Intelligent control method and system for industrial robot

Abstract

The invention discloses an intelligent control method and system for an industrial robot, and relates to the technical field of industrial automation, and the method comprises the steps: triggering a dynamic posture adjustment mechanism based on a preliminary comparison evaluation result, calculating an output structure stability index Estab, and quantifying the anti-interference capability of the current robot. And then combining the structural stability index Estab with the thermal disturbance aggregation trend index Rtend to calculate an attitude deviation angle Oproj. And further coupling the attitude deviation angle Oproj and the disturbance aggregation trend index Rtend, calculating to obtain an adjustment trigger decision function Jctrl, and performing secondary comparison evaluation with a set trigger threshold Cth to accurately judge whether the pipeline cleaning robot enters attitude adjustment logic or not. Through the above steps, linkage modeling and response control of the disturbance trend, the stability capability and the attitude risk can be realized, and the dynamic precision and robustness of the attitude adjustment system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial automation, and particularly to an intelligent control method and system for industrial robots. Background Art

[0002] The present invention relates to the technical field of industrial automation, and particularly to an industrial robot attitude control method in a pipeline operation environment. Specifically, it relates to an intelligent control method based on the fusion of thermal disturbance streamline modeling and structural stable state, which is applicable to a cleaning industrial robot system operating in a closed pipeline with high temperature, high humidity, and severe water mist disturbance. By constructing a thermal convection disturbance field model and the dynamic state of the robot's structural attitude in real time, it predicts the potential instability trend during its operation and takes control compensation in advance to achieve stable, continuous, and safe execution of operation tasks.

[0003] At the present stage, in the internal pipeline cleaning process of some high-temperature and high-pressure industrial processes, such as chemical industry, oil refining, and boiler systems, robots must penetrate into closed, curved, and high-temperature water mist-filled pipelines to perform cleaning tasks. Due to phenomena such as vortices and turbulent flows formed by the thermal convection disturbance streamline caused by continuous high-temperature liquid impact and complex structures, robots are extremely prone to problems such as attitude deviation, decreased stability, and even overturning. Conventional vision / gyro systems are extremely unreliable in this environment.

[0004] The core reason for the above deficiencies is that traditional control systems only perform closed-loop feedback control based on their own attitude states, while ignoring the prior trend information of the external disturbance environment. Especially in pipeline cleaning operations, the water mist formed by a large number of high-pressure nozzles will form a local convection vortex structure after encountering heat. Such disturbances have "spatial invisibility", "development suddenness", and "direction inconsistency". Without being modeled and perceived, they are extremely likely to act on the robot's shell to generate deflection torque and vibration fluctuations. If the system fails to identify the trend at the initial stage of the disturbance, resulting in a delay in control response, it will further cause serious abnormalities such as the robot overturning, getting stuck in the turning section, the attitude angle deviating sharply, and even damaging the execution structure, affecting the continuity and safety of the cleaning task. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides an intelligent control method and system for industrial robots, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: including the following steps:

[0007] S1. Set collection points on the pipeline cleaning robot, construct a thermal flow disturbance sensor array, and collect thermal disturbance data in real time through the thermal capacity density sensor array, and then transmit the thermal disturbance data to the central control server through a wired direct connection data line;

[0008] S2. Extracting features of the thermal disturbance data in the central control server to obtain a thermal disturbance feature set, and preprocessing the thermal disturbance feature set to obtain a standardized digital set;

[0009] S3. Calculate and output the thermal disturbance aggregation trend index Rtrend based on the standardized digital set, set the trend interval threshold and the thermal disturbance aggregation trend index Rtrend for preliminary comparative evaluation, and trigger the self-stabilizing control mechanism based on the preliminary comparative evaluation results;

[0010] S4, triggering dynamic posture adjustment based on preliminary comparative evaluation, by extracting the structural stability index Estab of the current robot and combining it with the thermal disturbance aggregation trend index Rtrend to calculate and output the posture deviation angle Oproj;

[0011] S5. Comprehensively calculate the thermal disturbance aggregation trend index Rtrend and the attitude deviation angle Oproj to output the adjustment trigger decision function Jctrl, and set the trigger threshold Cth for secondary comparison and evaluation to trigger the control logic.

