Industrial robot multi-type test equipment adaptive collaborative control method

Abstract

The present application relates to the technical field of robot control, and discloses an adaptive collaborative control method for industrial robots and multiple types of test equipment, comprising: obtaining microphysical property feedforward parameters to generate a global safety dynamic envelope sphere by executing boundary inflation on an end effector; mapping interference boundary data of the test equipment into a dynamic geometric envelope body, and establishing an environmental thermal radiation scalar field; combining end feedback sensor data to calculate target clamping force; calculating an effective Euclidean approach distance according to the envelope sphere and the envelope body to generate an avoidance acceleration threshold to limit underlying servo drive for a loading action; when abnormal conditions exist, a composite cost function is constructed according to the aforementioned parameters to solve and control asynchronous evacuation actions; when there is no abnormality, preset working condition mapping logic is called to solve and output control instructions. The present application prevents equipment collision and material splashing by limiting servo output, and ensures safe evacuation of the equipment in case of failure by planning a path away from the heat source using a composite cost function.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, specifically to an adaptive collaborative control method for various types of industrial robot testing equipment. Background Technology

[0002] In the processing operations of various types of testing equipment, industrial robots need to cooperate with testing instruments to perform material handling and loading. Existing robot control systems use fixed motion trajectories and static clamping parameters. When faced with powder materials with different bulk densities and angles of repose, executing a uniform acceleration and deceleration control curve will cause the powder materials to slip and splash during the handling process.

[0003] Laboratory equipment includes moving mechanisms and high-temperature heating components. Existing control systems lack calculations for the dynamic changes of interference boundaries and the distribution of environmental heat radiation. When the robot approaches the laboratory equipment, it will physically interfere with and collide with the moving parts in the open state, and the end effector will be directly damaged by high-temperature radiation.

[0004] When encountering abnormal operating conditions such as communication interruption or equipment alarm, the conventional safety procedure directly cuts off the servo power, causing the robotic arm to stop at the fault coordinate position. The sudden loss of driving force of the robotic arm will cause the clamped material to overturn. The actuators that remain in the high-temperature area will suffer hardware damage due to continuous heating. The existing fixed evacuation logic lacks the ability to optimize the judgment of spatial interference obstacles and temperature gradients. When executing the forced retreat evacuation action, it will cause the robotic arm to collide with the surrounding equipment, causing secondary collisions and interrupting the automated testing process. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an adaptive collaborative control method for industrial robots with multiple types of testing equipment. This method solves the problems of material splashing and scattering, physical interference and collision caused by the use of fixed control strategies in existing control systems for multiple types of testing operations, as well as equipment damage and secondary collisions caused by rigid evacuation logic under abnormal working conditions.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] An adaptive collaborative control method for multiple types of industrial robot testing equipment, the method comprising the following steps:

[0008] Obtain the feedforward parameters of microscopic properties, and generate a global safety dynamic envelope sphere by performing boundary expansion on the end effector;

[0009] Read the interference boundary data and operating temperature data of the testing equipment, map the interference boundary data into a dynamic geometric envelope, and establish an environmental thermal radiation scalar field based on the operating temperature data;

[0010] Acquire end-feedback sensor data and calculate target clamping force by combining the microscopic property feedforward parameters; calculate effective Euclidean approach distance based on the global safety dynamic envelope sphere and the dynamic geometric envelope, generate avoidance acceleration threshold by combining the microscopic property feedforward parameters, and use the target clamping force and the avoidance acceleration threshold to perform low-level servo drive restrictions on the loading action;

[0011] When an abnormal working condition is determined, a soft interrupt signal is generated to suspend the transport trajectory. A composite cost function is constructed and solved based on the effective Euclidean approach distance, the microscopic material property feedforward parameters and the environmental thermal radiation scalar field. The retreat path parameters are then generated to control the execution of asynchronous evacuation actions.

[0012] When the abnormal operating conditions are not triggered, the preset operating condition mapping logic is invoked using the microscopic property feedforward parameters to perform calculation operations, generate and output control commands to drive the testing operation.

[0013] The innovative principle of this invention lies in its fusion of front-end microscopic physical property parameters with terminal environmental geometric interference data and temperature field data to construct a multi-dimensional environmental model. At the control execution level, the physical spatial proximity distance and physical property parameters are used to jointly constrain the dynamic output of the underlying servo driver. When an anomaly occurs, the optimal safe descent direction is found through a composite cost function involving both temperature gradient and spatial distance, enabling the robot to autonomously avoid obstacles and heat and evacuate. The control mode established by this invention achieves adaptive coordination across unit working conditions.

[0014] Furthermore, the step of obtaining microscopic property feedforward parameters and performing boundary expansion on the end effector to generate a global safety dynamic envelope sphere includes:

[0015] The nominal bulk density, dynamic powder angle of repose, and dry ash-free volatile matter parameters of the coal sample were extracted as feedforward parameters for the microscopic properties.

[0016] Extract the tool center point of the end effector as the spatial geometric sphere center, and extract the maximum physical contour size and superimpose the anti-collision reserve to generate the safety envelope radius;

[0017] A three-dimensional boundary expansion process is performed in the global coordinate system using the spatial geometric sphere center and the safe envelope radius to generate the global safe dynamic envelope sphere of the spherical interference boundary.

[0018] This step establishes the safety defense boundary range of the end effector in three-dimensional space by extracting specific physical property parameters and equipment geometric features.

[0019] Furthermore, the steps of reading the interference boundary data and operating temperature data of the testing equipment, mapping the interference boundary data into a dynamic geometric envelope, and establishing an environmental thermal radiation scalar field based on the operating temperature data include:

[0020] Based on the physical characteristics of the actuator of the laboratory equipment, the moving parts with linear displacement and rotational opening and closing characteristics are dimensionally reduced and mapped into dynamic capsule bodies and dynamic fan-shaped swept bodies to form the dynamic geometric envelope.

[0021] Extract the geometric center coordinates of the opening surface of the heating component, and based on the geometric center coordinates and the operating temperature data, construct the environmental thermal radiation scalar field with spatial gradient distribution characteristics using the natural base decay law.

[0022] This step performs geometric feature dimensionality reduction on equipment with specific shapes to reduce the computational power consumption of spatial interference calculations and quantifies the radiation distribution characteristics of high-temperature equipment.

[0023] Furthermore, the step of acquiring end-feedback sensor data and calculating the target clamping force in conjunction with the microscopic property feedforward parameters includes:

[0024] The end-feedback sensor data includes contact pressure and spatial acceleration values; the equivalent transport mass is calculated using the nominal bulk density of the coal sample in the microscopic property feedforward parameters, and the dynamic gain term is obtained by performing inertial product compensation calculation in combination with the spatial acceleration value;

[0025] The dynamic gain term is linearly superimposed with the basic holding force, and the target holding force is obtained by performing closed-loop error correction based on the contact pressure value.

[0026] This step avoids the risk of slippage of materials with different densities during acceleration and deceleration by introducing dynamic inertia compensation and sensor closed-loop feedback.

[0027] Furthermore, the steps of calculating the effective Euclidean approach distance based on the global safety dynamic envelope sphere and the dynamic geometric envelope, generating an avoidance acceleration threshold by combining the microscopic material property feedforward parameters, and using the target clamping force and the avoidance acceleration threshold to implement low-level servo drive restrictions on the loading action include:

[0028] The shortest spatial distance from the center of the global safety dynamic envelope sphere to the boundary of the dynamic geometric envelope is calculated as the effective Euclidean approach distance;

[0029] A nonlinear decay model is established that correlates the effective Euclidean approach distance with the dynamic powder repose angle in the microscopic property feedforward parameters, and the upper limit of the limit maneuver is generated as the avoidance acceleration threshold.

[0030] The target clamping force is converted into a driving command, and the avoidance acceleration threshold is converted into the allowable angular acceleration limit corresponding to each joint axis to constrain the driver output.

