Electro-hydraulic control system for autonomous crushing of intelligent crushing robot and control method of electro-hydraulic control system

Abstract

The invention discloses an electro-hydraulic control system for autonomous crushing of an intelligent crushing robot and a control method thereof. According to the system, two working ports of an electro-hydraulic proportional reversing valve are connected with working ports A and B of a crushing hammer body through a first working oil way and a second working oil way correspondingly; an oil inlet and an oil outlet of the overflow valve are connected with the first working oil way and the second working oil way respectively. The filter assembly A and the energy accumulator are connected in series in the working oil way II; the heat dissipation assembly is connected to the main oil return way in series. The first pressure sensor and the second pressure sensor are connected with the first working oil way and the second working oil way respectively. The image point cloud collector is used for collecting image point cloud information; the crushing state recognition solver is connected with the pressure sensor and the image point cloud collector through the data collector; and the intelligent controller is connected with the crushing state recognition solver. The method comprises the steps that in the autonomous crushing control mode, multi-source monitoring data are collected in real time, feature extraction is conducted, the working state is classified and judged through the crushing state recognition model, and the intelligent controller executes a control strategy matched with the classification result. The intelligent crushing device can realize intelligent crushing operation.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for metal mining engineering machinery, specifically an electro-hydraulic control system and control method for autonomous crushing of an intelligent crushing robot. Background Technology

[0002] Intelligent crushing robots can perform operations such as ore crushing, rapid crushing of hard rock roof in metal mine roadways, and efficient mining of ore in metal mine tunneling roadways. The electro-hydraulic control system for autonomous crushing is a key component for achieving autonomous operation of the hydraulic breaker, playing a crucial role in the field of ore crushing. Crushing operations rely on the reciprocating motion of a piston within a cylinder, converting hydraulic energy into impact energy for rock crushing while impacting the chisel rod. When different ore crushing states are detected, the intelligent controller receives a signal and adjusts the opening of the proportional valve, thereby regulating the hydraulic oil pressure to output different impact kinetic energies. This addresses the current problem of the inability to autonomously determine the operation and effectively meets the requirement of stable impact kinetic energy output.

[0003] Currently, existing ore crushing control technologies do not identify and differentiate ore crushing conditions. When encountering complex working environments that cause constantly changing impact kinetic energy demands, the system struggles to determine the current crushing state and achieve precise impact kinetic energy output, easily leading to low crushing efficiency and thus failing to achieve high-efficiency crushing operations.

[0004] Based on the above problems, there is an urgent need to provide an electro-hydraulic control system and control method for crushing operations that can autonomously judge the crushing state and accurately control the impact kinetic energy output, so that the intelligent crushing robot can autonomously and adaptively adjust the impact kinetic energy according to the crushing state of the ore, and can perform autonomous control during the crushing operation, so as to ensure that the intelligent crushing robot can perform efficient crushing operations under different ore conditions. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an electro-hydraulic control system and its control method for autonomous crushing of an intelligent crushing robot. This system has a simple structure and ensures efficient crushing operations for the intelligent crushing robot under different ore conditions. It supports both manual and autonomous control, ensuring safe and efficient crushing operations. The method is simple to implement, low in cost, and highly intelligent. It can achieve high-precision identification of the crushing state and dynamically adjust the control strategy for different working conditions, enabling adaptive optimization of the crushing process. This significantly improves crushing efficiency, reduces energy consumption and equipment wear rate, and ensures the safety and stability of the crushing operation.

[0006] To achieve the above objectives, the present invention provides an electro-hydraulic control system for autonomous crushing of an intelligent crushing robot, comprising a crushing hammer body, a hydraulic control unit and an electrical control unit, wherein the crushing hammer body includes a crushing cylinder and a chisel rod. The hydraulic control unit includes a high-pressure oil supply mechanism, an electro-hydraulic proportional directional valve, an overflow valve, a filter assembly A, an accumulator, and a heat dissipation assembly. The working ports A and B of the electro-hydraulic proportional directional valve are connected to the working ports A and B of the hydraulic breaker body via working oil circuit one and working oil circuit two, respectively. The inlet P of the electro-hydraulic proportional directional valve is connected to the high-pressure oil supply mechanism, and its return port T is connected to the oil tank via the main return oil circuit. The inlet and outlet of the overflow valve are connected to working oil circuit one and working oil circuit two, respectively. The working ports of the filter assembly A and the accumulator are connected in series in working oil circuit two. The filter assembly A includes a filter A and a check valve A connected in parallel. The heat dissipation assembly is connected in series in the main return oil circuit. The heat dissipation assembly includes a back pressure valve one, a radiator, and a back pressure valve two. The back pressure valve one and the radiator are connected in series to form a heat dissipation branch oil circuit, and the back pressure valve two is connected in parallel with the heat dissipation branch oil circuit. The set pressure of the back pressure valve two is higher than the set pressure of the back pressure valve one. The electronic control unit includes a pressure sensor 1, a flow sensor, a pressure sensor 2, an acceleration sensor, an image point cloud collector, a temperature sensor, a data collector, a crushing state identification solver, an intelligent controller, and a control handle. Pressure sensor 1 and the flow sensor are both connected to working oil circuit 1; pressure sensor 2 is connected to working oil circuit 2; the acceleration sensor is embedded in the chisel rod; the image point cloud collector is mounted on one side of the breaker body and is used to collect dynamic image data and point cloud information of the ore during the breaker's operation; the temperature sensor is connected to a heat dissipation branch; the data collector is connected to pressure sensor 1, the flow sensor, pressure sensor 2, the acceleration sensor, the temperature sensor, and the image point cloud collector; the crushing state identification solver is connected to the data collector and is used to classify the crushing working state; the intelligent controller is connected to the crushing state identification solver, the electro-hydraulic proportional directional valve, and the radiator, and is used to implement intelligent control strategies matched to the working state; the control handle is connected to the intelligent controller.

[0007] As a preferred embodiment, the hydraulic control unit further includes a bidirectional hydraulic lock, with two connecting oil circuits of the bidirectional hydraulic lock respectively connected in series in working oil circuit one and working oil circuit two; the bidirectional hydraulic lock includes a first hydraulic control check valve and a second hydraulic control check valve, with the hydraulic control port of the first hydraulic control check valve connected to the oil outlet of the second hydraulic control check valve, and the hydraulic control port of the second hydraulic control check valve connected to the oil outlet of the first hydraulic control check valve. Through the bidirectional hydraulic lock, the rod-side and rodless chambers of the breaker cylinder can be bidirectionally pressure-maintained and locked, preventing the piston from shifting due to external forces or system pressure fluctuations, thus ensuring the stability, safety, and accuracy of the breaker operation.

[0008] As a preferred embodiment, the hydraulic control unit further includes a filter assembly B, which is connected in series in the main return oil circuit and located downstream of the heat dissipation assembly. The filter assembly B includes a filter B and a one-way valve B connected in parallel. The filter assembly B effectively prevents impurities from entering the oil tank. Furthermore, the parallel connection of the one-way valve and the filter ensures that oil flow interruptions will not occur due to filter blockage.

[0009] As a preferred embodiment, the image point cloud acquisition device includes an explosion-proof camera and a lidar; the acquisition angle of the explosion-proof camera is 45 degrees to the horizontal plane. By simultaneously setting up an explosion-proof camera and a lidar, the advantages of visual information and three-dimensional spatial information can be complemented, overcoming the limitations of a single sensor, and outputting high-precision and highly robust ore volume data.