[0012] Preferably, said S1 includes S11 and S12;

[0013] S11, setting collection points at symmetrical positions of the front, back, left and right of the pipeline cleaning robot, and installing a thermal flow disturbance sensor array in each collection point to collect thermal disturbance data of each collection point in real time, wherein the number of thermal flow disturbance sensor arrays set is greater than or equal to 4 groups;

[0014] The thermal flow disturbance sensor array includes an array-type MEMS thermal sensor, a TOF ultrasonic microwave transmitter, a TOF ultrasonic microwave receiver and an infrared thermal imaging sensor;

[0015] The thermal disturbance data includes the temperature T of the i-th collection point i , return signal, heat convection main direction vector vflow and robot forward direction vector vmove;

[0016] S12. The thermal disturbance data collected in real time is transmitted to the central control server of the pipeline cleaning robot through a data transmission line directly connected to the back end of the pipeline cleaning robot.

[0017] Preferably, S2 includes S21 and S22;

[0018] S21. Extracting features of thermal disturbance data in the central control server to obtain a thermal disturbance feature set;

[0019] The thermal disturbance feature set includes the thermal gradient Rtgrad(i) at the i-th acquisition point, the disturbance vortex density Pvortex(i) at the i-th acquisition point, and the convection direction deviation Ddiv(i) at the i-th acquisition point;

[0020] The thermal gradient Rtgrad is calculated and extracted by dividing the temperature difference T between each pair of array-type MEMS thermosensitive sensor points by the distance between the sensor point pairs;

[0021] The disturbance vortex density Pvortex is obtained by the TOF ultrasonic microwave transmitter emitting ultrasonic pulses, and then the TOF ultrasonic microwave receiver receiving the return signal. The disturbance waveform spectrum of the return signal is extracted using the fast Fourier transform FFT. Based on the disturbance waveform spectrum and setting the disturbance peak determination rule, the proportion of the number of disturbance events in the statistical unit time window in the spatial region is derived and obtained;

[0022] The convection direction deviation Ddiv is calculated and extracted by performing the arccosine function arccos based on the main thermal convection direction vector vflow and the robot forward direction vector vmove;

[0023] S22. Preprocess the thermal disturbance feature set to obtain a standardized digital set;

[0024] The preprocessing includes outlier detection and normalization processing;

[0025] The outlier detection uses the IQR interquartile range method to remove extreme fluctuations in the thermal disturbance feature set, and then uses the mean vector method to complete the data of the thermal disturbance feature set at the removed positions;

[0026] The normalization processing uses the Min-Max normalization method to normalize physical quantities of different dimensions and scales to the interval [0, 1] for the thermal disturbance feature set after outlier detection, so as to eliminate the influence of dimensions.

[0027] Preferably, S3 includes S31 and S32;

[0028] S31. Based on the standardized digital set obtained from all acquisition points, calculate and output the disturbance aggregation trend index Rtrend, and share the thermal disturbance focusing degree, direction deflection degree, and spatial complexity within the acquisition points;

[0029] The disturbance aggregation trend index Rtrend is calculated and output through the following algorithm formula;

[0030] ;

[0031] In the formula, N represents the total number of acquisition points, and sin represents the sine function.

