physical properties of the actual excavated medium. This leads to the use of uniform operating parameters when facing different working conditions such as sand and rock, resulting in low energy efficiency, increased equipment wear and tear, and insufficient operational accuracy. To address this industry pain point, this invention creatively proposes to organically combine multi-source sensing technology, feature vector analysis, and intelligent recognition models to construct an integrated "perception-decision-execution" adaptive control system. This method uses a sensor array mounted on the hydraulic cylinder to capture vibration and strain data in real time, extracts feature vectors that characterize the mechanical properties of the medium, and then uses a pre-trained excavated medium recognition model for intelligent classification. Finally, it achieves dynamic adjustment of the hydraulic system by controlling the switching valve group. The significance of this method lies in upgrading excavation operations from a manual experience-based mode to a data-driven intelligent operation mode. This not only significantly improves the equipment's adaptability to different working conditions and operational economy but also provides a concrete technical path for the digital transformation of construction machinery, possessing significant practical value in promoting the industry towards intelligence, precision, and efficiency.

[0083] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0084] Figure 1 Flowchart of the excavator operation status control method provided in this application Figure 1 ,like Figure 1 As shown, the method includes:

[0085] S101. Acquire vibration signals and strain data collected by the sensor array.

[0086] In this embodiment, by integrating an array of multiple types of sensors onto the hydraulic cylinder, two key physical signals during operation can be acquired at high frequency and synchronously. Vibration signals directly reflect the impact and periodic fluctuations generated when the bucket contacts and cuts different media (such as soft soil and hard rock), and their frequency domain and amplitude characteristics directly reflect the hardness and density of the media. Strain data, on the other hand, realistically depicts the minute deformations of the hydraulic cylinder body caused by external loads, accurately conveying the changes in digging resistance. This step transforms complex and ambiguous on-site conditions into digital information that can be quantitatively analyzed and processed, providing a high-quality data foundation for subsequent intelligent identification and fundamentally avoiding the shortcomings of traditional single-sensor information, such as insufficient dimensionality and susceptibility to interference.

[0087] S102. Extract the vibration feature vector of the vibration signal and extract the strain feature vector of the strain data.

[0088] In this embodiment, it is difficult to directly obtain discriminative information from the original vibration signals and strain data. Therefore, it is necessary to extract the feature indicators that best represent the current working state through signal processing algorithms, quantify these indicators into structured feature vectors, realize the transformation from high-noise raw data to low-dimensional, high-value information, prepare a standardized "input language" for the efficient and accurate judgment of the intelligent model, and greatly improve the information density and recognizability of the data.

[0089] S103. Input the vibration feature vector and strain feature vector into the excavation medium identification model to obtain the current excavation medium type of the excavator.

[0090] In this embodiment, a pre-trained mining medium identification model is used to collect the fused multimodal feature vectors. Through complex internal nonlinear calculations, the model learns and matches the mapping relationship between different feature combinations and medium types, and finally outputs a probabilistic judgment about the current mining medium type.

[0091] S104. Based on the control type, determine the control operation corresponding to the current excavation medium type.

[0092] In this embodiment, pre-set control types optimized for different media types are used. For example, a "efficiency-first" mode with high flow and fast action is used for soft soil, while a "protection-first" mode with low flow, high pressure, and slow crushing is used for hard rock. This step transforms the cognitive results of the previous step into specific execution strategies. It achieves the refinement and personalization of control strategies, ensuring that the excavator's action parameters (such as the flow and pressure of the hydraulic system) can actively adapt to the physical characteristics of the work object, thereby pursuing optimal operating efficiency while ensuring equipment safety.

[0093] S105. Send the control command corresponding to the control operation to the valve assembly so that the valve assembly can perform the control operation.

[0094] In this embodiment, by equipping the excavator's hydraulic cylinders with sensor arrays and installing switching valve assemblies at the cylinder's inlet and outlet ports, specific, executable digital commands generated by the controller (such as adjusting valve opening or changing pump displacement) are sent to the switching valve assemblies at the cylinder's inlet and outlet ports. These high-response electro-hydraulic switching valve assemblies precisely adjust the flow and pressure of the hydraulic oil, thereby changing the output force and speed of the hydraulic cylinders, ultimately driving the excavator bucket to move in a manner best suited to the current medium. This step constitutes a complete closed loop of perception-decision-execution, effectively transforming the intelligent analysis results of the preceding steps into precise and adaptive adjustments to the equipment's operating state, ultimately achieving the core objectives of improving operational efficiency, protecting equipment, and reducing energy consumption.