[0010] In this invention, an overflow valve connects working oil circuits one and two. In the event of stuck drill or difficult impact conditions, the overpressure oil in working oil circuit one can flow through a flow valve into working oil circuit two, and then into the accumulator for storage, or flow back to the oil tank through the return port T of the electro-hydraulic proportional directional valve. In the heat dissipation assembly, the set pressure of back pressure valve two is higher than that of back pressure valve one, allowing the return oil to preferentially flow back to the oil tank through the heat dissipation branch oil circuit. This facilitates cooling of the hydraulic oil using a radiator when the oil temperature exceeds the limit, thereby ensuring the working efficiency of the hydraulic system. Pressure signals from working oil circuits one and two are collected using pressure signal one and pressure signal two, respectively. This allows for real-time pressure data analysis to determine whether the hydraulic impact hammer is operating normally under normal conditions. Furthermore, in the event of stuck drill, the impact operation can be stopped by switching the valve position of the electro-hydraulic proportional directional valve, and the pressure status of the working oil circuits can be further monitored for rapid reset of the stuck drill. By installing an accumulator in the second working oil circuit, pressure pulsations and impacts during the operation of the hydraulic breaker can be absorbed, thus ensuring the stability of the hydraulic system pressure. The filter assembly A prevents impurities from entering the accumulator or the electro-hydraulic proportional directional valve, ensuring the service life of hydraulic components. Simultaneously, the parallel connection of the check valve and filter ensures that oil flow interruptions due to filter blockage will not occur. The flow sensor facilitates the acquisition of flow signals from the hydraulic cylinder supply circuit. The acceleration sensor facilitates the real-time acquisition of the chisel rod's acceleration signals. The image point cloud acquisition device facilitates the real-time acquisition of images and point cloud data of the ore during operation. The temperature sensor facilitates real-time monitoring of the oil temperature in the cooling branch circuit, allowing automatic radiator activation when the temperature exceeds a set threshold. The data acquisition device facilitates noise reduction and spatiotemporal synchronization processing of multi-source monitoring signals, ensuring the quality of the monitoring data. The crushing state identification solver facilitates accurate classification of the working state based on multi-source monitoring data. By configuring the intelligent controller, it is easy to execute appropriate intelligent control strategies based on the classification results of working conditions. By configuring the control handle, it is easy to receive real-time control from the operator, so as to adapt to complex working conditions. Therefore, the system can easily support both manual and automatic control modes.

[0011] The system has a simple structure and can ensure that the intelligent crushing robot can carry out efficient crushing operations under different ore conditions. It supports both manual and autonomous control operations and can ensure the safe and efficient operation of crushing operations.

[0012] This invention also provides a control method for an electro-hydraulic control system for autonomous crushing of an intelligent crushing robot, comprising the following steps: Step 1: Selecting the working mode; The operator sends a manual remote control signal or an autonomous crushing signal through the control handle in the cab. After the intelligent controller receives the manual remote control signal, it executes step two; after receiving the autonomous crushing signal, it executes step three. Step 2: Manual remote control mode; The operator sends control signals through the control handle in the cab, and the intelligent controller controls the electro-hydraulic proportional directional valve according to the received control signals, so that the main body of the breaker hammer works in manual remote control mode. Step 3: Autonomous Crushing Control Mode; S1: Data collection during autonomous crushing operations; Within a set time period, the intelligent controller automatically controls the switching action of the electro-hydraulic proportional directional valve according to the preset timing control logic, enabling the breaker hammer to autonomously perform crushing operations. At the same time, the data acquisition unit collects image data and point cloud information of the ore during the operation through the image point cloud acquisition unit, collects the acceleration signal of the chisel rod during the operation through the acceleration sensor, and collects pressure signal one and pressure signal two in working oil circuit one and working oil circuit two through pressure sensor one and pressure sensor two, respectively. After the data acquisition unit performs noise reduction processing on the multi-source monitoring data, it sends it to the crushing state identification solver. S2: Construct a classification dataset through feature extraction; S21: Based on pressure data, a moving average is calculated on the pressure values ​​within a fixed time window, and the resulting average pressure value is used as the pressure feature vector. S22: Based on acceleration data, peak acceleration data are statistically analyzed within a fixed time window, and the impact frequency is calculated as the acceleration feature vector. S23: Within a fixed time window, input the image data and point cloud information into a deep learning model based on an improved Mask R-CNN, extract the pixel size of each ore bounding box, and calculate the average stone area as the image feature vector. S24: Transfer the pressure feature vector Acceleration eigenvectors and image feature vectors The features are concatenated to form a fused feature vector. , S24: Based on fused feature vectors A sample dataset is constructed based on the corresponding working status, and the sample dataset is divided into a training set and a test set according to a set ratio. S3: Construct a fractured state identification model; The hyperparameters of the SVM state classification model are optimized based on the sparrow search algorithm to obtain the optimal hyperparameters; S33: using fused feature vectors Using the working state as the input data and the working state as the output data, the SVM state classification model is trained using the training set based on the optimal hyperparameters, and the performance of the trained SVM state classification model is evaluated using the test set. The SVM state classification model with the best performance is saved to the breakage state identification solver as the breakage state identification model. S4: Autonomous crushing operation based on real-time crushing status recognition; S41: The data acquisition unit collects multi-source monitoring data through an image point cloud acquisition unit, an accelerometer, pressure sensor one, and pressure sensor two, and sends the data to the fracture state identification solver after noise reduction processing; S42: A fused feature vector is extracted based on the multi-source monitoring data. As input data, the working state is classified and determined using the crushing state recognition model, and the classification result is output to the intelligent controller; S43: The intelligent controller controls the crushing process by adopting a control strategy that matches the classification result, so that the intelligent crushing robot works in an autonomous crushing mode based on the crushing state recognition.

[0013] As a preferred approach, in step S42 of step three, the classification results include normal crushing 1, normal crushing 2, dry running, stuck drill, and crushed work area. In step S43 of step three, the control strategy matching the classification results is as follows: When it is normal crushing 1 or normal crushing 2, the conventional incremental PID control method based on energy error feedback is used to control the crushing process; when it is dry running, the electro-hydraulic proportional directional valve is de-energized and operates in the neutral position to stop the impact operation, and a reminder signal is sent to the intelligent crushing robot to adjust the working position of the main body of the breaker hammer; when it is stuck drill, the pressure state of the chisel rod is first judged by pressure sensor one and pressure sensor two, then the electro-hydraulic proportional directional valve is de-energized and operates in the neutral position to stop the impact operation. Then, when the pressure sensor one and pressure sensor two are stable, the electro-hydraulic proportional directional valve is switched to the reverse impact position to release the chisel rod by reversing the impact force; when it is a crushed work area, a reminder signal is sent to the intelligent crushing robot to move to the new work area. Under normal operating conditions, incremental PID control is used to ensure that the impact energy is dynamically adjusted according to the load, thereby improving crushing efficiency. In dry operation, the machine is stopped in time and a position adjustment reminder is given to avoid ineffective impacts and reduce energy consumption and equipment wear. In the case of stuck drill, the machine is stopped first and then reversed to quickly release the chisel rod, reducing the risk of equipment failure. The machine reminds users to move the work area after the crushed area has been crushed, avoiding repetitive work and improving overall work efficiency.

[0014] As a preferred option, in step S21 of step three, the first step is obtained according to formula (1). Average pressure over a time window According to formula (2), the first... Impact frequency within a time window According to formula (3), the first... Area of ​​each stone ; (1); In the formula, For window size, Relative to the current moment The A historical pressure value; (2); In the formula, For the first The number of peak values ​​within a window The length of the window; (3); In the formula, This represents the total number of stones detected in the image.