[0032] Preferably, in S32, the trend interval threshold is set by an experimental calibration method. The trend interval threshold includes a first trend threshold F1 and a second trend threshold F2. When the disturbance aggregation trend index Rtrend obtained in real time is preliminarily compared with the trend interval threshold to evaluate the temperature disturbance, vortex density, and deflection direction of the current pipeline cleaning robot, and regional division is performed based on the preliminary comparison and evaluation results. The specific evaluation content is as follows;

[0033] When the disturbance aggregation trend index Rtrend < the first trend threshold F1, it indicates that the disturbance is dispersed. At this time, it is divided into a stable flow area, and the current operating state is continued;

[0034] When the first trend threshold F1 ≤ the disturbance aggregation trend index Rtrend < the second trend threshold F2, it indicates that the disturbance in the current area is aggregated. At this time, it is divided into a controllable disturbance area, and the attitude warning compensation mode is started;

[0035] In the attitude warning compensation mode, by extracting the standardized digital set of the next cycle, the disturbance aggregation trend index Rtrend is calculated and output after secondary calculation, and secondary preliminary evaluation is performed. When it is still in the range of the first trend threshold F1 ≤ the disturbance aggregation trend index Rtrend < the second trend threshold F2, the current area is divided into a disturbance abnormal area at this time, and dynamic attitude adjustment is triggered;

[0036] When the disturbance aggregation trend index Rtrend ≥ the second trend threshold F2, it indicates that the thermal disturbance vortices are concentrated and the direction is significantly skewed. At this time, it is divided into a disturbance abnormal area, and dynamic attitude adjustment is triggered.

[0037] Preferably, the S4 includes S41 and S42;

[0038] S41. After the dynamic attitude adjustment is triggered in the preliminary comparison and evaluation, the built-in three-axis gyroscope, acceleration IMU module, and anti-seismic IMU of the pipeline cleaning robot are started to collect the attitude data of the robot in real time. After the attitude data is normalized, the structural stability index Estab is calculated and extracted;

[0039] The attitude data includes the moment of inertia Irot of the robot around the center, the current angular velocity w, and the vibration impact force Fvib;

[0040] The structural stability index Estab is calculated and output through the following algorithm formula;

[0041] ;

[0042] In the formula, Mbase represents the self-weight of the robot, and g represents the acceleration due to gravity, both of which are initially set by the user and are dimensionless.

[0043] Preferably, in S42, based on the obtained structure stability index Estab and the thermal disturbance aggregation trend index Rtrend, the attitude offset angle Oproj is calculated and output, and the maximum attitude offset angle of the pipeline cleaning robot due to disturbance effects is analyzed;

[0044] The attitude offset angle Oproj is calculated and output through the following algorithm formula;

[0045] ;

[0046] In the formula, represents the disturbance amplification factor, which is dimensionless and is set by the user according to the sensitivity of the robot structure to the disturbance trend.

[0047] Preferably, the S5 includes S51 and S52;

[0048] S51, based on the obtained attitude offset angle Oproj and the disturbance aggregation trend index Rtrend, couples its own adjustment ability with the trend development speed of the current external disturbance, and comprehensively calculates and outputs the adjustment trigger decision function Jctrl to measure the attitude of the current pipeline cleaning robot during operation;

[0049] The adjustment trigger decision function Jctrl is calculated and output through the following algorithm formula;

[0050] ;

[0051] In the formula, wcomp represents the compensation rate, which is dimensionless, d represents the integral function, and dt represents the time integral function, represents a very small number to prevent division by zero, represents the disturbance trend growth rate, represents the weight factor of the disturbance trend growth rate, which is dimensionless and is set by the user.

[0052] Preferably, in S52, by extracting the maximum compensation rate wcomp of the robot's tracked differential control max , and at the same time based on the attitude offset angle Oproj under the maximum controllable disturbance max , the maximum compensation rate wcomp max and the attitude offset angle Oproj max are ratio-calculated, and a disturbance development speed superposition factor is added to set the trigger threshold Cth. Then, the real-time obtained adjustment trigger decision function Jctrl is compared and evaluated with the trigger threshold Cth for the second time to analyze the attitude operation of the current pipeline cleaning robot, and trigger control logic is performed according to the results of the second comparison and evaluation. The specific evaluation content is as follows;