[0031] This step translates the risk of spatial collisions into an angular acceleration limiting output at the bottom of the actuator, ensuring safe approach while preventing powder material from splashing due to excessive rotation of the joint shaft.

[0032] Furthermore, the step of generating a soft interrupt signal to suspend the transport trajectory when determining the existence of abnormal operating conditions includes:

[0033] The system detects communication heartbeat, equipment fault status, temperature of heating components, and location feedback parameters. If any abnormal state persists for a preset confirmation time, it determines that the abnormal operating condition exists and generates the soft interrupt signal.

[0034] The transport trajectory is suspended using the soft interrupt signal, the pitch and roll angles in the wrist attitude matrix of the end effector are frozen while the translational motion degrees of freedom are preserved, and a fault-based three-dimensional coordinate system is established based on the absolute position data of the joint encoder.

[0035] This step ensures that the robot's posture is locked in space the instant an anomaly occurs, preventing material from tipping over and providing an evacuation starting position.

[0036] Furthermore, the steps of constructing and solving a composite cost function based on the effective Euclidean approach distance, the microscopic property feedforward parameters, and the environmental thermal radiation scalar field, and generating retreat path parameters to control the execution of asynchronous evacuation actions, include:

[0037] The composite cost function is constructed by introducing a repulsive potential field penalty term based on the effective Euclidean proximity distance and a penalty term based on the temperature gradient of the ambient thermal radiation scalar field.

[0038] Starting from the three-dimensional coordinates of the fault dwelling, the displacement vector is calculated to generate candidate search directions. After collision and safety verification, the direction with the largest decrease in the composite cost function is selected to generate the retreat path parameters.

[0039] When performing the asynchronous evacuation action, the spatial acceleration command corresponding to the retreat path parameters is envelope-limited using the avoidance acceleration threshold.

[0040] This step ensures that the planned evacuation path avoids high-temperature heat sources and physical obstacles by calculating the gradient optimization of multi-objective penalty factors.

[0041] Furthermore, before the step of using the microscopic property feedforward parameters to call the preset working condition mapping logic to perform the calculation, the method further includes the step of establishing the preset working condition mapping logic, specifically including:

[0042] A historical mapping dataset is constructed by collecting the feedforward parameters of microscopic physical properties from historical test batches and the corresponding optimal operational feedback values.

[0043] The historical mapping dataset is input into a single-layer fully connected mapping neural network, and the connection weight parameters and intrinsic bias parameters of the neurons are iteratively updated using the backpropagation algorithm until the training converges.

[0044] Extract the fixed connection weight parameters and the inherent bias parameters to construct the preset working condition mapping logic.

[0045] This step establishes a data-driven model training process to achieve numerical correlation between physical parameters and process instructions.

[0046] Furthermore, the steps of using the microscopic property feedforward parameters to call the preset working condition mapping logic to perform the calculation, and generating and outputting control commands to drive the testing operation include:

[0047] Based on the microscopic property feedforward parameters, the initial command value is calculated by performing the solution operation through matrix multiplication and bias compensation algorithm using the connection weight parameters and inherent bias parameters in the preset working condition mapping logic.

[0048] The initial command value is compared item by item with a preset set of physical limit safety thresholds. When the value is determined to exceed the limit, the limiting function is called to forcibly constrain the exceeding index within the safe working range, and the control command is generated.

[0049] This step ensures that the computational results output by the neural network model are subject to the mandatory constraints of physical boundaries, thereby improving the determinism of the control system.

[0050] Furthermore, the step of generating and outputting control commands to drive the testing operation further includes:

[0051] The control command after the amplitude limiting process is sent to the control register of the corresponding test unit to drive each test unit to perform the test operation;

[0052] During the operation of each testing unit, the built-in sensor performs closed-loop tracking of the target control quantity, and combines the real-time feedback of interference boundary data and operating temperature data to connect the front-end physical property data and the terminal execution status to form a control data closed loop of the control system.

[0053] This step establishes data connectivity between the industrial robot and the testing equipment, ensuring the stable execution of control commands.

[0054] This invention provides an adaptive cooperative control method for multiple types of laboratory equipment using industrial robots. It offers the following advantages:

[0055] 1. This invention generates a global safety dynamic envelope sphere by obtaining microscopic material property feedforward parameters to expand the execution boundary of the end effector. Combined with the dynamic geometric envelope of the testing equipment, the effective Euclidean approach distance is calculated, and then an avoidance acceleration threshold is generated to limit the underlying servo drive of the filling action. By limiting the joint axis angular acceleration output of the underlying layer, the robot can automatically adjust its actions according to the physical space distance and material characteristics when approaching the testing equipment, preventing physical collisions between the equipment and avoiding the splashing of powder materials, thereby improving the safety level of the filling operation.

[0056] 2. This invention suspends the transport trajectory by generating a soft interrupt signal when abnormal operating conditions are detected, freezes the pitch and roll angles in the wrist attitude matrix of the end effector while retaining the translational motion degrees of freedom, and generates retreat path parameters by constructing a composite cost function based on the effective Euclidean approach distance, microscopic material property feedforward parameters, and environmental thermal radiation scalar field. The frozen attitude matrix locks the spatial attitude of the end effector at the moment of abnormality to prevent material overturning. At the same time, the retreat path away from the high-temperature heat source and physical interference obstacles is calculated by solving the composite cost function, ensuring the safe evacuation of equipment when the system faces a fault.

[0057] 3. This invention utilizes microscopic property feedforward parameters to call preset working condition mapping logic to perform calculations. It compares the initial command value with a preset set of physical limit safety thresholds item by item. When the limit is exceeded, it calls a limiting function to forcibly constrain the control command to the safe working range. Combined with closed-loop tracking to connect front-end data and terminal status, it performs the process of numerical comparison and calling the limiting function. This enables the control system to directly match the operation command of the testing unit based on the physical and chemical indicators of the coal sample, eliminates the risk of execution exceeding the limit caused by model output divergence, and improves the stability of multi-type equipment collaborative operation. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the system timing interaction of the present invention;

[0059] Figure 2 This is a schematic diagram of the method flow of the present invention;

[0060] Figure 3 This is a schematic diagram of the pre-data baseline establishment and initialization process of the present invention;

[0061] Figure 4 This is a schematic diagram of the physical workspace digital modeling process of the present invention;

[0062] Figure 5 This is a schematic diagram of the simulation curve for calculating the environmental thermal radiation scalar field of the present invention;

[0063] Figure 6 This is a schematic diagram of the underlying servo drive limitation process for the cross-device loading action of the present invention;

[0064] Figure 7 This is a schematic diagram of the constraint model curve of the correlation Euclidean distance and the maximum allowable acceleration according to the present invention;

[0065] Figure 8 This is a schematic diagram of the adaptive solution process for abnormal evacuation and testing instructions in this invention. Detailed Implementation

[0066] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] See attached document Figure 1 This invention provides an adaptive collaborative control system for multiple types of laboratory equipment for industrial robots. The system includes an interaction module, a mapping module, a constraint module, a monitoring module, and a control module.

[0068] The interaction module communicates with an external coal quality pretreatment information system to obtain the microscopic property feedforward parameters of the coal sample to be processed. The interaction module performs boundary expansion processing on the end effector of the industrial robot in the global Cartesian coordinate system to generate a global safety dynamic envelope sphere. The interaction module synchronously outputs the microscopic property feedforward parameters and the global safety dynamic envelope sphere to the associated modules inside the system.

[0069] The mapping module is connected to the programmable logic controller of the testing equipment to read the dynamic spatial interference boundary data and operating temperature data of the testing equipment. The mapping module reduces the dynamic spatial interference boundary data to a dynamic geometric envelope that changes over time. At the same time, the mapping module generates an environmental thermal radiation scalar field with spatial gradient distribution characteristics in the global Cartesian coordinate system based on the operating temperature data. The mapping module outputs the dynamic geometric envelope and the environmental thermal radiation scalar field to the constraint module and the monitoring module, respectively.