[0015] As a preferred option, in step three, S43, the control process of the conventional incremental PID control method based on energy error feedback is as follows: S43-1: First, identify the size of the ore volume based on real-time image data and point cloud information, and then calculate the ore mass based on the ore density. The unit is kg; then, the theoretical impact energy required for the hydraulic breaker is calculated according to formula (4). ; (4); In the formula, The piston speed in the hydraulic crusher cylinder is expressed in m / s. S43-2: Based on theoretical impact energy using fuzzy algorithms The initial control current signal is mapped and output to the electro-hydraulic proportional directional valve. At the same time, based on the feedback pressure and flow data, the energy transmission loss rate is calculated according to formula (5). ; (5); In the formula, The time required for the piston to complete one impact, measured in seconds; The pressure data is based on pressure sensor 1, and the unit is Pa. The data is based on flow sensor data, and the unit is L / s; The impulse frequency is expressed in Hz. , In time The number of impacts measured internally; S43-3: Calculate the actual impact energy required according to formula (6). ; (6); S43-4: Calculate the first according to formula (7) Error at each sampling time; (7); S43-5: Using the incremental PID control algorithm, the control increment is calculated according to formula (8). ; (8); In the formula, Sampling period; The sampling sequence number; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; S43-6: Obtain the control output at the current moment according to formula (9). ; (9); S43-7: Obtain the proportional valve current signal at the current moment according to formula (10). ; (10); In the formula, , These are the minimum and maximum allowable currents for the electro-hydraulic proportional directional valve, respectively, in mA. Limit the PID output; S43-8: Output proportional valve current signal This leads to the electro-hydraulic proportional directional valve, enabling precise control of the valve opening within the valve and simultaneously performing adaptive control adjustments. The adaptive control adjustment process is as follows: If... Then increase the proportional valve current signal By increasing the valve opening to increase the energy of the next impact, if This reduces the proportional valve current signal. The energy of the next impact is reduced by decreasing the valve opening. At the same time, the ore damage is fed back in real time by combining image data and point cloud information to ensure that the ore crushing degree meets the requirements, and the feedback on ore damage is used as an auxiliary basis for subsequent control and adjustment.

[0016] As a preferred option, in step S1 of step three, the multi-source monitoring data is denoised using discrete wavelet transform according to formula (11); (11); In the formula, These are discrete wavelet coefficients; The wavelet transform scaling parameter; These are translation parameters; The original input signal; It is a wavelet function.

[0017] As a preferred option, in step S3 of step three, the process of optimizing the hyperparameters of the SVM state classification model based on the sparrow search algorithm is as follows: S31: Initialize SSA parameters; set population size n, maximum number of iterations G, number of producers PD, number of scouts SD, and safety threshold. , ; S32: Initial population position generation; randomly generate the initial positions of N sparrow individuals. The position of each sparrow is represented by a two-dimensional vector (C, g), where C is the error penalty factor and g is the kernel function parameter, that is, each sparrow corresponds to a set of parameters of the SVM state classification model. S33: Iterative optimization and position update; S33-1: Fitness Calculation and Extreme Value Screening; Calculate the fitness values ​​of all sparrows, screen out the current globally best and worst individuals, and randomly generate alarm values. , ; S33-2: Update producer position; Update producer position according to formula (12) ; (12); In the formula, This represents the number of iterations. For the first Only sparrows in the first In the nth iteration The value of the dimension; This represents the maximum number of iterations. These are random numbers that follow a normal distribution. For elements all equal to 1 matrix; ; It is a random number. S33-3: Forager position update; update the forager's position according to formula (13). ; (13); In the formula, The optimal position occupied by the producer; This is the worst position globally at present; For elements that are all 1 or -1 matrix; ; S33-4: Scout position update; update the scout's position according to formula (14). ; (14); In the formula, This is the current globally optimal position; These are normally distributed random numbers with a mean of 0 and a variance of 1. It is a random number. ; This represents the current fitness value of the sparrow. , These are the current best and worst fitness values, respectively. The smallest constant to avoid division by zero errors; S33-5: Position update judgment; compare the fitness value of each sparrow's new position with its original position. If the new position has a better fitness value, then update to the new position; otherwise, retain the original position. S34: Iteration Termination and Result Output; Determine if the current iteration count has reached the preset maximum iteration count G. If it has, stop the iteration and output the global optimal position. and global optimal fitness If the target is not reached, return to S33 for the next iteration.

[0018] This invention provides a control method for an intelligent crushing robot's autonomous electro-hydraulic control system. First, the operating mode includes either manual remote control or autonomous crushing control, supporting switching between the two modes. This retains the operator's right to intervene manually in complex and special working conditions while enabling autonomous operation under normal conditions, thus improving the flexibility of crushing operations. Next, in manual remote control mode, the operator can directly send commands via a control handle in the cab, conforming to traditional engineering machinery operating habits. Simultaneously, the intelligent controller directly drives the electro-hydraulic proportional directional valve based on the operator's manual control, precisely controlling the start / stop and impact intensity of the breaker hammer to meet personalized operational needs. Second, in autonomous crushing control mode, by collecting three key data types—image point cloud data, acceleration signals, and hydraulic pressure signals—the system can reflect the ore crushing morphology, chisel impact characteristics, and hydraulic system operating conditions, ensuring the integrity of the basic data dimensions and providing sufficient basis for accurate subsequent status identification. Furthermore, by denoising the monitoring data, invalid information such as sensor noise and environmental interference can be effectively filtered out, improving the quality of the original data and reducing errors in subsequent model training. Subsequently, differentiated extraction methods were designed for different types of data. For pressure data, the moving average method was used to effectively highlight the steady-state characteristics of pressure. For acceleration data, peak values ​​and impact frequencies were statistically analyzed to effectively reflect impact intensity and frequency. For image point cloud data, the size and area of ​​ore were extracted by improving Mask R-CNN, which can fully characterize the crushing effect, making the features highly correlated with the crushing state. The three types of features—pressure, acceleration, and image—were concatenated into a fusion feature vector F to achieve multi-dimensional information complementarity, avoid the limitations of single features, and improve the accuracy of subsequent model classification. Furthermore, the Sparrow Search Algorithm (SSA) was used to optimize the error penalty factor C and kernel function g of the SVM. Compared with traditional methods such as grid search, it has the characteristics of fast convergence speed and strong optimization ability, and can quickly find the optimal hyperparameter combination. The crushing state recognition model constructed using the SVM model has the advantages of simple structure and low computational requirements. At the same time, combined with the hyperparameters optimized by SSA, it can run quickly in the intelligent controller to meet the needs of real-time state recognition. Then, multi-source data is collected in real time and fused features are extracted. The fragmentation state is quickly determined through a pre-trained model, realizing a closed-loop control of "collection-identification-decision" and ensuring real-time response. Finally, dedicated control strategies are formulated for different working states to accurately match the requirements of the working conditions.

[0019] This method is simple to implement, low in cost, and highly intelligent. By fusing three types of data—image data, acceleration, and pressure—and combining feature engineering with the SSA-SVM model, it achieves high-precision identification of the crushing state and dynamically adjusts the control strategy for different working conditions. This enables adaptive optimization of the crushing process, significantly improving crushing efficiency, reducing energy consumption and equipment wear, and ensuring the safety and stability of crushing operations. Furthermore, this method can precisely adjust the impact frequency of the crushing robot and the pressure of the hydraulic system, solving the problem of insufficient impact kinetic energy preventing normal operation. This allows the intelligent crushing robot to autonomously perform crushing operations based on the size of the ore, enhancing its autonomous operation capabilities. Attached Figure Description

[0020] Figure 1 This is the electro-hydraulic schematic diagram of the system in this invention; Figure 2 This is a flowchart of the method in this invention; Figure 3 This is a flowchart illustrating the optimization of hyperparameters of the SVM state classification model based on the sparrow search algorithm in this invention.