[0053] When the adjustment trigger decision function Jctrl ≤ the trigger threshold Cth, it indicates that the current pipeline cleaning robot has the self-stabilization ability. At this time, the first control logic is triggered;

[0054] When the adjustment trigger decision function Jctrl > the trigger threshold Cth, it indicates that the disturbance development of the current Hiroshima cleaning robot exceeds the upper limit of the control ability. At this time, the second control logic is triggered;

[0055] The first control logic symmetrically adjusts the wheel speed of the crawlers of the pipeline cleaning robot to maintain forward stability, and at the same time activates the lateral micro-push state and does not activate the high-energy-consuming device;

[0056] The second control logic automatically activates the self-stabilizing bracket of the pipeline cleaning robot and decelerates by 50%.

[0057] An intelligent control system for an industrial robot includes a sensing and acquisition module, a sensing and processing module, a disturbance analysis module, an attitude analysis module, and an integrated control module;

[0058] The sensing and acquisition module sets acquisition points on the pipeline cleaning robot to construct a heat flux disturbance sensor array, and real-time collects heat disturbance data through the heat capacity density sensor array, and then transmits the heat disturbance data to the central control server through a wired direct connection data line;

[0059] The sensing and processing module extracts features from the heat disturbance data in the central control server to obtain a heat disturbance feature set, and preprocesses the heat disturbance feature set to obtain a standardized digital set;

[0060] The disturbance analysis module calculates and outputs the heat disturbance aggregation trend index Rtrend based on the standardized digital set, sets a trend interval threshold for a preliminary comparison and evaluation with the heat disturbance aggregation trend index Rtrend, and triggers the self-stabilization control mechanism based on the preliminary comparison and evaluation results;

[0061] The attitude analysis module triggers dynamic attitude adjustment based on the preliminary comparison and evaluation, and calculates and outputs the attitude offset angle Oproj by extracting the structural stability index Estab of the current robot and combining it with the heat disturbance aggregation trend index Rtrend;

[0062] The integrated control module comprehensively calculates and outputs the adjustment trigger decision function Jctrl by combining the heat disturbance aggregation trend index Rtrend and the attitude offset angle Oproj, and sets a trigger threshold Cth for a secondary comparison and evaluation to trigger the control logic.

[0063] The present invention provides an intelligent control method and system for an industrial robot. It has the following beneficial effects:

[0064] (1) By setting symmetric acquisition points in the front, back, left, and right directions on the pipeline cleaning robot, a heat flux disturbance sensor array including MEMS thermal sensors, TOF ultrasonic microwave devices, and infrared thermal imaging modules is constructed. This enables real-time acquisition of heat disturbance data in complex disturbance environments inside the pipeline, such as high temperature, water mist, and structural corners, and transmits the data to the central control server via a wired data link. After preprocessing the heat disturbance data through feature extraction, outlier detection, and normalization, a standardized digital set is formed to ensure the accuracy and stability of subsequent calculations. By performing aggregation operations on the standardized digital set, the disturbance aggregation trend index Rtrend is obtained, and combined with the set trend interval threshold calibrated through experiments, the classification of the disturbance state level and the prediction of trend evolution are realized, significantly improving the robot's recognition ability and response foresight before the occurrence of disturbances.

[0065] (2) This method triggers the dynamic attitude adjustment mechanism based on the preliminary comparison evaluation results, and calculates and outputs the structural stability index Estab by using the rotational inertia Irot, angular velocity w, and vibration impact force Fvib collected by the IMU sensor to quantify the current anti-interference ability of the robot. Subsequently, the structural stability index Estab is combined with the heat disturbance aggregation trend index Rtrend, and by setting the disturbance amplification factor , the predicted attitude deviation angle Oproj is calculated to effectively estimate the maximum deviation risk that the robot may encounter within the next Δt time. Further coupling the attitude deviation angle Oproj and the disturbance aggregation trend index Rtrend, combining the current maximum compensation rate wcomp and the disturbance development speed, a regulation trigger decision function Jctrl is constructed and compared with the set trigger threshold Cth for evaluation to accurately determine whether the robot enters the attitude adjustment logic. Through the above steps, the linkage modeling and response control of the disturbance trend, stability ability, and attitude risk can be realized, improving the dynamic accuracy and robustness of the attitude adjustment system.