[0070] The constraint module is connected to the interaction module and the mapping module respectively. The constraint module acquires the end-effector feedback sensor data of the industrial robot and calculates the target clamping force by combining the microscopic property feedforward parameters. The constraint module calculates the effective Euclidean approach distance based on the center parameters of the global safety dynamic envelope sphere and the dynamic geometric envelope. The constraint module uses the effective Euclidean approach distance and the microscopic property feedforward parameters to generate the normal avoidance acceleration threshold. Then, the constraint module uses the target clamping force and the normal avoidance acceleration threshold to perform low-level drive constraints on the loading action of the industrial robot.

[0071] The monitoring module acquires system communication heartbeat packets and sensor threshold status. When a single testing equipment node is found to have abnormal operating conditions, the monitoring module outputs a soft interrupt signal to suspend the regular handling trajectory queue. The monitoring module locks the pitch and roll angles in the end-wrist attitude matrix of the industrial robot. The monitoring module constructs a composite cost function based on the effective Euclidean approach distance, microscopic material property feedforward parameters, and environmental thermal radiation scalar field. The monitoring module performs a minimization solution on the composite cost function to output the backoff path parameters.

[0072] The control module is connected to the interaction module to receive microscopic property feedforward parameters. The control module uses a single-layer fully connected mapping neural network model pre-trained based on historical datasets to map the underlying working conditions of each downstream testing unit according to the microscopic property feedforward parameters. Then, the control module calculates and outputs the target ignition energy of the calorimeter, the cleaning execution cycle and high-pressure water jet intensity of the oxygen bomb cleaning mechanism, the duty cycle output of the furnace temperature controller of the industrial analyzer, and the target clamping force of the hydrocarbon analysis sample packaging mechanism based on the microscopic property feedforward parameters.

[0073] See attached document Figure 2 This invention provides an adaptive collaborative control method for multiple types of industrial robot testing equipment, the method comprising the following steps:

[0074] S1: Obtain the microscopic property feedforward parameters of the coal sample to be processed sent by the external coal quality pretreatment information system. Perform three-dimensional boundary expansion processing on the end effector of the industrial robot in the established global Cartesian coordinate system to generate a global safety dynamic envelope sphere. Initialize and load the microscopic property feedforward parameters and the global safety dynamic envelope sphere into the system memory to establish the pre-data benchmark of the control system.

[0075] S2 polls the programmable logic controller of the testing equipment, reads the dynamic spatial interference boundary data and operating temperature data of the testing equipment, reduces the dynamic spatial interference boundary data to a dynamic geometric envelope that changes over time, extracts the geometric center coordinates of the opening surface of the heating component of the testing equipment, establishes an environmental thermal radiation scalar field in the global Cartesian coordinate system based on the geometric center coordinates and operating temperature data, and assigns spatial gradient distribution characteristics to the environmental thermal radiation scalar field to complete the digital modeling of the physical working space.

[0076] S3 receives end-effector feedback sensor data from the industrial robot, initializes the loaded microscopic property feedforward parameters using system memory, calculates the dynamic target clamping force using the end-effector feedback sensor data, calculates the effective Euclidean approach distance based on the center parameters of the global safety dynamic envelope sphere and the dynamic geometric envelope, substitutes the effective Euclidean approach distance and microscopic property feedforward parameters into the preset avoidance constraint equation to generate the normal avoidance acceleration threshold, and uses the dynamic target clamping force and the normal avoidance acceleration threshold to implement low-level servo drive restrictions on the cross-equipment loading action of the industrial robot to ensure the physical safety of the coal sample handling process.

[0077] S4, according to the set frequency, detects the communication heartbeat packets and sensor status of the system. Under the trigger logic of determining that a single laboratory equipment node has an abnormal working condition, a soft interrupt signal is generated. The soft interrupt signal is used to immediately suspend the regular handling trajectory queue of the industrial robot, freeze the pitch angle and roll angle in the attitude matrix of the end wrist of the industrial robot, extract the effective Euclidean approach distance, microscopic material property feedforward parameters and environmental thermal radiation scalar field parameters, use the extracted parameters to construct a composite cost function that correlates geometric collision avoidance and thermal characteristics, call the optimal solver to perform a minimization solution operation on the composite cost function, generate retreat path parameters and control the industrial robot to perform asynchronous retreat actions.

[0078] S5, under the condition that the cross-equipment loading operation has not triggered abnormal operating conditions and the end effector of the industrial robot has successfully released the coal sample to be processed and safely withdrawn from the dynamic spatial interference boundary of the testing equipment, the operating condition mapping logic of the downstream testing unit is established based on the microscopic property feedforward parameters. The operating condition mapping logic is implemented by an offline trained neural network model. Based on the microscopic property feedforward parameters, adaptive calculation is performed to output the target ignition energy of the calorimeter, the cleaning execution cycle and high-pressure water jet intensity of the oxygen bomb cleaning mechanism, the duty cycle output of the furnace temperature controller of the industrial analyzer, and the target clamping force of the hydrocarbon analysis sample packaging mechanism. The target ignition energy, cleaning execution cycle, high-pressure water jet intensity, duty cycle output, and target clamping force are sent to the control register of the corresponding testing unit to drive the corresponding testing unit to perform adaptive testing operations, connecting the upstream property data and the terminal execution unit to form a control data closed loop of the control system.

[0079] See attached document Figure 3 In specific implementations, step S1 provided by the present invention may include the following steps:

[0080] S11, the interaction module establishes a communication connection with the external coal quality pretreatment information system through the industrial Ethernet interface, and sends a data request message with a timestamp to the external coal quality pretreatment information system. Then, it obtains the data packet containing the physicochemical properties of the coal sample to be processed sent by the external coal quality pretreatment information system, and then uses the existing data parsing algorithm to decode the data packet, and finally extracts the microscopic physical property feedforward parameters.

[0081] The feedforward parameters of microscopic physical properties specifically include the nominal bulk density of the coal sample, the dynamic angle of repose of the powder under specific particle size conditions, and the dry ash-free volatile matter parameters. The specific particle size conditions are limited to the range of 0.2 mm to 3.0 mm. The nominal bulk density of the coal sample is calculated based on the coal sample mass collected by the weighing sensor and the coal sample volume parameters obtained by the visual measurement device. The dynamic angle of repose of the powder is calculated based on the coal sample packing profile collected by the visual measurement device. The dry ash-free volatile matter parameters are obtained from the detection results or prediction results already obtained by the upstream coal quality detection link through the external coal quality pretreatment information system.

[0082] The numerical range of the microscopic physical property feedforward parameters is controlled by the physicochemical property benchmark of the current batch of coal. Specifically, the nominal bulk density of the coal sample is limited to a range of 0.8 g / cm³. 3 Up to 1.2 g / cm 3 Between these values, the dynamic powder repose angle is limited to 28° to 45°, and the dry ash-free volatile matter parameter is limited to 10% to 40%.

[0083] S12, the control system establishes a global Cartesian coordinate system with the center of the physical base of the industrial robot as the origin. The interaction module reads the basic structural shape and size parameters of the end effector of the industrial robot. The end effector includes a mechanical gripper that performs the grasping action and a test crucible that carries the powdered coal sample. Then, based on the basic structural shape and size parameters, the tool center point of the end effector is extracted as the spatial geometric sphere center. The Euclidean distance value covering the maximum physical contour size of the mechanical gripper and the test crucible is extracted. A fixed anti-collision reserve margin is superimposed on the Euclidean distance value to generate a safety envelope radius. The anti-collision reserve margin is set according to the steady-state tracking error limit of the industrial robot servo motor.