[0021] In the diagram: 1. Oil pump, 2. Motor, 3. Oil tank, 4. Electro-hydraulic proportional directional valve, 5. Two-way hydraulic lock, 6. Working oil circuit one, 7. Working oil circuit two, 8. Overflow valve, 9. Second hydraulically controlled check valve, 10. Crushing cylinder, 11. Accumulator, 12. Pressure sensor one, 13. Pressure sensor two, 14. Flow sensor, 15. Accelerometer, 16. Crusher body, 17. Chisel rod, 18. Image point cloud collector, 19. Data collector, 20. Crushing state identification solver, 21. Intelligent controller, 22. Main return oil circuit, 23. Back pressure valve one, 24. Radiator, 25. Filter B, 26. Back pressure valve two, 27. Check valve B, 28. Control handle, 29. Filter A, 30. Check valve A, 31. First hydraulically controlled check valve. Detailed Implementation

[0022] The invention will now be further described with reference to the accompanying drawings.

[0023] like Figure 1 As shown, the present invention provides an electro-hydraulic control system for autonomous crushing of an intelligent crushing robot, including a crusher body 16, a hydraulic control unit and an electrical control unit. The crusher body 16 includes a crushing cylinder 10 and a chisel rod 17. A nitrogen chamber is provided on one side of the cylinder end of the crushing cylinder 10 and nitrogen is added. The hydraulic control unit includes a high-pressure oil supply mechanism, an electro-hydraulic proportional directional valve 4, an overflow valve 8, a filter assembly A, an accumulator 11, and a heat dissipation assembly. The working ports A and B of the electro-hydraulic proportional directional valve 4 are connected to the working ports A and B of the hydraulic breaker body 16 via working oil circuit 1 6 and working oil circuit 2 7, respectively. The inlet P of the electro-hydraulic proportional directional valve 4 is connected to the high-pressure oil supply mechanism, and its return port T is connected to the oil tank 3 via the main return oil circuit 22. The inlet and outlet of the overflow valve 8 are connected to working oil circuit 1 6 and working oil circuit 2 7, respectively. 2.7; The working oil ports of the filter assembly A and the accumulator 11 are connected in series in the working oil circuit 2.7; The filter assembly A includes a filter A29 and a one-way valve A30 connected in parallel; The heat dissipation assembly is connected in series on the main return oil circuit 22; The heat dissipation assembly includes a back pressure valve 23, a radiator 24 and a back pressure valve 26, the back pressure valve 23 and the radiator 24 are connected in series to form a heat dissipation branch oil circuit, the back pressure valve 26 is connected in parallel with the heat dissipation branch oil circuit, and the set pressure of the back pressure valve 26 is higher than the set pressure of the back pressure valve 23; As a preferred embodiment, the electro-hydraulic proportional directional valve 4 is a three-position four-way directional valve. When it is energized and operating in the upper position (positive impact position), the oil circuit between its inlet port P and working port A is connected, and the oil circuit between its return port T and working port B is connected. When it is de-energized and operating in the middle position, its inlet port P, return port T, working port A and working port B are all cut off. When it is energized and operating in the lower position (reverse impact position), the oil circuit between its inlet port P and working port B is connected, and the oil circuit between its return port T and working port A is connected. As a preferred option, the accumulator 11 is a diaphragm accumulator; As a preferred embodiment, the high-pressure oil supply mechanism includes an oil pump 1 and a motor 2; the oil pump 1's suction port is connected to the oil tank 3 through an oil suction branch, and the motor 2 is coaxially connected to the oil pump 1; The electronic control unit includes a pressure sensor 12, a flow sensor 14, a pressure sensor 13, an acceleration sensor 15, an image point cloud collector 18, a temperature sensor 31, a data collector 19, a breakage state identification solver 20, an intelligent controller 21, and a control handle 28. The pressure sensor 12 and flow sensor 14 are both connected to the working oil circuit 6, and are used to collect pressure signal 1 and flow signal 2, respectively. The pressure sensor 13 is connected to the working oil circuit 7, and is used to collect pressure signal 2. The acceleration sensor 15 is embedded in the chisel rod 17 and is used to collect acceleration signals. The image point cloud collector 18 is supported on one side of the outside of the breaker hammer body 16 and is used to collect... The system collects dynamic image data and point cloud information of the ore during the operation of the hydraulic breaker body 16; the temperature sensor 31 is connected to the heat dissipation branch and is used to collect the temperature signal of the heat dissipation branch oil circuit; the data acquisition unit 19 is connected to the pressure sensor 12, the flow sensor 14, the pressure sensor 13, the acceleration sensor 15, and the image point cloud acquisition unit 18 respectively; the crushing state identification solver 20 is connected to the data acquisition unit 19 and is used to classify the working state; the intelligent controller 21 is connected to the crushing state identification solver 20, the electro-hydraulic proportional directional valve 4, and the radiator 24 respectively and is used to implement an intelligent control strategy that matches the working state; the control handle 28 is connected to the intelligent controller 21.

[0024] As a preferred embodiment, the hydraulic control unit further includes a bidirectional hydraulic lock 5, with two connecting oil circuits of the bidirectional hydraulic lock 5 respectively connected in series in working oil circuit one 6 and working oil circuit two 7; the bidirectional hydraulic lock 5 includes a first hydraulic control check valve 31 and a second hydraulic control check valve 32, the hydraulic control port of the first hydraulic control check valve 31 is connected to the oil outlet of the second hydraulic control check valve 32, and the hydraulic control port of the second hydraulic control check valve 32 is connected to the oil outlet of the first hydraulic control check valve 31. Through the bidirectional hydraulic lock, the rod chamber and rodless chamber of the breaker cylinder can be bidirectionally pressure maintained and locked, preventing the piston from shifting due to external force or system pressure fluctuations, thus ensuring the stability, safety, and accuracy of the breaker operation.

[0025] As a preferred embodiment, the hydraulic control unit further includes a filter assembly B, which is connected in series in the main return oil circuit 22 and located downstream of the heat dissipation assembly. The filter assembly B includes a filter B25 and a one-way valve B27 connected in parallel. The filter assembly B effectively prevents impurities from entering the oil tank. Furthermore, the parallel connection of the one-way valve and the filter ensures that oil flow interruptions will not occur due to filter blockage.

[0026] As a preferred embodiment, the image point cloud acquisition device 18 includes an explosion-proof camera and a lidar; the explosion-proof camera is positioned at a 45-degree angle to the horizontal plane, allowing for the acquisition of video information about the ore crushing process at a 45-degree angle. The lidar is used to acquire the point cloud information of the ore. By simultaneously setting up the explosion-proof camera and lidar, the advantages of visual information and three-dimensional spatial information can be complemented, overcoming the limitations of a single sensor and outputting high-precision, highly robust ore volume data.