[0066] (3) By setting a double-layer evaluation mechanism that adjusts the trigger decision function Jctrl and the trigger threshold Cth, this method can differentially control the response strategies of the robot under different disturbance levels. Among them, if the adjusted trigger decision function Jctrl is less than or equal to the trigger threshold Cth, the first control logic is triggered, and fine-tuning compensation is completed only through symmetric adjustment of the crawler wheel speed and lateral micro-pushing state to maintain low-power operation; when the adjusted trigger decision function Jctrl exceeds the trigger threshold Cth, the second control logic is entered, and the self-stabilizing bracket is automatically activated and a 50% speed reduction measure is implemented to ensure stable operation in a severely disturbed environment. By organically integrating the attitude prediction model, compensation ability evaluation, and disturbance growth trend modeling, this method realizes the on-demand allocation and hierarchical response of system control resources, effectively avoiding the energy consumption redundancy and control overshoot problems caused by the traditional system's response lag or full-time high-energy consumption operation, thereby improving the energy efficiency, safety, and long-term stability of the overall system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Schematic diagram of the steps of an intelligent control method for an industrial robot according to the present invention;

[0068] Figure 2 Schematic diagram of the process of an intelligent control system for an industrial robot according to the present invention;

[0069] Figure 3 Schematic diagram of the collection point settings of the pipeline cleaning robot according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment 1

[0072] Please refer to Figure 1 , the present invention provides an intelligent control method for an industrial robot. To achieve the above objectives, the present invention is implemented through the following technical solutions: including the following steps:

[0073] S1. Set collection points on the pipeline cleaning robot, construct a heat flow disturbance sensor array, and collect heat disturbance data in real time through the heat capacity density sensor array, and then transmit the heat disturbance data to the central control server through a wired direct connection data line;

[0074] S2. Extract features from the thermal disturbance data in the central control server to obtain a thermal disturbance feature set, and preprocess the thermal disturbance feature set to obtain a standardized digital set;

[0075] S3. Calculate and output the thermal disturbance aggregation trend index Rtrend based on the standardized digital set, set a trend interval threshold for a preliminary comparison and evaluation with the thermal disturbance aggregation trend index Rtrend, and trigger a self-stabilizing control mechanism based on the preliminary comparison and evaluation results;

[0076] S4. Trigger dynamic attitude adjustment based on the preliminary comparison and evaluation. By extracting the structural stability index Estab of the current robot and combining it with the thermal disturbance aggregation trend index Rtrend, calculate and output the attitude offset angle Oproj;

[0077] S5. Perform a comprehensive calculation on the thermal disturbance aggregation trend index Rtrend and the attitude offset angle Oproj to output a regulation trigger decision function Jctrl, set a trigger threshold Cth for a secondary comparison and evaluation, and trigger the control logic.