[0084] Subsequently, the interactive module performs 3D boundary dilation processing on the end effector of the industrial robot in the established global Cartesian coordinate system using the spatial geometric sphere center and the safety envelope radius. Simultaneously, using the result of this 3D boundary dilation processing, the irregular polygonal physical mesh is replaced with a regular spherical interference boundary, thereby generating a global safety dynamic envelope sphere. The 3D spatial surface equation of the global safety dynamic envelope sphere satisfies the following mathematical relationship:

[0085] ;

[0086] In the formula, Let X be the X-axis coordinate of any given point in space in the world coordinate system; The Y-axis coordinate of any given point in space in the world coordinate system; Let Z be the Z-axis coordinate of any given point in space in the world coordinate system; Let X be the X-axis coordinate of the robot tool's center point in the world coordinate system; The Y-axis coordinate of the robot tool's center point in the world coordinate system; The Z-axis coordinate of the robot tool's center point in the world coordinate system; The radius of the safety envelope sphere at the center point of the robot tool.

[0087] S13, the interaction module allocates the extracted microscopic property feedforward parameters to the preset memory register address space inside the control system, and writes the three-dimensional spatial surface equation parameters of the global safety dynamic envelope sphere into the memory stack of the industrial robot's underlying motion control program, thereby completing the system memory initialization loading operation. Subsequently, a data mapping link is established between the data storage layer and the underlying physical execution layer, thereby establishing the front-end data reference of the control system. Finally, the underlying servo drive logic calls the front-end data reference to perform correlation calculations.

[0088] See attached document Figure 4 and attached Figure 5 In specific implementations, step S2 provided by the present invention may include the following steps:

[0089] S21, the mapping module polls the programmable logic controller of the testing equipment via the industrial fieldbus protocol, thereby reading the dynamic spatial interference boundary data and operating temperature data of the testing equipment. The dynamic spatial interference boundary data includes the position encoder feedback value of the testing equipment actuator, and the operating temperature data includes the real-time sampling value of the thermocouple built into the heating component. The mapping module adds an absolute timestamp to the acquired dynamic spatial interference boundary data and operating temperature data. At the same time, the mapping module pushes the data carrying the absolute timestamp into the real-time database synchronization queue of the control system. The control system extracts the data frame with the smallest time difference from the real-time database synchronization queue according to the robot interpolation cycle for subsequent calculation. The robot interpolation cycle parameter parameter configured in the control system is in the range of 2ms to 8ms. The mapping module uses the timestamp alignment mechanism to compensate for the timing deviation caused by the bus transmission delay, reducing the spatial calculation error caused by asynchronous data participating in the same robot interpolation cycle calculation.

[0090] S22, during the equipment debugging phase, the control system uses the industrial robot base coordinate system as the global Cartesian coordinate system and establishes the corresponding local coordinate system for each testing device. The control system uses preset calibration points to calculate the homogeneous transformation matrix of each device's local coordinate system to the global Cartesian coordinate system. At the same time, the specific number of preset calibration points is limited to between 3 and 6, and the homogeneous transformation matrix is ​​stored in the parameter storage area of ​​the control system. The mapping module uses the corresponding homogeneous transformation matrix to uniformly transform the coordinates of the moving parts generated by the position encoder feedback and the geometric feature coordinates in the three-dimensional computer-aided design model to the global Cartesian coordinate system.

[0091] The mapping module retrieves dynamic spatial interference boundary data from the real-time database and performs mathematical dimensionality reduction operations in conjunction with the 3D computer-aided design dimensional features of the laboratory equipment actuator. This reduces the dynamic spatial interference boundary data to a dynamic geometric envelope that changes over time. For moving parts with linear displacement characteristics, the dynamic geometric envelope is specifically represented as a dynamic capsule sliding along a straight trajectory in space. The mapping module uses the position encoder feedback value and the mechanical motion speed of the moving part to generate the real-time 3D coordinates of the two vertices at the two ends of the central axis of the dynamic capsule. Furthermore, the mapping module extracts the maximum circumcircle radius of the moving part as the expansion radius of the dynamic capsule. Finally, the geometric algebraic envelope equation of the dynamic capsule is constructed using the spatial line segments at the two vertices and the expansion radius.

[0092] S23, the dynamic geometric envelope is specifically manifested as a dynamic fan-shaped sweeping body for the moving parts of the testing equipment with rotational opening and closing characteristics. The mapping module extracts the physical hinge points of the rotating moving parts as the fixed rotation center coordinates of the dynamic fan-shaped sweeping body. The mapping module uses the real-time angle value fed back by the position encoder and the set angle prediction step size to generate the dynamic fan-shaped sweeping body. At the same time, the specific value of the set angle prediction step size is set between 1.5° and 4.5°. Then, the mapping module uses the dynamic capsule body and the dynamic fan-shaped sweeping body to replace the polygonal mesh. The mapping module parameterizes the dynamic spatial interference boundary of the heterogeneous testing equipment and sends it into the control system to participate in the interference correlation calculation.

[0093] S24, the mapping module retrieves the 3D computer-aided design model and extracts the geometric center coordinates of the opening surface of the heating component of the testing equipment. The opening surface of the heating component represents the main physical boundary for heat radiation and dissipation into the external physical working space. Based on the geometric center coordinates and operating temperature data, the mapping module establishes an environmental thermal radiation scalar field in the established global Cartesian coordinate system and assigns spatial gradient distribution characteristics to the environmental thermal radiation scalar field. The mapping module uses the natural base decay law to establish an internal thermodynamic mapping relationship, and then integrates the dynamic geometric envelope and the environmental thermal radiation scalar field to complete the digital modeling of the physical working space. The 3D spatial intensity calculation function of the environmental thermal radiation scalar field satisfies the following mathematical relationship:

[0094] ;

[0095] In the formula, Spatial coordinate points The calculated real-time temperature value at the location; The initial reference temperature of the environment is set as a fixed constant between 20°C and 25°C by default. This represents the highest core temperature at the center of the heat source. is the base of the natural logarithm; The attenuation coefficient of thermal radiation in the space medium is given, and the specific value range of the attenuation coefficient is limited to between 0.05 and 0.15 based on the field calibration data. Let X be the X-axis coordinate of any given point in space in the world coordinate system; The Y-axis coordinate of any given point in space in the world coordinate system; Let Z be the Z-axis coordinate of any given point in space in the world coordinate system; The X-axis coordinate of the heat source center point in the world coordinate system; The Y-axis coordinate of the heat source center point in the world coordinate system; The Z-axis coordinate of the heat source center point in the world coordinate system.

[0096] The control system pre-enters an empirical matching table of heat source attenuation constants for associated equipment models. This table is established through offline temperature measurement calibration during the equipment commissioning phase. For different laboratory equipment models, opening surface geometric areas, and typical operating temperatures, steady-state temperature data is collected at different spatial distances from the opening surface of the heating component. Based on the steady-state temperature data, the corresponding heat source attenuation constant is fitted. The control system establishes a correspondence between the equipment model, opening surface geometric area, and heat source attenuation constant and writes it into the empirical matching table. The mapping module calls the matched heat source attenuation constant based on the current laboratory equipment model and opening surface geometric area.

[0097] See attached document Figure 6 and attached Figure 7 In specific implementations, step S3 provided by the present invention may include the following steps:

[0098] S31, the constraint module receives end-effector feedback sensor data from the industrial robot, and the end-effector feedback sensor data includes the real-time contact pressure value fed back by the piezoelectric thin film sensor arranged at the fingertips of the end gripper and the instantaneous spatial acceleration value fed back by the inertial measurement unit arranged at the end flange. The constraint module extracts the nominal bulk density of the coal sample from the microscopic property feedforward parameters initialized and loaded in the system memory.

[0099] Simultaneously, the constraint module calculates the equivalent transport mass using the nominal bulk density of the coal sample combined with the fixed volume parameters of the test crucible. The constraint module sets the basic holding force as a fixed constant determined based on the self-weight of the test crucible, and the specific value of the basic holding force is limited to between 15N and 25N. Then, the constraint module performs inertial product compensation calculation using the equivalent transport mass and the instantaneous spatial acceleration value to obtain the dynamic gain term. The constraint module linearly superimposes the dynamic gain term with the basic holding force and performs closed-loop error correction based on the real-time contact pressure value. Thus, it calculates the dynamic target holding force using the microscopic property feedforward parameters initialized in the system memory and the end feedback sensor data.