[0095] The excavator operation status control method provided in this application embodiment collects vibration signals and strain data in real time by arranging a sensor array on the hydraulic cylinder, and extracts representative vibration feature vectors and strain feature vectors from them to form a key parameter system for identifying the characteristics of the medium. These feature vectors are input into a pre-trained excavation medium identification model, which comprehensively judges the type of medium at the current working surface, such as soil, sand, gravel, or mixed materials with different physical properties. The system automatically matches a preset control strategy according to the identified medium type, generates control commands optimally adapted to the current working condition, and sends them to the switching valve groups at the inlet and outlet of the hydraulic cylinder. By adjusting the hydraulic flow and pressure, precise control of the excavation action is achieved. This method realizes intelligent operation throughout the entire process from working condition perception and medium identification to execution control, effectively overcoming the limitations of traditional excavators that rely on manual experience for adjustment when facing complex and variable media. It significantly improves the equipment's adaptability to the working environment and overall construction efficiency, while also helping to reduce equipment wear and energy consumption, providing a feasible technical path for the intelligent upgrading of construction machinery.

[0096] Figure 2 Flowchart of the excavator operation status control method provided in this application Figure 2 ,like Figure 2 As shown, in this embodiment... Figure 1 Based on the embodiments, a method for controlling the operating status of an excavator is described in detail, which includes:

[0097] S201. Acquire the original vibration signal and original strain data collected by the sensor at a preset sampling rate; perform bandpass filtering on the original vibration to obtain the vibration signal; perform temperature compensation correction on the original strain data to obtain the strain data.

[0098] In this embodiment, the sensor array mounted on the hydraulic cylinder directly acquires raw signals containing a large amount of environmental noise, equipment mechanical noise, and temperature drift, among other interference factors. Directly using this raw data would severely interfere with the accuracy of subsequent feature extraction and model recognition. Therefore, step S201 uses targeted algorithms to filter and compensate for different types of noise and interference, providing "clean" input data for subsequent analysis.

[0099] In practical implementation, for vibration signals, since their effective components are usually concentrated within a specific frequency range (e.g., the low-to-medium frequency vibration generated by the bucket colliding with the medium during excavator operation), while high-frequency vibrations mostly originate from interference from unrelated components such as the engine, bandpass filtering becomes crucial. Therefore, it is necessary to set a frequency passband strongly correlated with the operation (e.g., 5Hz to 200Hz), allowing signal components within this band to pass through while significantly attenuating noise components outside this band, thereby obtaining a vibration signal that truly reflects the essence of the excavation action.

[0100] Strain gauges, while sensing mechanical deformation, are also extremely sensitive to changes in ambient temperature, leading to spurious strain readings caused by temperature variations in the measured values. Therefore, this step employs a temperature compensation correction algorithm. A "compensator" is placed in the measuring bridge, unloaded but under the same temperature environment. Through hardware or software algorithms, the pure temperature effect sensed by this compensator is subtracted from the total output, thus separating the true strain data purely caused by the mechanical load.

[0101] Through the targeted signal preprocessing described above, the obtained vibration signals and strain data exhibit higher signal-to-noise ratios and greater accuracy. For example, after bandpass filtering, the specific low-frequency impact waveform generated when the bucket cuts into the rock can be captured more clearly, rather than being drowned out by the high-frequency vibration of the engine; the strain data after temperature compensation can accurately reflect the changes in digging resistance, without exhibiting artificially high load readings when the temperature rises in the afternoon. This high-quality data input enables subsequent feature extraction and model recognition to be based on real working condition information, significantly improving the accuracy of media type identification and the robustness and reliability of the entire control system. This is a crucial first step in realizing the leap from "blind operation" to "precise perception" for excavators.

[0102] S202. The vibration signal is processed based on the wavelet packet decomposition algorithm or the fast Fourier transform algorithm to obtain the vibration feature vector; the strain data is mechanically analyzed to obtain the strain feature vector.