[0078] In this embodiment, the method accurately collects raw thermal disturbance data by setting up a thermal disturbance sensor array, and ensures real-time and reliable data access to the central control server through wired transmission; by performing feature extraction, outlier elimination and normalization processing on the raw thermal disturbance data, a standardized digital set of unified scale is constructed to provide basic data guarantee for subsequent calculations. By calculating the output disturbance aggregation trend index Rtrend and performing a preliminary comparative evaluation based on the set multi-level trend interval threshold, the disturbance distribution state can be identified and the self-stabilizing control mechanism can be triggered in stages. After entering S4, the real-time collected posture data is combined to construct the structural stability index Estab with rotational inertia, angular velocity and impact force as inputs, and the model is jointly built with Rtrend to predict the output posture deviation angle Oproj, so as to identify the maximum posture risk that may occur within the future Δt. Finally, in S5, the above-mentioned posture deviation angle Oproj and the disturbance aggregation trend index Rtrend are incorporated into the construction of the adjustment trigger decision function Jctrl, and the system compensation rate and the disturbance trend change rate are introduced as control parameters. The dynamic trigger threshold Cth is used to complete the secondary comparative evaluation, and the control logic is intelligently switched to achieve the optimal match between fine-tuning compensation and active posture adjustment mechanism. Through the above-mentioned implementation mode, the present invention realizes the closed-loop control process of "disturbance trend modeling, posture prediction analysis, intelligent decision-making evaluation and action response execution", and has the following beneficial effects: on the one hand, compared with the existing control system that only relies on angle feedback, the method can make a predictive response before the disturbance occurs, effectively avoiding the problems of drastic posture deviation and control lag; on the other hand, with the help of multi-level response mechanism and system compensation capability evaluation strategy, the control resource utilization efficiency and system energy saving are effectively improved, and the false triggering rate of high-energy consumption adjustment actions is reduced, thereby significantly enhancing the stable operation capability, adaptability and operation continuity of the pipeline cleaning robot in a high-disturbance and high-uncertainty operating environment.

[0079] Example 2

[0080] See also Figure 1 and Figure 3 , specifically: S1 includes S11 and S12;

[0081] S11, setting collection points at symmetrical positions of the front, back, left and right of the pipeline cleaning robot, and installing a thermal flow disturbance sensor array in each collection point to collect thermal disturbance data of each collection point in real time, wherein the number of thermal flow disturbance sensor arrays set is greater than or equal to 4 groups;

[0082] The thermal flow disturbance sensor array includes an array MEMS thermal sensor, a TOF ultrasonic microwave transmitter, a TOF ultrasonic microwave receiver and an infrared thermal imaging sensor;

[0083] Thermal disturbance data includes the temperature T of the i-th acquisition point i, return signal, main direction vector vflow of heat convection, and robot forward direction vector vmove;

[0084] S12. Through the data transmission line directly connected to the rear end of the pipeline cleaning robot, transmit the real-time collected heat disturbance data to the central control server of the pipeline cleaning robot.

[0085] In this embodiment, the method constructs a uniformly distributed and fast-responsive disturbance perception network by arranging acquisition points symmetrically around the pipeline cleaning robot and installing no less than four groups of heat flow disturbance sensor arrays at each acquisition point; the sensor array includes a variety of heterogeneous sensing elements such as arrayed MEMS thermal sensors, TOF ultrasonic microwave transmitters and receivers, and infrared thermal imaging sensors, which can collect key disturbance data such as temperature information T, acoustic wave return signal, main direction vector vflow of heat convection, and robot forward direction vector vmove in real time, realizing the full-dimensional identification of the intensity, direction deviation, and vortex structure of heat convection disturbance in the pipeline. By directly uploading the heat disturbance data collected at each acquisition point to the central control server via the direct-connected wired data transmission path at the rear end of the robot, the problem of transmission attenuation of wireless signals in water mist or high-temperature metal pipelines is avoided, ensuring high real-time performance, high integrity, and low error rate of the sensing data, and providing highly reliable original basic data for subsequent heat disturbance feature extraction and trend analysis.

[0086] Embodiment 3

[0087] Please refer to Figure 1 and Figure 3 , specifically: S2 includes S21 and S22;

[0088] S21. Extract features from the heat disturbance data in the central control server to obtain a heat disturbance feature set;

[0089] The heat disturbance feature set includes the heat gradient Rtgrad(i) of the i-th acquisition point, the disturbance vortex density Pvortex(i) of the i-th acquisition point, and the convection direction deviation Ddiv(i) of the i-th acquisition point;

[0090] The heat gradient Rtgrad is calculated and extracted by dividing the temperature difference T between pairs of points of each arrayed MEMS thermal sensor by the distance between the sensor point pairs;