[0100] S32, the constraint module extracts the center parameters of the global safety dynamic envelope sphere within the interpolation sampling period of a low-level controller. At the same time, the constraint module extracts the spatial algebraic feature points of the dynamic geometric envelope. Then, for the dynamic capsule, the constraint module calculates the shortest spatial distance from the center of the spatial geometric sphere to the central axis of the dynamic capsule, and subtracts the expansion radius of the dynamic capsule and the safety envelope radius of the global safety dynamic envelope sphere from the shortest spatial distance to obtain the corresponding effective Euclidean approach distance.

[0101] For a dynamic sector sweeping body, the constraint module calculates the shortest spatial distance from the center of the spatial geometric sphere to the outer boundary of the dynamic sector sweeping body, and subtracts the safety envelope radius of the global safety dynamic envelope sphere from the shortest spatial distance to obtain the corresponding effective Euclidean approach distance. The safety envelope radius already includes the maximum physical contour size of the end effector and the collision avoidance margin. Therefore, it is not necessary to repeatedly deduct the collision avoidance margin or the physical size of the end effector during the calculation of the effective Euclidean approach distance. The effective Euclidean approach distance characterizes the true physical safety margin between the end effector of the industrial robot and the dynamic interference boundary of the heterogeneous testing equipment.

[0102] S33, the constraint module extracts the dynamic powder repose angle from the microscopic property feedforward parameters and establishes a nonlinear attenuation mathematical model relating the effective Euclidean approach distance and the dynamic powder repose angle. The constraint module determines the nearest boundary point on the dynamic geometric envelope that has the smallest distance to the center of the global safe dynamic envelope sphere, and establishes a spatial unit vector pointing from the nearest boundary point to the center of the global safe dynamic envelope sphere. This spatial unit vector is determined as the normal avoidance direction within the current control cycle. Then, the constraint module substitutes the effective Euclidean approach distance and the microscopic property feedforward parameters into the preset avoidance constraint equation to generate a normal avoidance acceleration threshold. The normal avoidance acceleration threshold characterizes the upper limit of the end effector's extreme maneuver output in the normal spatial vector direction. Under the constraint of the normal avoidance acceleration threshold, the control system maintains that the peak inertial force does not exceed the frictional resistance of the coal sample's repose angle, and the preset avoidance constraint equation satisfies the following mathematical relationship:

[0103] ;

[0104] In the formula, This represents the maximum allowable acceleration under the current system conditions. The maximum acceleration that the system hardware or physical mechanism can achieve; is the base of the natural logarithm; The coefficient of friction of the contact surface or the sliding damping coefficient of the system is specified, and the specific value of the coefficient of friction or the sliding damping coefficient of the system is specified between 0.2 and 0.6. The feature distance of the current trajectory, and The value is equal to the effective Euclidean approach distance calculated by the constraint module; This is the operator for the tangent trigonometric function; The angle of repose or the slope angle of the terrain at the working contact surface.

[0105] S34, the constraint module converts the dynamic target clamping force into a drive current control command and sends it to the gripper drive controller of the end effector. The constraint module establishes the robot Jacobian matrix based on the current joint position of the industrial robot, and combines it with the current interpolation sampling period to convert the normal avoidance acceleration threshold of the end effector into the allowable angular acceleration limit corresponding to each joint axis. The constraint module writes the allowable angular acceleration limit corresponding to each joint axis into the limiting register of the corresponding servo driver, thereby limiting the maximum speed derivative output of each joint servo motor in the interpolation following period.

[0106] Thus, the constraint module uses dynamic target clamping force and normal avoidance acceleration threshold to implement low-level servo drive restrictions on the cross-equipment loading action of the industrial robot. During the cross-equipment loading operation, the constraint module maintains the morphological stability of the coal sample powder and the geometric safety of the mechanical trajectory. Finally, by integrating the microscopic powder mechanical properties and macroscopic spatial interference avoidance boundary constraint logic, the physical safety of the coal sample handling process is ensured.

[0107] See attached document Figure 8 In specific implementations, steps S4 and S5 provided by this invention may include the following steps:

[0108] S41, abnormal operating conditions include continuous communication heartbeat timeout, valid programmable logic controller fault status bit, operating temperature of heating component exceeding set safety limit, position feedback of actuator exceeding allowable position deviation, data of end feedback sensor exceeding corresponding safety threshold, or end feedback sensor failure. Specifically, the set safety limit of heating component is limited to between 800℃ and 1000℃, the allowable position deviation of actuator is limited to between 2mm and 5mm, and the corresponding safety threshold of end feedback sensor for contact pressure is limited to no more than 50N. The monitoring module generates a soft interrupt signal after any abnormal operating condition continues for a preset confirmation time, and the specific duration of the preset confirmation time is set to between 50ms and 150ms.

[0109] Then, the monitoring module immediately suspends the industrial robot's regular handling trajectory queue using a soft interrupt signal. During the equipment debugging phase, the control system completes the calibration of the industrial robot's tool coordinate system and records the pitch and roll angles corresponding to the test crucible being in a horizontal state as the horizontal attitude reference. After the soft interrupt is triggered, the monitoring module maintains the pitch and roll angles corresponding to the horizontal attitude reference, only allowing the yaw angle to be adjusted according to the retreat path direction, while retaining the motion degrees of freedom of the end effector along the three translational directions of the global Cartesian coordinate system.

[0110] Furthermore, the absolute position data of each joint encoder at the moment of triggering the interruption are extracted. Then, based on the absolute position data of each joint encoder and the existing robot forward kinematics solving algorithm, the fault-dwelling three-dimensional coordinates of the current end effector in the global Cartesian coordinate system are calculated, thereby establishing the initial state benchmark for fault recovery.

[0111] S42, the monitoring module reads the on-site environmental status data containing the three-dimensional coordinates of the fault location and starts an asynchronous calculation thread independent of the main loop control logic. The monitoring module then extracts the effective Euclidean approach distance, microscopic property feedforward parameters, and environmental thermal radiation scalar field parameters. Using the extracted parameters, the monitoring module constructs a composite cost function that correlates geometric collision avoidance and thermal characteristics. This composite cost function introduces a repulsive potential field based on spatial distance and a thermal radiation scalar value based on the temperature gradient. The composite cost function guides the industrial robot away from high-temperature areas and dynamic interference boundaries. The composite cost function satisfies the following mathematical relationship:

[0112] ;

[0113] In the formula, This represents the total cost function value for system trajectory planning or control optimization. The weighting coefficients for the position and yaw angle deviation penalty terms are set between 10 and 20. The actual position or attitude parameters of the system at the current moment; The pre-defined system safety position or reference attitude reference parameters; The weighting coefficient for the obstacle avoidance safety distance penalty item is set between 50 and 100. The Euclidean distance from the current position to the nearest obstacle; The weighting coefficient for the thermal radiation environment penalty term is set between 5 and 15. Spatial coordinate points The calculated real-time temperature value at the location.

[0114] S43, the monitoring module calls the optimal solver in the asynchronous calculation thread to perform a minimization operation on the composite cost function, and calculates the spatial displacement vector that makes the gradient of the composite cost function decrease the fastest, using the fault-dwelling three-dimensional coordinates as the search starting point. When it is detected that the gradient magnitude of the composite cost function is continuously lower than the set convergence threshold and the industrial robot has not yet reached the safe return point, the monitoring module generates multiple candidate search directions using Gaussian random perturbation parameters. The specific value of the convergence threshold is set to 0.001 to 0.005, and the standard deviation of the Gaussian random perturbation parameters is set to 0.05m to 0.15m. The Gaussian random perturbation parameters are only used for candidate search direction generation and do not directly form the motion execution command of the industrial robot.