[0027] In this invention, an overflow valve connects working oil circuits one and two. In the event of stuck drill or difficult impact conditions, the overpressure oil in working oil circuit one can flow through a flow valve into working oil circuit two, and then into the accumulator for storage, or flow back to the oil tank through the return port T of the electro-hydraulic proportional directional valve. In the heat dissipation assembly, the set pressure of back pressure valve two is higher than that of back pressure valve one, allowing the return oil to preferentially flow back to the oil tank through the heat dissipation branch oil circuit. This facilitates cooling of the hydraulic oil using a radiator when the oil temperature exceeds the limit, thereby ensuring the working efficiency of the hydraulic system. Pressure signals from working oil circuits one and two are collected using pressure signal one and pressure signal two, respectively. This allows for real-time pressure data analysis to determine whether the hydraulic impact hammer is operating normally under normal conditions. Furthermore, in the event of stuck drill, the impact operation can be stopped by switching the valve position of the electro-hydraulic proportional directional valve, and the pressure status of the working oil circuits can be further monitored for rapid reset of the stuck drill. By installing an accumulator in the second working oil circuit, pressure pulsations and impacts during the operation of the hydraulic breaker can be absorbed, thus ensuring the stability of the hydraulic system pressure. The filter assembly A prevents impurities from entering the accumulator or the electro-hydraulic proportional directional valve, ensuring the service life of hydraulic components. Simultaneously, the parallel connection of the check valve and filter ensures that oil flow interruptions due to filter blockage will not occur. The flow sensor facilitates the acquisition of flow signals from the hydraulic cylinder supply circuit. The acceleration sensor facilitates the real-time acquisition of the chisel rod's acceleration signals. The image point cloud acquisition device facilitates the real-time acquisition of images and point cloud data of the ore during operation. The temperature sensor facilitates real-time monitoring of the oil temperature in the cooling branch circuit, allowing automatic radiator activation when the temperature exceeds a set threshold. The data acquisition device facilitates noise reduction and spatiotemporal synchronization processing of multi-source monitoring signals, ensuring the quality of the monitoring data. The crushing state identification solver facilitates accurate classification of the working state based on multi-source monitoring data. By configuring the intelligent controller, it is easy to execute appropriate intelligent control strategies based on the classification results of working conditions. By configuring the control handle, it is easy to receive real-time control from the operator, so as to adapt to complex working conditions. Therefore, the system can easily support both manual and automatic control modes.

[0028] The system has a simple structure and can ensure that the intelligent crushing robot can carry out efficient crushing operations under different ore conditions. It supports both manual and autonomous control operations and can ensure the safe and efficient operation of crushing operations.

[0029] like Figure 2 and Figure 3As shown, the present invention also provides a control method for an electro-hydraulic control system for autonomous crushing of an intelligent crushing robot, comprising the following steps: Step 1: Selecting the working mode; The operator sends a manual remote control signal or an autonomous crushing signal through the control handle 28 in the cab. After the intelligent controller 21 receives the manual remote control signal, it executes step two; after receiving the autonomous crushing signal, it executes step three. Step 2: Manual remote control mode; The operator sends a control signal through the control handle 28 in the cab. The intelligent controller 21 controls the action of the electro-hydraulic proportional directional valve 4 according to the received control signal, so that the main body of the breaker hammer 16 works in the manual remote control working state. Step 3: Autonomous Crushing Control Mode; S1: Data collection during autonomous crushing operations; Within a set time period, the intelligent controller 21 automatically controls the switching action of the electro-hydraulic proportional directional valve 4 according to the preset timing control logic, so that the main body of the breaker hammer 16 can perform crushing operations autonomously. At the same time, the data acquisition unit 19 collects image data and point cloud information of the ore during the operation through the image point cloud acquisition unit 18, collects the acceleration signal of the chisel rod 17 during the operation through the acceleration sensor 15, and collects the pressure signal 1 and pressure signal 2 in the working oil circuit 16 and working oil circuit 27 respectively through the pressure sensor 12 and pressure sensor 23. After the data acquisition unit 19 performs noise reduction processing on the multi-source monitoring data, it sends it to the crushing state identification solver 20. S2: Construct a classification dataset through feature extraction; S21: Based on pressure data, a moving average is calculated on the pressure values ​​within a fixed time window, and the resulting average pressure value is used as the pressure feature vector. S22: Based on acceleration data, peak acceleration data are statistically analyzed within a fixed time window, and the impact frequency is calculated as the acceleration feature vector. S23: Within a fixed time window, input the image data and point cloud information into a deep learning model based on an improved Mask R-CNN, extract the pixel size of each ore bounding box, and calculate the average stone area as the image feature vector. S24: Transfer the pressure feature vector Acceleration eigenvectors and image feature vectors The features are concatenated to form a fused feature vector. , S24: Based on fused feature vectors A sample dataset is constructed based on the corresponding working status, and the sample dataset is divided into a training set and a test set according to a set ratio. S3: Construct a fractured state identification model; The optimal hyperparameters (error penalty factor C and kernel function g) of the SVM state classification model are obtained by optimizing the Sparrow Search Algorithm (SSA); S33: The feature vector is fused. As input data, the working state is used as output data. Based on the optimal hyperparameters, the SVM state classification model is trained using the training set, and the performance of the trained SVM state classification model is evaluated using the test set. The SVM state classification model with the best performance is saved to the breakage state identification solver 20 as the breakage state identification model. S4: Autonomous crushing operation based on real-time crushing status recognition; S41: Data acquisition unit 19 acquires multi-source monitoring data through image point cloud acquisition unit 18, accelerometer 15, pressure sensor 12, and pressure sensor 13. Specifically, image point cloud acquisition unit 18 acquires image data and point cloud information of the ore during operation, accelerometer 15 acquires acceleration signal of chisel rod 17 during operation, and pressure sensor 12 and pressure sensor 13 acquire pressure signal 1 and pressure signal 2 in working oil circuit 6 and working oil circuit 7, respectively. After noise reduction processing, the data is sent to crushing state identification solver 20. S42: A fusion feature vector is extracted based on the multi-source monitoring data. As input data, the working state is classified and determined using the crushing state recognition model, and the classification result is output to the intelligent controller 21; S43: The intelligent controller 21 controls the crushing process by adopting a control strategy that matches the classification result, so that the intelligent crushing robot works in an autonomous crushing mode based on the crushing state recognition.

[0030] As a preferred option, in step S42 of step three, the classification results include normal crushing 1, normal crushing 2, dry drilling, stuck drill and crushed work area, as shown in Table 1; Table 1 shows that the working states are divided into normal crushing 1, normal crushing 2, dry drilling, stuck drill, and broken zone. When the crushing state is normal crushing - large rock, the SVM state classification model outputs 0; when the crushing state is normal crushing - small to medium rock, the SVM state classification model outputs 1; when the crushing state is dry drilling, the SVM state classification model outputs 2; when the crushing state is stuck drill, the SVM state classification model outputs 3; and when the crushing state is the broken zone, the SVM state classification model outputs 4. Table 1: Classification of Crushing Working Conditions As a preferred option, in step S43 of step three, the control strategy matching the classification result is as follows: When it is normal crushing 1 and normal crushing 2, the conventional incremental PID control method based on energy error feedback is used to control the crushing process (under this condition, the impact energy changes with the load, and the actuator works normally); when it is dry-running, the electro-hydraulic proportional directional valve 4 is de-energized and operates in the neutral position to stop the impact operation, and a reminder signal is sent to the intelligent crushing robot to adjust the working position of the breaker hammer body 16; when the drill is stuck, the pressure state of the chisel rod 17 is first judged by pressure sensor 12 and pressure sensor 23, then the electro-hydraulic proportional directional valve 4 is de-energized and operates in the neutral position to stop the impact operation. Then, when the pressure sensor 12 and pressure sensor 23 are stable, the electro-hydraulic proportional directional valve 4 is switched to the reverse impact position to release the chisel rod 17 by reversing the impact force; when it is a crushed work area, a reminder signal is sent to the intelligent crushing robot to move to the new work area. Under normal operating conditions, incremental PID control is used to ensure that the impact energy is dynamically adjusted according to the load, thereby improving crushing efficiency. In dry operation, the machine is stopped in time and a position adjustment reminder is given to avoid ineffective impacts and reduce energy consumption and equipment wear. In the case of stuck drill, the machine is stopped first and then reversed to quickly release the chisel rod, reducing the risk of equipment failure. The machine reminds users to move the work area after the crushed area has been crushed, avoiding repetitive work and improving overall work efficiency.