[0103] In this embodiment, the vibration feature vector of the vibration signal includes, but is not limited to, the root mean square (RMS), peak value, kurtosis, margin, and the energy proportion, dominant frequency, and spectral centroid of the vibration signal in different frequency bands (e.g., 0-100Hz, 100-500Hz, 500-2000Hz) by performing a fast Fourier transform (FFT) on the vibration data window every 100 milliseconds. The strain feature vector of the strain data includes, but is not limited to, the mean and variance of the strain values, and the strain gradient along the piston rod, used to characterize the load size and the degree of eccentricity.

[0104] This step transforms vibration and strain waveforms into feature vectors with clear physical meaning and high discriminative power. This process significantly compresses the amount of data and extracts the most crucial classification information. Instead of processing thousands of vibration data points per second, it summarizes them into a vector composed of a few key features such as "30% energy in frequency band A, 50% energy in frequency band B, and a center-of-gravity frequency of 150Hz."

[0105] S203. Input the vibration feature vector and strain feature vector into the excavation medium identification model to obtain the current excavation medium type of the excavator, and output the current excavation medium type in real time.

[0106] In this embodiment, the excavation medium identification model receives a comprehensive feature vector composed of vibration features (such as energy values ​​in each frequency band) and strain features (such as peak value, mean, and variance) from multiple dimensions. The model performs weighted calculations and nonlinear transformations on these features, ultimately generating a probability distribution for various preset medium types at the output layer. For example, the output of the excavation medium identification model might be [rock: 85%, hard soil: 12%, other: 3%], selecting "rock" with the highest probability as the current excavation medium type. The entire calculation process is highly optimized, ensuring extremely high processing speed.

[0107] Optionally, this step can continuously respond to changes in the external medium at a frequency of 10 times per second. From the moment the excavator bucket cuts into the material to the identification of the medium type, the latency of the entire process can be as low as 100 milliseconds. The judgment result is provided in real time in two ways: first, it is displayed on the interactive screen in the cab in the form of a prominent icon and text (such as "Rock Mode"), providing the operator with the most intuitive understanding of the working condition; second, it can be supplemented by voice broadcast (such as "Attention, current is hard rock layer"), enhancing the reminder without distracting the operator's attention. This real-time, intuitive feedback mechanism seamlessly connects the model's "intelligent judgment" with "human cognition," greatly reducing the reliance on the operator's experience.

[0108] In one possible implementation, the training process of the mining medium identification model can be obtained according to the following method: constructing a training dataset, which includes multiple sets of training data, each set of training data including the sample mining medium type, sample vibration signal, and sample strain data; extracting the sample vibration features corresponding to the sample vibration signal, and extracting the sample strain features corresponding to the sample strain data;

[0109] The vibration characteristics and strain characteristics of the samples are fused to obtain the sample fusion characteristics; the sample fusion characteristics are input into the excavation medium identification model to obtain the predicted excavation medium type; the loss function value between the sample excavation medium type and the predicted excavation medium type is calculated based on the preset loss function; the excavation medium identification model is trained according to the loss function value until the preset loss function converges to obtain the trained excavation medium identification model.

[0110] During the training of the excavation medium identification model, the dataset needs to cover massive amounts of data collected by excavators under various typical working conditions. Each training dataset is a complete learning sample, which includes not only sample vibration signals and sample strain data synchronously collected by sensor arrays, but also sample excavation medium types accurately labeled in advance through field surveys or experimental design, such as "hard granite," "wet clay," or "loose gravel." Feature fusion technology is used to obtain sample fusion features that can more comprehensively describe the current excavation state. These fusion features are input into the initial excavation medium identification model to be trained (its structure can be selected from classifiers such as neural networks and support vector machines). The model will output a predicted probability distribution for the medium type, i.e., the predicted excavation medium type. The training algorithm then uses a preset loss function, such as the cross-entropy loss function, to quantify the difference between the model's prediction results and the actual sample excavation medium types, objectively reflecting the prediction performance under the current model parameters. The essence of model training is an optimization process aimed at minimizing the loss value. Through iterative training, until the value of the preset loss function is observed to converge to a stable low point on the validation set and no longer decreases significantly, a well-trained mining media recognition model that can be used for actual deployment is obtained.

[0111] S204. Based on the control type, determine the control operation corresponding to the current excavation medium type; send the control command corresponding to the control operation to the switch valve group so that the switch valve group can execute the control operation.