[0115] The monitoring module performs dynamic geometric envelope collision verification and environmental thermal radiation scalar field safety threshold verification on the candidate search directions. It also eliminates candidate search directions whose effective Euclidean approach distance is lower than the preset minimum safety distance or whose environmental thermal radiation scalar value exceeds the preset safety threshold. Specifically, the preset minimum safety distance is limited to between 50mm and 100mm, and the preset safety threshold is set to between 65℃ and 80℃. Subsequently, the direction with the largest decrease in composite cost function is selected from the candidate search directions that have passed the safety verification as the new spatial displacement vector. Then, the spatial displacement vector is mapped to the joint angular velocity command queue of the industrial robot using the existing Jacobian inverse kinematics matrix.

[0116] During the generation of retreat path parameters, the monitoring module reads the normal avoidance acceleration threshold calculated in real time in step S33, and uses the normal avoidance acceleration threshold to perform envelope limiting on the Cartesian space acceleration command corresponding to the retreat path. After completing the envelope limiting, the monitoring module uses the inverse kinematic mapping relationship corresponding to the robot's Jacobian matrix to convert the limited Cartesian space motion command into the joint motion command of the industrial robot, thereby eliminating motion peaks that exceed the allowable acceleration range.

[0117] This operation ensures that during emergency asynchronous evacuation, the rapid retraction acceleration of the industrial robot remains controlled by the natural angle of repose of the coal sample to be processed, thereby preventing dangerous high-temperature coal powder from being scattered during the evacuation process and causing secondary safety hazards.

[0118] S44, the monitoring module sends the joint angular velocity command queue containing the retreat path parameters to the servo driver of each joint axis through the underlying private communication protocol. The monitoring module also uses the retreat path parameters to control the industrial robot to perform asynchronous evacuation actions. The monitoring module guides the industrial robot to move from the fault-dwelling three-dimensional coordinates to the safe return point according to the retreat path parameters. At the same time, it continuously monitors the changing trend of the environmental thermal radiation scalar field parameters during the execution of the asynchronous evacuation action.

[0119] When the effective Euclidean approach distance is lower than the preset minimum safe distance or the environmental thermal radiation scalar value exceeds the preset safety threshold, the monitoring module triggers a protective stop command and suspends the current joint movement. During the protective stop, the mapping module continuously updates the dynamic geometric envelope and the environmental thermal radiation scalar field parameters. The monitoring module calls the optimal solver again to generate new retreat path parameters based on the updated environmental parameters. After the new retreat path parameters pass the dynamic collision check, thermal radiation safety check, and joint movement range check, the monitoring module controls the industrial robot to continue to perform the controlled retreat action. Finally, the monitoring module maintains the physical integrity of the core hardware and test samples under abnormal working conditions.

[0120] S51, when the control module determines that the cross-equipment loading operation has not triggered abnormal operating conditions, it extracts the microscopic physical property feedforward parameters cached in the system. The microscopic physical property feedforward parameters include the nominal bulk density of the coal sample, the dynamic angle of repose of the powder, and the dry ash-free volatile matter parameters.

[0121] The control module establishes the operating condition mapping logic of the downstream testing unit based on the feedforward parameters of microscopic physical properties. Furthermore, the control module uses the mapping weight matrix to project the physical property characteristics of coal samples with multiple dimensions into multidimensional operating indicators of heterogeneous testing equipment. The control system establishes a basic mapping framework in the industrial field control network, thereby constructing a numerical correlation between the front-end coal sample status and the back-end equipment drive commands. Finally, the control module completes the initial loading operation of the status determination and mapping logic.

[0122] Before the control module completes the initial loading of state determination and mapping logic, the mapping weight matrix and static bias constant column vector are obtained by offline training of the operating condition mapping neural network model of the control system. The specific steps for creating and training the operating condition mapping neural network model include:

[0123] Microscopic physical property feedforward parameters from historical test batches are collected via fieldbus as input feature data. Specifically, the input feature data includes the nominal bulk density of the coal sample, the dynamic angle of repose of the powder, and the volatile matter parameters on a dry ash-free basis. Simultaneously, the control module retrieves the optimal operating feedback values, calibrated by human experts and the automated platform after multiple rounds of testing and optimization in the corresponding historical test batches, as target label data. Specifically, the target label data includes the target ignition energy, cleaning execution cycle, high-pressure water jet intensity, furnace temperature controller duty cycle output, and target clamping force. Then, the input feature data and target label data are paired and matched to construct a historical mapping dataset. The historical mapping dataset is then randomly divided into a training set and a validation set at a ratio of 80% to 20%.

[0124] A single-layer fully connected mapping neural network is constructed as the model carrier for the working condition mapping logic. To strictly correspond with the underlying mathematical mapping equations, the single-layer fully connected mapping neural network has only one input layer and one output layer and does not contain any hidden layers. The number of neurons in the input layer is consistent with the feature dimension of the microscopic material property feedforward parameters, and the number of neurons in the output layer is strictly consistent with the instruction dimension of the system control commands. In terms of connection configuration, a dense fully connected method is used to establish a data link between the input layer and the output layer, and the neurons in the output layer are uniformly configured with linear activation functions. The synaptic weight parameters of the input layer neurons connecting to the output layer neurons directly correspond to the mapping weight matrix required by the control system. At the same time, the inherent bias parameters of each neuron node in the output layer directly correspond to the static bias constant column vector required by the control system.

[0125] An adaptive moment estimation optimizer was selected as the optimization solver for model training. The initial learning rate of the single-layer fully connected mapping neural network was set to 0.005. The training batch size was set to 32 sets of feature data, and the maximum number of iterations in the training process was set to 2000. The mean squared error loss function was then configured as the loss function of the single-layer fully connected mapping neural network. The control system used the mean squared error loss function to quantify the numerical deviation between the forward prediction command and the optimal running feedback value. The tolerance step size of the model early stopping mechanism was set to 50 consecutive iterations.

[0126] The microscopic physical property feedforward parameter features from the training set are fed in batches into the input layer of a single-layer fully connected mapping neural network. The single-layer fully connected mapping neural network then calculates the forward-predicted control quantities of the testing equipment through the underlying fully connected logic. The loss gradient between the control quantities of the testing equipment and the target label data is calculated using the mean square error loss function. The synaptic connection weights and intrinsic bias parameters of the neurons are iteratively updated using the backpropagation algorithm combined with the adaptive moment estimation optimizer.

[0127] After each training iteration, the control system inputs the validation set into the single-layer fully connected mapping neural network to calculate the generalization error of the model. When the generalization error of the validation set does not show a decreasing trend for 50 consecutive iterations or the number of training iterations reaches the maximum of 2000, the control system forcibly terminates the model training process. After determining that the model training has reached the convergence state, the control system extracts the currently fixed connection weight parameters and bias parameters of the output layer, and assigns the fixed connection weight parameters to the mapping weight matrix of the adaptive solution operation layer. At the same time, the control system assigns the fixed bias parameters to the static bias constant column vector of the adaptive solution operation layer.

[0128] S52, the control module calls the operating condition mapping logic and performs adaptive calculation based on the microscopic property feedforward parameters. Simultaneously, the control module outputs the target ignition energy of the calorimeter, the cleaning cycle and high-pressure water jet intensity of the oxygen bomb cleaning mechanism, the duty cycle output of the furnace temperature controller of the industrial analyzer, and the target clamping force of the hydrocarbon analysis sample packaging mechanism through matrix multiplication and bias compensation algorithms. Thus, the control module converts the microscopic powder characteristics into digital signals recognizable by the testing terminal. Furthermore, the underlying mathematical model of the adaptive calculation satisfies the following linear spatial mapping relationship:

[0129] ;

[0130] In the formula, These are the low-level control commands or control signals that the system ultimately outputs. To map the planning parameters to the weight matrix or proportional gain coefficients of the control commands; The microscopic property feedforward parameter vector is the input to the system, and the microscopic property feedforward parameter vector specifically includes the nominal bulk density of the coal sample, the dynamic powder repose angle, and the dry ash-free volatile matter parameters; This is the inherent bias or steady-state error compensation constant of the control system.