[0031] As a preferred option, in step S21 of step three, the first step is obtained according to formula (1). Average pressure over a time window According to formula (2), the first... Impact frequency within a time window According to formula (3), the first... Area of ​​each stone ; (1); In the formula, For window size, Relative to the current moment The A historical pressure value; (2); In the formula, For the first The number of peak values ​​within a window The length of the window; (3); In the formula, This represents the total number of stones detected in the image.

[0032] As a preferred option, in step three, S43, the control process of the conventional incremental PID control method based on energy error feedback is as follows: S43-1: First, identify the size of the ore volume based on real-time image data and point cloud information, and then calculate the ore mass based on the ore density. The unit is kg; then, the theoretical impact energy required for the hydraulic breaker is calculated according to formula (4). ; (4); In the formula, The piston speed in the hydraulic crusher cylinder 10 is expressed in m / s. S43-2: Based on theoretical impact energy using fuzzy algorithms The initial control current signal is mapped and output to the electro-hydraulic proportional directional valve 4. At the same time, based on the feedback pressure and flow data, the energy transmission loss rate is calculated according to formula (5). ; (5); In the formula, The time required for the piston to complete one impact, measured in seconds; The pressure data is based on pressure sensor 12, and the unit is Pa. The data is based on the flow rate sensor 14, and the unit is L / s; The impulse frequency is expressed in Hz. , In time The number of impacts measured internally; S43-3: Calculate the actual impact energy required according to formula (6). ; (6); S43-4: Calculate the first according to formula (7) Error at each sampling time; (7); S43-5: Using the incremental PID control algorithm, the control increment is calculated according to formula (8). ; (8); In the formula, Sampling period; The sampling sequence number; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; S43-6: Obtain the control output at the current moment according to formula (9). ; (9); S43-7: Obtain the proportional valve current signal at the current moment according to formula (10). ; (10); In the formula, , These are the minimum and maximum allowable currents for the electro-hydraulic proportional directional valve 4, respectively, in mA. Limit the PID output; S43-8: Output proportional valve current signal This leads to the electro-hydraulic proportional directional valve 4, enabling precise control of the valve opening within the valve 4, while simultaneously performing adaptive control adjustments. The adaptive control adjustment process is as follows: If... Then increase the proportional valve current signal By increasing the valve opening to increase the energy of the next impact, if This reduces the proportional valve current signal. The energy of the next impact is reduced by decreasing the valve opening. At the same time, the ore damage is fed back in real time by combining image data and point cloud information to ensure that the ore crushing degree meets the requirements, and the feedback on ore damage is used as an auxiliary basis for subsequent control and adjustment.

[0033] As a preferred option, in step S1 of step three, the multi-source monitoring data is denoised using discrete wavelet transform according to formula (11); (11); In the formula, These are discrete wavelet coefficients; The wavelet transform scaling parameter; These are translation parameters; The original input signal; It is a wavelet function.

[0034] As a preferred option, in step S3 of step three, the process of optimizing the hyperparameters of the SVM state classification model based on the sparrow search algorithm is as follows: S31: Initialize SSA parameters; set population size n, maximum number of iterations G, number of producers PD, number of scouts SD, and safety threshold. , ; S32: Initial population position generation; randomly generate the initial positions of N sparrow individuals. The position of each sparrow is represented by a two-dimensional vector (C, g), where C is the error penalty factor and g is the kernel function parameter, that is, each sparrow corresponds to a set of parameters of the SVM state classification model. S33: Iterative optimization and position update; S33-1: Fitness Calculation and Extreme Value Screening; Calculate the fitness values ​​of all sparrows, screen out the current globally best and worst individuals, and randomly generate alarm values. , ; S33-2: Update producer position ((i=1:PD)); Update producer position according to formula (12) That is, when At that time, the producer enters wide-area search mode, and its location is updated to... ,when At that time, the producers led the herd to a safe area; the location was updated to... ; (12); In the formula, This represents the number of iterations. For the first Only sparrows in the first In the nth iteration The value of the dimension; This represents the maximum number of iterations. These are random numbers that follow a normal distribution. For elements all equal to 1 matrix; ; It is a random number. S33-3: Forager position update ((i=n:PD+1)); Update the forager's position according to formula (13). That is, when At this time, foragers with poor adaptability may fly to other places to forage, and the location of foragers with poor adaptability will be updated to... In other cases, the location of foragers with better adaptability is updated to... ; (13); In the formula, The optimal position occupied by the producer; This is the worst position globally at present; For elements that are all 1 or -1 matrix; ; S33-4: Scout position update ((i=n:PD+1)); Update the scout's position according to formula (14). That is, when At that time, sparrows located on the edge of the sparrow flock moved towards the center, and their positions were updated to... ;when At that time, the sparrow located in the center of the sparrow flock randomly moves closer to other sparrows, and its position is updated. ; (14); In the formula, This is the current globally optimal position; These are normally distributed random numbers with a mean of 0 and a variance of 1. It is a random number. ; This represents the current fitness value of the sparrow. , These are the current best and worst fitness values, respectively. The smallest constant to avoid division by zero errors; S33-5: Position update judgment; compare the fitness value of each sparrow's new position with its original position. If the new position has a better fitness value, then update to the new position; otherwise, retain the original position. S34: Iteration Termination and Result Output; Determine if the current iteration count has reached the preset maximum iteration count G. If it has, stop the iteration and output the global optimal position. and global optimal fitness If the target is not reached, return to S33 for the next iteration.

[0035] This invention provides a control method for an intelligent crushing robot's autonomous electro-hydraulic control system. First, the operating mode includes either manual remote control or autonomous crushing control, supporting switching between the two modes. This retains the operator's right to intervene manually in complex and special working conditions while enabling autonomous operation under normal conditions, thus improving the flexibility of crushing operations. Next, in manual remote control mode, the operator can directly send commands via a control handle in the cab, conforming to traditional engineering machinery operating habits. Simultaneously, the intelligent controller directly drives the electro-hydraulic proportional directional valve based on the operator's manual control, precisely controlling the start / stop and impact intensity of the breaker hammer to meet personalized operational needs. Second, in autonomous crushing control mode, by collecting three key data types—image point cloud data, acceleration signals, and hydraulic pressure signals—the system can reflect the ore crushing morphology, chisel impact characteristics, and hydraulic system operating conditions, ensuring the integrity of the basic data dimensions and providing sufficient basis for accurate subsequent status identification. Furthermore, by denoising the monitoring data, invalid information such as sensor noise and environmental interference can be effectively filtered out, improving the quality of the original data and reducing errors in subsequent model training. Subsequently, differentiated extraction methods were designed for different types of data. For pressure data, the moving average method was used to effectively highlight the steady-state characteristics of pressure. For acceleration data, peak values ​​and impact frequencies were statistically analyzed to effectively reflect impact intensity and frequency. For image point cloud data, the size and area of ​​ore were extracted by improving Mask R-CNN, which can fully characterize the crushing effect, making the features highly correlated with the crushing state. The three types of features—pressure, acceleration, and image—were concatenated into a fusion feature vector F to achieve multi-dimensional information complementarity, avoid the limitations of single features, and improve the accuracy of subsequent model classification. Furthermore, the Sparrow Search Algorithm (SSA) was used to optimize the error penalty factor C and kernel function g of the SVM. Compared with traditional methods such as grid search, it has the characteristics of fast convergence speed and strong optimization ability, and can quickly find the optimal hyperparameter combination. The crushing state recognition model constructed using the SVM model has the advantages of simple structure and low computational requirements. At the same time, combined with the hyperparameters optimized by SSA, it can run quickly in the intelligent controller to meet the needs of real-time state recognition. Then, multi-source data is collected in real time and fused features are extracted. The fragmentation state is quickly determined through a pre-trained model, realizing a closed-loop control of "collection-identification-decision" and ensuring real-time response. Finally, dedicated control strategies are formulated for different working states to accurately match the requirements of the working conditions.