[0091] The perturbation vortex density Pvortex emits ultrasonic pulses through a TOF ultrasonic microwave transmitter, and then receives the returned signal through a TOF ultrasonic microwave receiver. The fast Fourier transform FFT is used to extract the perturbation waveform spectrogram of the returned signal. Based on the perturbation waveform spectrogram and setting the perturbation peak determination rule, the proportion of the number of perturbation events in the statistical unit time window in the spatial region is obtained. For example, a determination threshold is set. If the amplitude in a certain frequency band exceeds the set value, each time a perturbation peak cluster is recognized, it is regarded as a perturbation event;

[0092] The significance of extracting the perturbation vortex density Pvortex is that in a pipeline, when high-temperature water mist contacts the cold metal wall surface, encounters a bend, structural change, reverse impact of a nozzle, etc., it will cause perturbations in the vortex structure at the micro scale;

[0093] The convection direction deviation Ddiv is extracted by performing an arccos calculation based on the main thermal convection direction vector vflow and the robot's forward direction vector vmove. The specific algorithm formula is:

[0094] , where arccos represents the inverse cosine function, which is used to convert the direction similarity into an angle. It is used for normalization to ensure that the formula result is in the range of [0, 1]. The significance of extraction is the deviation angle between the main convection direction and the robot's forward direction, which reflects the perturbation direction deflection trend;

[0095] S22. Preprocess the thermal perturbation feature set to obtain a standardized digital set;

[0096] The preprocessing includes outlier detection and normalization;

[0097] Outlier detection uses the IQR interquartile range method to remove extreme fluctuations in the thermal perturbation feature set, and then uses the mean vector method to complement the data of the thermal perturbation feature set at the removed positions;

[0098] Normalization processing uses the Min - Max normalization method for the thermal perturbation feature set after outlier detection to normalize physical quantities of different dimensions and scales into the range of [0, 1] to eliminate the influence of dimensionality.

[0099] In this embodiment, the method extracts features from the real-time collected thermal disturbance data through the central control server, and constructs a thermal disturbance feature set consisting of the thermal gradient Rtgrad (i) of the i-th collection point, the disturbance vortex density Pvortex (i) of the i-th collection point, and the convection direction deviation Ddiv (i) of the i-th collection point. Among them, the thermal gradient Rtgrad is obtained by calculating the temperature difference and spacing ratio between the MEMS thermistor point pairs, accurately reflecting the local temperature field change rate, and the disturbance vortex density Pvortex is realized by the time-frequency analysis (FFT) of the TOF ultrasonic signal and the disturbance peak cluster recognition algorithm to achieve micro-scale vortex event statistics; and the convection direction deviation Ddiv is calculated by the arccosine function conversion of the angle between the heat flow vector vflow and the forward vector vmove, which is used to evaluate the degree of deviation between the main convection direction and the path direction. The three together construct the spatial structural feature expression of the thermal disturbance environment. At the same time, in order to ensure the stability and uniformity of the thermal disturbance feature set, further preprocessing operations are performed on it: first, the IQR interquartile range method is used to eliminate extreme fluctuation points in the data, and then the mean vector method is used to fill in the missing data area to ensure data continuity and noise resistance; then the Min-Max normalization method is used to standardize each physical quantity to the [0,1] interval, completely eliminating the dimensional differences and scale offset problems between the physical quantities, so that it is suitable for subsequent index construction and formula calculation.

[0100] Example 4

[0101] See also Figure 1 , specifically: S3 includes S31 and S32;

[0102] S31. Based on the standardized digital set obtained from all the collection points, calculate and output the disturbance aggregation trend index Rtrend, and share the degree of thermal disturbance focusing, directional deflection and spatial complexity within the collection point;

[0103] The disturbance aggregation trend index Rtrend is calculated and output by the following algorithm formula;

[0104] ;

[0105] In the formula, N represents the total number of acquisition points, and sin represents the sine function;

[0106] Indicates the deviation direction amplitude correction factor. The larger the value, the more serious the deviation of the pipeline cleaning robot.