[0112] In this embodiment, the implementation of specific control logic adapts to different operational needs through preset "control types." One possible implementation involves setting the control type to active protection control and identifying the current excavated medium as hard material. The control actions are determined to be audible and visual alarms and hydraulic system depressurization. Active protection control focuses on equipment safety and personnel alerts. When this mode is set and the current excavated medium is identified as hard material (such as high-strength rock or reinforced concrete), it immediately triggers multi-level protection responses.

[0113] Specifically, the control operation first includes activating the audible and visual alarms. Warning lights in the cab flash and a regular beeping sound occurs, while the display screen highlights "Hard medium, beware of impact," strongly alerting the operator to the potential overload risk to the equipment. Simultaneously, a pressure relief command is sent to the hydraulic system's valve assembly. This command does not completely stop operation but rather finely adjusts the valve opening to maintain the hydraulic cylinder's working pressure below a safe upper limit, intelligently limiting digging force and preventing irreversible damage to critical structural components such as the hydraulic rod and bucket from a direct impact. This method of warning before limiting force ensures safety while minimizing disruption to operational continuity.

[0114] In one possible implementation, when the control type is autonomous operation control, the digging strategy corresponding to the current digging medium type is determined based on the correspondence between the medium type and the digging strategy; the digging strategy is either an efficient digging strategy or a demolition strategy. Under this control type, the decision logic is upgraded from passive protection to active strategy planning. In this mode, a "medium type - digging strategy" mapping knowledge base is built-in. For example, when the medium is identified as loose sand, an efficient digging strategy is automatically matched. The control command generated at this time will guide the switching valve group to increase the hydraulic flow, so that the boom and bucket operate at a faster speed and with a larger stroke, maximizing the material handling capacity per unit time. Conversely, when the medium is identified as hard rock or a working condition requiring fine crushing, a demolition strategy is activated. At this time, the control command will instruct the switching valve group to adjust to a high-frequency, low-flow, high-pressure mode, so that the hydraulic cylinder drives the bucket to perform a short-stroke, high-frequency "pecking" action. This strategy can efficiently crush hard materials while minimizing the impact load on the equipment itself.

[0115] One possible implementation also needs to handle other common media types and demonstrate fine-grained control. For example, when the medium is identified as sticky mud, its tendency to "stick" to the bucket is assessed, triggering an anti-adhesion strategy. Under this strategy, during the bucket digging and unloading phases, the control command includes a subtle, high-frequency vibration signal sent to the switching valve assembly. This causes the hydraulic cylinder to drive the bucket in a slight vibration, effectively promoting mud slippage, keeping the bucket clean, and maintaining continuous and efficient operation. This refined control for specific working conditions fully demonstrates the significant advantages of intelligent control systems over traditional single control modes. It transforms the excavator from a blindly powerful tool into an intelligent partner capable of sensing the environment, devising strategies, and executing precisely.

[0116] In one possible implementation, the reason why the switching valve group can precisely control the hydraulic cylinder according to different excavation media is that it directly links the output of the excavation media identification model with the fine-grained flow and direction control capabilities of the hydraulic system. Eight independently controlled switching valves form a combinable flow distribution and pressure regulation structure, allowing the hydraulic cylinder's action characteristics to be dynamically adjusted according to the physical properties of the media, such as hardness and resistance. In terms of design principles, by determining the control operations corresponding to different media, quantitative requirements for the inlet and outlet flow, pressure, and speed of the two chambers of the hydraulic cylinder are obtained. The edge computing unit then formulates these requirements into opening and closing combinations and opening commands for the eight switching valves. For example, when facing hard rock, the inlet flow of the rodless chamber is reduced, the pressure is increased, and the return speed of the rod chamber is limited to enhance thrust and prevent shock. In soft soil conditions, the flow is increased and the pressure is reduced to achieve rapid operation. The independent controllability of the eight valves allows the excavator to combine main and bypass, high-speed and low-speed channels in the same hydraulic circuit, thereby achieving fine-grained adjustments similar to binary weights to meet the nonlinear demands of different strategies on hydraulic power.