[0131] S53, the control module extracts the set of physical limit safety thresholds specified in the hardware manual of the testing equipment. The set of physical limit safety thresholds specifically includes a heater temperature limit set to a maximum of 900°C, a high-pressure water jet pressure limit set to a maximum of 15MPa, and a mechanical clamping force peak value set to a maximum of 120N. The control module performs item-by-item comparison calculations between the values ​​in the system control command output column vector and the set of physical limit safety thresholds. When any value is found to exceed the limit, the system calls the built-in saturation limiting function, thereby forcibly constraining the over-limit indicators within the safe operating range. This eliminates the risk of equipment damage such as heating wire burnout or sample foil crushing due to abnormal calculation parameters before performing adaptive testing operations. The control module maintains that the terminal execution parameters output by the solution are always within the mechanical and electrical safety envelope of the downstream testing unit, ultimately establishing a hardware-level action protection barrier.

[0132] S54, the control module uses the fieldbus of the programmable logic controller to send the target ignition energy, cleaning execution cycle, high-pressure water jet intensity, duty cycle output and target clamping force to the control register of the corresponding test unit. The control module uses the modified register values ​​to drive the corresponding test unit to perform adaptive test operations. During the execution of control instructions, the programmable logic controller of each test unit uses the built-in sensor of the corresponding test unit to perform closed-loop tracking of the target control quantity.

[0133] The mapping module continuously reads the operating temperature data and the position feedback of the actuator, while the monitoring module continuously reads the system communication heartbeat packets and sensor threshold status. The terminal execution status re-enters the subsequent control cycle, thus forming a control data closed loop consisting of microscopic physical property feedforward parameters, equipment control commands, terminal execution status, and operating condition feedback. The control system eliminates sample overburning and cleaning residue physical phenomena caused by fixed test parameters throughout the entire process. Finally, it completes adaptive parameter adjustment while maintaining stable operation of multiple types of test units. The control system realizes closed-loop control from coal sample characteristic identification to terminal test parameter execution.

[0134] To aid in understanding the technical solution of this invention, the following provides an application example of adaptive collaborative control of multiple types of industrial robot testing equipment.

[0135] The application example receives microscopic property feedforward parameters sent by an external coal quality pretreatment information system and low-level state data output by multi-source sensors on site. At the same time, the application example pre-sets system memory parameters including an empirical table of heat source attenuation constants and a configuration file of a single-layer fully connected mapping neural network model trained offline. The application example uses the pre-set system memory parameters as a benchmark for the closed-loop feedback and effect comparison of the robot's physical collision avoidance constraints and terminal test parameters adaptive adjustment.

[0136] The interaction module sends a data request message with an absolute timestamp to the external coal quality pretreatment information system via an industrial Ethernet interface. It then extracts feedforward parameters of microscopic properties from the decoded coal sample, including nominal bulk density, dynamic powder repose angle, and dry ash-free volatile matter. Next, the interaction module establishes a global Cartesian coordinate system in system memory. It extracts the spatial Euclidean distance between the mechanical gripper and the maximum physical contour of the testing crucible, and adds a collision avoidance margin to this Euclidean distance. Finally, the interaction module performs three-dimensional boundary expansion processing on the end effector of the industrial robot to generate a global safety dynamic envelope sphere. Finally, the interaction module synchronously pushes the feedforward parameters of the microscopic properties and the three-dimensional spatial surface equations of the global safety dynamic envelope sphere into the memory stack of the underlying servo drive logic, thereby establishing the pre-data benchmark for the control system.

[0137] The mapping module polls the programmable logic controller of the heterogeneous testing equipment based on the underlying fieldbus protocol. The mapping module synchronously captures the position encoder feedback value of the actuator of the testing equipment and the operating temperature sampling value of the built-in thermocouple of the heating component. In addition, the mapping module combines the robot interpolation cycle of the control system and uses the timestamp alignment mechanism to compensate for the timing deviation of the bus transmission.

[0138] Subsequently, the mapping module performs mathematical dimensionality reduction operations based on the dimensional features of the 3D computer-aided design. Then, the mapping module transforms the position encoder feedback value into a dynamic capsule composed of spatial line segments and expansion radii, or the mapping module transforms the position encoder feedback value into a dynamic fan-shaped swept body composed of a fixed rotating center. In parallel, the mapping module constructs an environmental thermal radiation scalar field with spatial gradient distribution characteristics based on the geometric center coordinates of the opening surface of the heating component and the real-time sampled operating temperature data, using the natural base decay law. Then, it integrates the environmental thermal radiation scalar field and the dynamic geometric envelope to complete the digital modeling of the physical working space.

[0139] The constraint module continuously receives real-time data from the piezoelectric thin film sensor and inertial measurement unit arranged at the end gripper and flange. The constraint module performs inertial product compensation calculation to obtain the dynamic gain term by combining the equivalent transport mass. The constraint module then performs linear superposition of the dynamic gain term with the basic holding force of the test crucible to generate the dynamic target holding force.

[0140] Within the underlying interpolation sampling period, the constraint module calculates the effective Euclidean approach distance from the global safety dynamic envelope sphere to the dynamic geometric envelope. The constraint module also extracts the dynamic powder repose angle from the microscopic property feedforward parameters and substitutes it into the nonlinear decay mathematical model to generate the normal avoidance acceleration threshold. Finally, the constraint module uses the robot Jacobian matrix to convert the normal avoidance acceleration threshold into the allowable angular acceleration limit corresponding to each joint axis. The constraint module writes the allowable angular acceleration limit across processes into the limiting register of the corresponding servo driver, and then uses the dynamic target clamping force and the allowable angular acceleration limit to execute the underlying physical safety drive limit for cross-device loading actions.

[0141] The monitoring module continuously monitors the system communication heartbeat packets and sensor threshold status based on an independent background thread. When the abnormal working condition is determined to have reached the preset confirmation time, the monitoring module immediately generates a soft interrupt signal. Then, the monitoring module uses the soft interrupt signal to suspend the industrial robot's regular handling trajectory queue and freezes the pitch and roll angles in the end-wrist posture matrix of the industrial robot. Next, the monitoring module extracts the effective Euclidean approach distance and environmental thermal radiation scalar field parameters to construct a composite cost function.

[0142] The monitoring module calls the optimal solver to perform a minimization descent search on the composite cost function. Under the guidance of Gaussian random perturbation parameters, the monitoring module generates a spatial displacement vector that passes dynamic collision verification and thermal radiation safety verification. Finally, the monitoring module combines the normal avoidance acceleration threshold to perform envelope limiting on the Cartesian spatial acceleration command corresponding to the spatial motion displacement vector. The monitoring module uses the inverse kinematic mapping relationship to generate a smooth joint angular velocity command queue and sends the joint angular velocity command queue to the underlying servo driver, thereby driving the industrial robot to smoothly perform the asynchronous evacuation and return action to prevent spillage under the rest angle constraint.

[0143] After determining that the cross-equipment loading operation has not triggered abnormal operating conditions and that the industrial robot has safely withdrawn from the dynamic spatial interference boundary of the testing equipment, the control module calls the system to initialize the loaded mapping weight matrix and static bias constant column vector. Based on the microscopic property feedforward parameters, the control module performs adaptive linear spatial matrix multiplication and bias compensation calculation, and then outputs forward the target ignition energy of the calorimeter, the cleaning execution cycle and high-pressure water jet intensity of the oxygen bomb cleaning mechanism, the duty cycle output of the furnace temperature controller of the industrial analyzer, and the target clamping force of the hydrocarbon analysis package sample mechanism.