[0036] This method is simple to implement, low in cost, and highly intelligent. By fusing three types of data—image data, acceleration, and pressure—and combining feature engineering with the SSA-SVM model, it achieves high-precision identification of the crushing state and dynamically adjusts the control strategy for different working conditions. This enables adaptive optimization of the crushing process, significantly improving crushing efficiency, reducing energy consumption and equipment wear, and ensuring the safety and stability of crushing operations. Furthermore, this method can precisely adjust the impact frequency of the crushing robot and the pressure of the hydraulic system, solving the problem of insufficient impact kinetic energy preventing normal operation. This allows the intelligent crushing robot to autonomously perform crushing operations based on the size of the ore, enhancing its autonomous operation capabilities.

Claims

1. An electro-hydraulic control system for autonomous crushing of an intelligent crushing robot, comprising a crushing hammer body (16), wherein the crushing hammer body (16) includes a crushing cylinder (10) and a chisel rod (17); characterized in that, It also includes a hydraulic control unit and an electronic control unit; The hydraulic control unit includes a high-pressure oil supply mechanism, an electro-hydraulic proportional directional valve (4), an overflow valve (8), a filter assembly A, an accumulator (11), and a heat dissipation assembly; the working ports A and B of the electro-hydraulic proportional directional valve (4) are connected to the working ports A and B of the hydraulic breaker body (16) through working oil circuit one (6) and working oil circuit two (7), respectively; the oil inlet P of the electro-hydraulic proportional directional valve (4) is connected to the high-pressure oil supply mechanism, and its return oil port T is connected to the oil tank (3) through the main return oil circuit (22); the oil inlet and outlet of the overflow valve (8) are connected to working oil circuit one (6) and working oil circuit two (7), respectively. 7); The working oil ports of the filter assembly A and the accumulator (11) are connected in series in the working oil circuit two (7); the filter assembly A includes a filter A (29) and a one-way valve A (30) connected in parallel; the heat dissipation assembly is connected in series on the main return oil circuit (22); the heat dissipation assembly includes a back pressure valve one (23), a radiator (24) and a back pressure valve two (26), the back pressure valve one (23) and the radiator (24) are connected in series to form a heat dissipation branch oil circuit, the back pressure valve two (26) is connected in parallel with the heat dissipation branch oil circuit, and the set pressure of the back pressure valve two (26) is higher than the set pressure of the back pressure valve one (23); The electronic control unit includes a pressure sensor (12), a flow sensor (14), a pressure sensor (13), an acceleration sensor (15), an image point cloud collector (18), a temperature sensor (31), a data collector (19), a breakage state identification solver (20), an intelligent controller (21), and a control handle (28). The pressure sensor (12) and the flow sensor (14) are both connected to the working oil circuit (6). The pressure sensor (13) is connected to the working oil circuit (7). The acceleration sensor (15) is embedded in the chisel rod (17). The image point cloud collector (18) is mounted on one side of the outside of the breaker body (16) and is used to collect data on the operation of the breaker body (16). Dynamic image data and point cloud information of ore; the temperature sensor (31) is connected to the heat dissipation branch; the data acquisition unit (19) is connected to pressure sensor one (12), flow sensor (14), pressure sensor two (13), acceleration sensor (15), temperature sensor (31) and image point cloud acquisition unit (18) respectively; the crushing state identification solver (20) is connected to the data acquisition unit (19) and is used to classify the crushing working state; the intelligent controller (21) is connected to the crushing state identification solver (20), electro-hydraulic proportional directional valve (4) and radiator (24) respectively and is used to implement intelligent control strategies that match the working state; the control handle (28) is connected to the intelligent controller (21).

2. The electro-hydraulic control system for autonomous crushing of an intelligent crushing robot according to claim 1, characterized in that, The hydraulic control unit also includes a two-way hydraulic lock (5), and the two connecting oil circuits of the two-way hydraulic lock (5) are respectively connected in series in working oil circuit one (6) and working oil circuit two (7); the two-way hydraulic lock (5) includes a first hydraulic control check valve (31) and a second hydraulic control check valve (32), the hydraulic control port of the first hydraulic control check valve (31) is connected to the oil outlet of the second hydraulic control check valve (32), and the hydraulic control port of the second hydraulic control check valve (32) is connected to the oil outlet of the first hydraulic control check valve (31).

3. The electro-hydraulic control system for autonomous crushing of an intelligent crushing robot according to claim 1, characterized in that, The hydraulic control unit also includes a filter assembly B, which is connected in series in the main return oil circuit (22) and located downstream of the heat dissipation assembly. The filter assembly B includes a filter B (25) and a one-way valve B (27) connected in parallel.

4. The electro-hydraulic control system for autonomous crushing of an intelligent crushing robot according to claim 3, characterized in that, The image point cloud acquisition device (18) includes an explosion-proof camera and a lidar; the acquisition angle of the explosion-proof camera is 45 degrees to the horizontal plane.

5. A control method for an intelligent crushing robot's autonomous electro-hydraulic control system, employing the intelligent crushing robot's autonomous electro-hydraulic control system as described in any one of claims 1 to 4, characterized in that, Includes the following steps: Step 1: Selecting the working mode; The operator sends a manual remote control signal or an autonomous crushing signal through the control handle (28) in the cab. After the intelligent controller (21) receives the manual remote control signal, it executes step two. After receiving the autonomous crushing signal, it executes step three. Step 2: Manual remote control mode; The operator sends a control signal through the control handle (28) in the cab. The intelligent controller (21) controls the action of the electro-hydraulic proportional directional valve (4) according to the received control signal, so that the main body of the breaker (16) works in the manual remote control working state. Step 3: Autonomous Crushing Control Mode; S1: Data collection during autonomous crushing operations; Within a set time period, the intelligent controller (21) automatically controls the switching action of the electro-hydraulic proportional directional valve (4) according to the preset timing control logic, so that the main body of the breaker (16) can perform crushing operations autonomously. At the same time, the data acquisition unit (19) collects the image data and point cloud information of the ore during the operation through the image point cloud acquisition unit (18), collects the acceleration signal of the chisel rod (17) during the operation through the acceleration sensor (15), and collects the pressure signal one and pressure signal two in the working oil circuit one (6) and working oil circuit two (7) respectively through the pressure sensor one (12) and pressure sensor two (13). After the data acquisition unit (19) performs noise reduction processing on the multi-source monitoring data, it sends it to the crushing state identification solver (20). S2: Construct a classification dataset through feature extraction; S21: Based on pressure data, a moving average is calculated on the pressure values ​​within a fixed time window, and the resulting average pressure value is used as the pressure feature vector. S22: Based on acceleration data, peak acceleration data are statistically analyzed within a fixed time window, and the impact frequency is calculated as the acceleration feature vector. S23: Within a fixed time window, input the image data and point cloud information into a deep learning model based on an improved Mask R-CNN, extract the pixel size of each ore bounding box, and calculate the average stone area as the image feature vector. S24: Transfer the pressure feature vector acceleration eigenvectors and image feature vectors The features are concatenated to form a fused feature vector. , S24: Based on fused feature vectors A sample dataset is constructed based on the corresponding working status, and the sample dataset is divided into a training set and a test set according to a set ratio. S3: Construct a fractured state identification model; The hyperparameters of the SVM state classification model are optimized based on the sparrow search algorithm to obtain the optimal hyperparameters; S33: using fused feature vectors As input data, the working state is used as output data. Based on the optimal hyperparameters, the SVM state classification model is trained using the training set, and the performance of the trained SVM state classification model is evaluated using the test set. The SVM state classification model with the best performance is saved into the broken state identification solver (20) as the broken state identification model. S4: Autonomous crushing operation based on real-time crushing status recognition; S41: The data acquisition unit (19) acquires multi-source monitoring data through the image point cloud acquisition unit (18), the accelerometer (15), the pressure sensor one (12), and the pressure sensor two (13), and sends the data to the fracture state identification solver (20) after noise reduction processing; S42: The fusion feature vector is extracted based on the multi-source monitoring data. As input data, the working state is classified and determined using the crushing state identification model, and the classification result is output to the intelligent controller (21); S43: The intelligent controller (21) controls the crushing process according to the classification result and adopts a control strategy that matches the classification result, so that the intelligent crushing robot works in an autonomous crushing mode based on the crushing identification state.