[0117] Compared to traditional hydraulic cylinders controlled solely by proportional valves or simple directional valves, the switching valve assembly in this application, with its eight independent switching valves, allows the hydraulic cylinder not only to switch directions but also to achieve precise multi-level flow and pressure ratios within the same direction. This is particularly crucial when dealing with complex geological conditions. For example, when the model identifies the current medium as hard and the control type as active protection control, it will issue an "audible and visual alarm and hydraulic system depressurization" command. At this time, the edge computing unit will mostly close the two valves in the rodless chamber for oil inlet and partially open the two valves in the rod chamber for oil outlet to reduce system pressure. This alerts the operator and prevents overload damage to the equipment or loss of control caused by hard rock. Under autonomous operation control, when the medium is identified as sand and gravel, an efficient excavation strategy may be selected. The two valves in the rodless chamber for oil inlet will be fully open to provide a large flow for rapid extension, while the two valves in the rod chamber for oil outlet will be moderately opened to ensure return speed. At the same time, a bypass valve will be used to stabilize pressure and avoid shock. This design upgrades the excavation process from a fixed action mode to a dynamic mode that adapts to the medium, significantly improving operational efficiency and equipment safety.

[0118] The significance of configuring eight valves in a binary sequence with rated flow rates lies in the fact that a wider flow range can be covered with fewer valve combinations. This reduces hardware complexity while maintaining sufficient adjustment accuracy, enabling the optimal hydraulic conditions that balance power and speed to be found under different media. In actual operation, when encountering mixed strata or sudden hard interlayers, the control strategy is instantly identified and switched, and the valve combination changes accordingly. The hydraulic cylinder action shifts from rapid, light digging to high-pressure, slow digging or demolition, avoiding overload and energy waste. During long-term operation, this adaptive control also reduces mechanical shock and hydraulic system heat generation, extending the life of seals and pumps / valvees. Overall, this solution, by introducing an independently controllable eight-valve structure and intelligent identification linkage, elevates the hydraulic cylinder from a passive actuator to a core node in intelligent operation. This provides a solid execution foundation for the automation, unmanned operation, and adaptation to complex working conditions of excavators, and significantly improves the safety, efficiency, and durability of the entire machine.

[0119] In one possible implementation, even without using a valve group consisting of eight independent switching valves, adaptive control based on excavating medium identification can still be achieved using the excavator's existing hydraulic control system, such as a conventional hydraulic circuit with proportional valves, servo valves, or variable pumps. The only difference lies in the level of precision and response method. In existing excavator systems, the excavating medium identification model can still acquire vibration and strain data in real time through a sensor array, determine the current medium type, and output corresponding control strategies. Subsequently, the controller sends continuously varying current or pulse width modulation signals to the existing proportional or servo valves to adjust the flow and pressure of hydraulic oil entering each chamber of the hydraulic cylinder, thereby changing the piston's speed and thrust. For example, in hard rock conditions, reducing the proportional valve opening lowers the flow rate and increases the system pressure to prevent overload; in soft soil conditions, increasing the opening achieves rapid action. The original hydraulic control system of the excavator has a mature structure and smooth control, and can achieve continuous parameter adjustment. However, its limitation lies in the relatively limited adjustment range or resolution of the traditional valve group. When facing multi-level flow subdivision that requires near-binary weight combination, it may be necessary to rely on more complex valve body design or multiple valves in parallel, which is not as simple as independent on / off valve group, which can directly map multiple flow / pressure modes with simple ON / OFF combination. However, with reasonable control algorithm and high-resolution proportional valve, the original hydraulic control system can still achieve the basic goal of media self-adaptation. It just requires higher computing power of the controller and valve response accuracy. In some extreme working conditions, the protection and strategy switching speed may be slightly inferior to the on / off valve group solution designed for discrete combination.

[0120] Figure 3 A schematic diagram of a hydraulic cylinder structure is provided for this application, such as... Figure 3As shown, the hydraulic cylinder includes: a hydraulic cylinder body 301, a sensor array 303 disposed on the hydraulic cylinder body, and a switching valve group 304 disposed at the oil inlet and outlet of the hydraulic cylinder body. The excavator is also equipped with an edge computing unit 302.

[0121] The edge computing unit 302 is communicatively connected to the sensor array 303 and the switching valve group 304, respectively. The edge computing unit is used to execute the method of steps S101 to S104. The sensor array 303 is used to collect vibration signals and strain data of the hydraulic cylinder body. The switching valve group 304 includes eight independently controlled switching valves. The switching valve group is used to adjust the opening degree of the corresponding switching valve according to the control command of the edge computing unit, so as to realize the execution of the control operation corresponding to the control command.