[0144] Subsequently, the control module performs item-by-item comparison calculations between each output value and the system's built-in set of physical limit safety thresholds. When any value is found to exceed the limit, the control module calls the built-in saturation limiting function to forcibly constrain the over-limit index. Finally, the control module uses the fieldbus to send the controlled target ignition energy, cleaning execution cycle, high-pressure water jet intensity, duty cycle output, and target clamping force to the control registers of each downstream testing unit, thereby connecting the front-end microscopic property feedforward parameters with the back-end testing terminal execution mechanism, ultimately realizing the action protection and adaptive parameter control closed loop of multiple types of testing equipment under extreme correlation conditions.

[0145] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result using substantially the same method falls within the scope of the present invention.

Claims

1. An adaptive cooperative control method for multiple types of industrial robot testing equipment, characterized in that, The method includes the following steps: Obtain the feedforward parameters of microscopic properties, and generate a global safety dynamic envelope sphere by performing boundary expansion on the end effector; Read the interference boundary data and operating temperature data of the testing equipment, map the interference boundary data into a dynamic geometric envelope, and establish an environmental thermal radiation scalar field based on the operating temperature data; Acquire end-feedback sensor data and calculate target clamping force by combining the microscopic property feedforward parameters; calculate effective Euclidean approach distance based on the global safety dynamic envelope sphere and the dynamic geometric envelope, generate avoidance acceleration threshold by combining the microscopic property feedforward parameters, and use the target clamping force and the avoidance acceleration threshold to perform low-level servo drive restrictions on the loading action; When an abnormal working condition is determined, a soft interrupt signal is generated to suspend the transport trajectory. A composite cost function is constructed and solved based on the effective Euclidean approach distance, the microscopic material property feedforward parameters and the environmental thermal radiation scalar field. The retreat path parameters are then generated to control the execution of asynchronous evacuation actions. When the abnormal operating conditions are not triggered, the preset operating condition mapping logic is invoked using the microscopic property feedforward parameters to perform calculation operations, generate and output control commands to drive the testing operation.

2. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The step of obtaining microscopic property feedforward parameters and performing boundary expansion on the end effector to generate a globally safe dynamic envelope sphere includes: The nominal bulk density, dynamic powder angle of repose, and dry ash-free volatile matter parameters of the coal sample were extracted as feedforward parameters for the microscopic properties. Extract the tool center point of the end effector as the spatial geometric sphere center, and extract the maximum physical contour size and superimpose the anti-collision reserve to generate the safety envelope radius; A three-dimensional boundary expansion process is performed in the global coordinate system using the spatial geometric sphere center and the safe envelope radius to generate the global safe dynamic envelope sphere of the spherical interference boundary.

3. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The steps of reading the interference boundary data and operating temperature data of the testing equipment, mapping the interference boundary data into a dynamic geometric envelope, and establishing an environmental thermal radiation scalar field based on the operating temperature data include: Based on the physical characteristics of the actuator of the laboratory equipment, the moving parts with linear displacement and rotational opening and closing characteristics are dimensionally reduced and mapped into dynamic capsule bodies and dynamic fan-shaped swept bodies to form the dynamic geometric envelope. Extract the geometric center coordinates of the opening surface of the heating component, and based on the geometric center coordinates and the operating temperature data, construct the environmental thermal radiation scalar field with spatial gradient distribution characteristics using the natural base decay law.

4. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The step of acquiring end-feedback sensor data and calculating the target clamping force by combining the microscopic physical property feedforward parameters includes: The end-feedback sensor data includes contact pressure and spatial acceleration values; the equivalent transport mass is calculated using the nominal bulk density of the coal sample in the microscopic property feedforward parameters, and the dynamic gain term is obtained by performing inertial product compensation calculation in combination with the spatial acceleration value; The dynamic gain term is linearly superimposed with the basic holding force, and the target holding force is obtained by performing closed-loop error correction based on the contact pressure value.

5. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The steps of calculating the effective Euclidean approach distance based on the global safety dynamic envelope sphere and the dynamic geometric envelope, generating an avoidance acceleration threshold by combining the microscopic material property feedforward parameters, and using the target clamping force and the avoidance acceleration threshold to implement low-level servo drive restrictions on the loading action include: The shortest spatial distance from the center of the global safety dynamic envelope sphere to the boundary of the dynamic geometric envelope is calculated as the effective Euclidean approach distance; A nonlinear decay model is established that correlates the effective Euclidean approach distance with the dynamic powder repose angle in the microscopic property feedforward parameters, and the upper limit of the limit maneuver is generated as the avoidance acceleration threshold. The target clamping force is converted into a driving command, and the avoidance acceleration threshold is converted into the allowable angular acceleration limit corresponding to each joint axis to constrain the driver output.

6. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The step of generating a soft interrupt signal to suspend the transport trajectory when an abnormal operating condition is determined includes: The system detects communication heartbeat, equipment fault status, temperature of heating components, and location feedback parameters. If any abnormal state persists for a preset confirmation time, it determines that the abnormal operating condition exists and generates the soft interrupt signal. The transport trajectory is suspended using the soft interrupt signal, the pitch and roll angles in the wrist attitude matrix of the end effector are frozen while the translational motion degrees of freedom are preserved, and a fault-based three-dimensional coordinate system is established based on the absolute position data of the joint encoder.

7. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The steps for constructing and solving a composite cost function based on the effective Euclidean approach distance, the microscopic property feedforward parameters, and the environmental thermal radiation scalar field, and generating retreat path parameters to control the execution of asynchronous evacuation actions, include: The composite cost function is constructed by introducing a repulsive potential field penalty term based on the effective Euclidean proximity distance and a penalty term based on the temperature gradient of the ambient thermal radiation scalar field. Starting from the three-dimensional coordinates of the fault dwelling, the displacement vector is calculated to generate candidate search directions. After collision and safety verification, the direction with the largest decrease in the composite cost function is selected to generate the retreat path parameters. When performing the asynchronous evacuation action, the spatial acceleration command corresponding to the retreat path parameters is envelope-limited using the avoidance acceleration threshold.

8. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, Before the step of using the microscopic property feedforward parameters to call the preset working condition mapping logic to perform the calculation, the method further includes the step of establishing the preset working condition mapping logic, specifically including: A historical mapping dataset is constructed by collecting the feedforward parameters of microscopic physical properties from historical test batches and the corresponding optimal operational feedback values. The historical mapping dataset is input into a single-layer fully connected mapping neural network, and the connection weight parameters and intrinsic bias parameters of the neurons are iteratively updated using the backpropagation algorithm until the training converges. Extract the fixed connection weight parameters and the inherent bias parameters to construct the preset working condition mapping logic.

9. The adaptive cooperative control method for multiple types of industrial robot testing equipment according to claim 1, characterized in that, The steps of using the aforementioned microscopic property feedforward parameters to call a preset working condition mapping logic to perform calculations, and generating and outputting control commands to drive the testing operation include: Based on the microscopic property feedforward parameters, the initial command value is calculated by performing the solution operation through matrix multiplication and bias compensation algorithm using the connection weight parameters and inherent bias parameters in the preset working condition mapping logic; The initial command value is compared item by item with a preset set of physical limit safety thresholds. When the value is determined to exceed the limit, the limiting function is called to forcibly constrain the exceeding index within the safe working range, and the control command is generated.

10. The adaptive cooperative control method for multi-type laboratory equipment using industrial robots according to claim 9, characterized in that, The step of generating and outputting control commands to drive the laboratory operation further includes: The control command after the amplitude limiting process is sent to the control register of the corresponding test unit to drive each test unit to perform the test operation; During the operation of each testing unit, the built-in sensor performs closed-loop tracking of the target control quantity, and combines the real-time feedback of interference boundary data and operating temperature data to connect the front-end physical property data and the terminal execution status to form a control data closed loop of the control system.

Images