6. The control method for an intelligent crushing robot's autonomous electro-hydraulic control system according to claim 5, characterized in that, In step S42 of step three, the classification results include normal crushing 1, normal crushing 2, dry crushing, stuck drill, and crushed work area; in step S43 of step three, the control strategy matching the classification results is as follows: when it is normal crushing 1 and normal crushing 2, the conventional incremental PID control method based on energy error feedback is used to control the crushing process; when it is dry crushing, the electro-hydraulic proportional directional valve (4) is de-energized and operates in the neutral position to stop the impact operation and send a reminder signal to the intelligent crushing robot to adjust the working position of the main body of the breaker hammer (16); when it is When the drill gets stuck, the pressure state of the chisel rod (17) is first determined by pressure sensor 1 (12) and pressure sensor 2 (13). Then, the electro-hydraulic proportional directional valve (4) is de-energized and operates in the neutral position to stop the impact operation. Then, when the pressure sensor 1 (12) and pressure sensor 2 (13) are stable, the electro-hydraulic proportional directional valve (4) is switched to the reverse impact position to release the chisel rod (17) by reversing the impact force. When it is a crushed work area, a reminder signal to move to the new work area is sent to the intelligent crushing robot.

7. The control method for an intelligent crushing robot's autonomous electro-hydraulic control system according to claim 6, characterized in that, In step S21 of step three, the first step is obtained according to formula (1). Average pressure over a time window According to formula (2), the first... Impact frequency within a time window According to formula (3), the first... Area of ​​each stone ; （1）； In the formula, For window size, Relative to the current moment The A historical pressure value; （2）； In the formula, For the first The number of peak values ​​within a window The length of the window; （3）； In the formula, This represents the total number of stones detected in the image.

8. The control method for an intelligent crushing robot's autonomous electro-hydraulic control system according to claim 7, characterized in that, In step S43 of step three, the control process of the conventional incremental PID control method based on energy error feedback is as follows: S43-1: First, identify the size of the ore volume based on real-time image data and point cloud information, and then calculate the ore mass based on the ore density. The unit is kg; then, the theoretical impact energy required for the hydraulic breaker is calculated according to formula (4). ; （4）； In the formula, The piston speed in the crushing cylinder (10) is expressed in m / s. S43-2: Based on theoretical impact energy using fuzzy algorithms The initial control current signal is mapped and output to the electro-hydraulic proportional directional valve (4). At the same time, based on the feedback pressure and flow data, the energy transmission loss rate is calculated according to formula (5). ; （5）； In the formula, The time required for the piston to complete one impact, measured in seconds; The pressure data is based on pressure sensor 1 (12), and the unit is Pa; The flow data is obtained based on the flow sensor (14), and the unit is L / s; The impulse frequency is expressed in Hz. , In time The number of impacts measured internally; S43-3: Calculate the actual impact energy required according to formula (6). ; （6）； S43-4: Calculate the first according to formula (7) Error at each sampling time; （7）； S43-5: Using the incremental PID control algorithm, the control increment is calculated according to formula (8). ; （8）； In the formula, Sampling period; The sampling sequence number; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; S43-6: Obtain the control output at the current moment according to formula (9). ; （9）； S43-7: Obtain the proportional valve current signal at the current moment according to formula (10). ; （10）； In the formula, , These are the minimum and maximum allowable currents for the electro-hydraulic proportional directional valve (4), respectively, in mA. Limit the PID output; S43-8: Output proportional valve current signal To the electro-hydraulic proportional directional valve (4), to achieve precise control of the valve port opening inside the electro-hydraulic proportional directional valve (4), and at the same time to perform adaptive control adjustment; The adaptive control adjustment process is as follows: If Then increase the proportional valve current signal By increasing the valve opening to increase the energy of the next impact, if This reduces the proportional valve current signal. The energy of the next impact is reduced by decreasing the valve opening. At the same time, the ore damage is fed back in real time by combining image data and point cloud information to ensure that the ore crushing degree meets the requirements, and the feedback on ore damage is used as an auxiliary basis for subsequent control and adjustment.

9. The control method of the intelligent crushing robot autonomous electro-hydraulic control system according to claim 8, characterized in that, In step S1 of step three, the multi-source monitoring data is denoised using discrete wavelet transform according to formula (11); （11）； In the formula, These are discrete wavelet coefficients; The wavelet transform scaling parameter; These are translation parameters; The original input signal; It is a wavelet function.

10. The control method of the intelligent crushing robot's autonomous electro-hydraulic control system according to claim 9, characterized in that, In step S3 of step 3, the process of optimizing the hyperparameters of the SVM state classification model based on the sparrow search algorithm is as follows: S31: Initialize SSA parameters; set population size n, maximum number of iterations G, number of producers PD, number of scouts SD, and safety threshold. , ; S32: Initial population position generation; randomly generate the initial positions of N sparrow individuals. The position of each sparrow is represented by a two-dimensional vector (C, g), where C is the error penalty factor and g is the kernel function parameter, that is, each sparrow corresponds to a set of parameters of the SVM state classification model. S33: Iterative optimization and position update; S33-1: Fitness Calculation and Extreme Value Screening; Calculate the fitness values ​​of all sparrows, screen out the current globally best and worst individuals, and randomly generate alarm values. , ; S33-2: Update producer position; Update producer position according to formula (12) ; （12）； In the formula, This represents the number of iterations. For the first Only sparrows in the first In the nth iteration The value of the dimension; This represents the maximum number of iterations. These are random numbers that follow a normal distribution. For elements all equal to 1 matrix; ; It is a random number. S33-3: Forager position update; update the forager's position according to formula (13). ; （13）； In the formula, The optimal position occupied by the producer; This is the worst position globally at present; For elements that are all 1 or -1 matrix; ; S33-4: Scout position update; update the scout's position according to formula (14). ; （14）； In the formula, This is the current globally optimal position; These are normally distributed random numbers with a mean of 0 and a variance of 1. It is a random number. ; This represents the current fitness value of the sparrow. , These are the current best and worst fitness values, respectively. The smallest constant to avoid division by zero errors; S33-5: Position update judgment; compare the fitness value of each sparrow's new position with its original position. If the new position has a better fitness value, then update it to the new position. Otherwise, retain the original position; S34: Iteration Termination and Result Output; Determine if the current iteration count has reached the preset maximum iteration count G. If it has, stop the iteration and output the global optimal position. and global optimal fitness If the target is not reached, return to S33 for the next iteration.

Images