[0122] like Figure 3 As shown, the hydraulic cylinder body 301 is a modified version of the original boom cylinder, including cylinder barrel 3011, piston 3012, piston rod 3013 and cylinder head 3014; the sensor array 303 includes vibration sensor 3031, pressure sensor 3032, temperature sensor 3033 and displacement sensor 3034; the switching valve group 304 contains 8 independently controlled switching valves, whose rated flow is configured in a binary sequence (1, 2, 4, 8, 16, 32, 64, 128 L / min) to achieve 256 levels of fine flow regulation.

[0123] In one possible implementation, vibration sensor 3031 consists of three triaxial MEMS accelerometers evenly distributed along a 120° circumference on the outer wall of the cylinder near the piston rod hinge end. This distributed layout can capture vibrations from different directions and, through data comparison, effectively eliminate interference from the excavator's own vibrations (such as those from the engine and slewing platform), focusing on the operational vibrations transmitted by the bucket teeth.

[0124] In one possible implementation, the pressure sensor 3032 is installed in the rodless chamber and the rod chamber of the hydraulic cylinder, with a range of 0-40 MPa.

[0125] In one possible implementation, the temperature sensor 3033 employs fiber optic grating (FBG) technology, which offers strong weather resistance and electromagnetic interference immunity. An armored optical fiber with 16 FBG sensing points internally etched is precisely bonded and cured spirally along the piston rod surface, starting from the hinge end. These 16 points can monitor the tensile, compressive, and bending strains at different locations on the piston rod during excavation. The fiber optic grating can also measure temperature for temperature compensation of the strain data.

[0126] In one possible implementation, the displacement sensor 3034 is located inside the hydraulic cylinder body 301, providing millimeter-level position feedback for the piston rod 3013.

[0127] Figure 4 A schematic diagram of the edge computing unit 302 for controlling the excavator's operating status provided in this application is shown below. Figure 4 As shown, the edge computing unit 302 for controlling the excavator's operating status provided in this embodiment includes:

[0128] The acquisition module 401 is used to acquire vibration signals and strain data collected by the sensor array;

[0129] Extraction module 402 is used to extract the vibration feature vector of the vibration signal and the strain feature vector of the strain data;

[0130] The processing module 403 is used to input the vibration feature vector and strain feature vector into the excavation medium identification model to obtain the current excavation medium type of the excavator;

[0131] The determination module 404 is used to determine the control operation corresponding to the current excavation medium type based on the control type;

[0132] The processing module 403 is also used to send control commands corresponding to the control operation to the switching valve group so that the switching valve group can perform the control operation.

[0133] In one possible implementation, the extraction module 402 is further used for:

[0134] The vibration signal is processed using wavelet packet decomposition or fast Fourier transform algorithms to obtain vibration feature vectors.

[0135] Mechanical analysis is performed on the strain data to obtain the strain characteristic vector.

[0136] In one possible implementation, the acquisition module 401 is further used for:

[0137] At a preset sampling rate, the original vibration signal and original strain data collected by the sensor are acquired;

[0138] The original vibration is bandpass filtered to obtain the vibration signal;

[0139] Temperature compensation correction is performed on the original strain data to obtain the strain data.

[0140] In one possible implementation, the determining module 404 is further used for:

[0141] When the control type is active protection control and the current excavation medium type is hard medium, the control operation is determined to be audible and visual alarm and hydraulic system depressurization.

[0142] In one possible implementation, the determining module 404 is further used for:

[0143] When the control type is autonomous operation control, the excavation strategy corresponding to the current excavation medium type is determined based on the correspondence between the medium type and the excavation strategy; the excavation strategy is either an efficient excavation strategy or a demolition strategy.

[0144] In one possible implementation, the processing module 403 is further used for:

[0145] Outputs the current excavation medium type in real time.

[0146] In one possible implementation, the processing module 403 is further used for:

[0147] Construct a training dataset, which includes multiple sets of training data. Each set of training data includes the sample mining medium type, sample vibration signal, and sample strain data.

[0148] Extract the sample vibration features corresponding to the sample vibration signal, and extract the sample strain features corresponding to the sample strain data;

[0149] The vibration characteristics and strain characteristics of the samples are fused to obtain the sample fused characteristics;

[0150] The sample fusion features are input into the mining medium identification model to obtain the predicted mining medium type;

[0151] The loss function value between the sample mining medium type and the predicted mining medium type is calculated based on the preset loss function.

[0152] The excavation medium identification model is trained based on the loss function value until the preset loss function converges, thus obtaining the trained excavation medium identification model.

[0153] The edge computing unit for controlling the excavator's operating status provided in this embodiment can execute the methods provided in the above-described method embodiments. Its implementation principle and technical effects are similar, and will not be described in detail here.

[0154] This embodiment describes an excavator equipped with a hydraulic cylinder as described in the above embodiment.

[0155] Figure 5 This is a hardware diagram of the edge computing unit for controlling the excavator's operating status provided in this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus 504.

[0156] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.

[0157] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0158] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0159] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0160] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0161] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0162] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0163] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0164] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0165] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, methods, or units, and may be electrical, mechanical, or other forms.

[0166] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0168] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0169] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0170] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A control method of an operation state of an excavator, a sensor array being provided on a hydraulic cylinder of the excavator, and a switching valve group being provided on an oil inlet and outlet of the hydraulic cylinder, characterized by, The method comprises: obtaining vibration signals and strain data collected by the sensor array; extracting vibration feature vectors of the vibration signals and strain feature vectors of the strain data; inputting the vibration feature vectors and the strain feature vectors into a medium identification model to obtain a current medium type of the excavator; determining a control operation corresponding to the current medium type based on a control type; sending a control instruction corresponding to the control operation to the switch valve group to enable the switch valve group to perform the control operation.

2. The method of claim 1, wherein, The method further comprises: processing the vibration signals based on a wavelet packet decomposition algorithm or a fast Fourier transform algorithm to obtain the vibration feature vectors; performing mechanical analysis on the strain data to obtain the strain feature vectors.

3. The method of claim 1, wherein, The method further comprises: obtaining original vibration signals and original strain data collected by the sensor at a preset sampling rate; performing band-pass filtering on the original vibration signals to obtain the vibration signals; performing temperature compensation correction on the original strain data to obtain the strain data.

4. The method of claim 1, wherein, The method further comprises: when the control type is active protection control and the current medium type is hard medium, determining the control operation to be sound-light alarm and hydraulic system pressure relief.

5. The method of claim 1, wherein, The method further comprises: when the control type is autonomous operation control, determining a digging strategy corresponding to the current medium type based on a correspondence between medium types and digging strategies; the digging strategy is a high-efficiency digging strategy or a breaking strategy.

6. The method of claim 1, wherein, The method further comprises: outputting the current medium type in real time.

7. The method of claim 1, wherein, The training process of the medium identification model comprises: constructing a training data set, wherein the training data set comprises a plurality of groups of training data, and each group of training data comprises a sample medium type, a sample vibration signal, and a sample strain data; extracting sample vibration features corresponding to the sample vibration signal and sample strain features corresponding to the sample strain data; fusing the sample vibration features and the sample strain features to obtain sample fusion features; inputting the sample fusion features into a medium identification model to obtain a predicted medium type; calculating a loss function value between the sample medium type and the predicted medium type based on a preset loss function; training the medium identification model based on the loss function value until the preset loss function converges, thereby obtaining a trained medium identification model.

8. A hydraulic cylinder characterized by, The hydraulic cylinder comprises: A hydraulic cylinder body, a sensor array arranged on the hydraulic cylinder body, and a switch valve group arranged on an oil inlet and outlet of the hydraulic cylinder body; The sensor array and an edge computing unit are connected, and the switch valve group and the edge computing unit are connected, and the edge computing unit is used for executing the method as claimed in any one of claims 1-7; The sensor array is used for collecting vibration signals and strain data of the hydraulic cylinder body; The switch valve group comprises eight independently controlled switch valves, and the switch valve group is used for adjusting the opening degree of the corresponding switch valve according to the control instruction of the edge computing unit, so as to realize the execution of the control operation corresponding to the control instruction.

9. The hydraulic cylinder of claim 8, wherein, The rated flow of the switch valve is configured in a binary sequence.

10. An excavator characterized by comprising: The excavator is provided with the hydraulic cylinder as claimed in claim 8 or 9.

Images