[0107] The calculation logic of the formula is to multiply the parameters in the standardized digital set to form the disturbance intensity of each collection point, and then perform summation and averaging logic to analyze the overall disturbance trend. If the disturbance in a certain direction is severely concentrated, even if other points are stable, the result will be pulled up. This is a kind of clustered trend identification, not whether it is currently oscillating, but a signal of multi-point convergent disturbance.

[0108] S32, setting the trend interval threshold by experimental calibration method, the trend interval threshold includes the first trend threshold F1 and the second trend threshold F2, and making a preliminary comparison and evaluation between the disturbance aggregation trend index Rtrend obtained in real time and the trend interval threshold, judging the temperature disturbance, vortex density and deflection direction of the current pipeline cleaning robot, further analyzing the robot's possible imbalance, slippage and deviating from the track due to the impact of the disturbance flow, and dividing the area based on the preliminary comparison and evaluation results, the specific evaluation content is as follows;

[0109] When the disturbance aggregation trend index Rtrend is less than the first trend threshold F1, it means that the disturbance is dispersed. At this time, it is divided into a stable flow area and continues to maintain the current operation state;

[0110] When the first trend threshold F1≤disturbance aggregation trend index Rtrend<second trend threshold F2, it indicates that disturbances are aggregated in the current area. At this time, it is divided into a controllable disturbance area and the attitude warning compensation mode is started;

[0111] The attitude warning compensation mode extracts the standardized digital set of the next cycle, performs secondary calculation to output the disturbance aggregation trend index Rtrend, and performs secondary preliminary evaluation. When the first trend threshold F1 ≤ disturbance aggregation trend index Rtrend < second trend threshold F2, the current area is divided into the abnormal disturbance area, triggering dynamic attitude adjustment.

[0112] When the disturbance aggregation trend index Rtrend ≥ the second trend threshold F2, it means that the thermal disturbance vortex is concentrated and the direction is obviously deflected. At this time, it is divided into the disturbance abnormal area and the dynamic attitude adjustment is triggered;

[0113] Experimental calibration method:

[0114] Collect a large amount of real data in a simulated environment;

[0115] Count the range of disturbance aggregation trend index Rtrend generated on the “eve of instability”;

[0116] The boundary values ​​of the medium and high risk intervals are set, with the first trend threshold F1 being 0.5 and the second trend threshold F2 being 1.2.

[0117] In this embodiment, based on the standardized digital set completed in the preprocessing in S2, the coupling operation is performed on the disturbance characteristic parameters of each acquisition point, and the thermal gradient Rtgrad, the disturbance vortex density Pvortex and the direction deviation Ddiv are comprehensively considered to construct the disturbance aggregation trend index Rtrend. By summing and averaging the product of the three parameters as the disturbance intensity factor, this index can effectively identify whether there is an aggregation evolution phenomenon of "consistent direction and multi-point convergence" in the internal disturbance of the pipeline, and then judge whether the robot is about to be in a risk state such as potential imbalance, deviation or slipping. It is a disturbance perception index with both structural, trend and predictive characteristics. By further combining the experimental calibration method, the first trend threshold F1 and the second trend threshold F2 are set as the basis for dividing the disturbance levels. After preliminary comparative evaluation, the area where the current robot is located can be automatically divided into a "stable flow area", a "controllable disturbance area" or a "disturbance abnormal area". By executing the steps in S3, the present invention effectively constructs a set of thermal disturbance trend index modeling and risk level determination mechanism, which not only breaks through the problem that traditional robots cannot identify the external disturbance trend, but also realizes a closed-loop control process from disturbance perception, trend modeling, multi-level risk prediction and attitude adjustment triggering, greatly improving the dynamic adaptability, control judgment accuracy and attitude stability maintenance efficiency of the robot system in a complex disturbance environment, and providing a clear trigger logic and quantitative basis for the reasonable execution of subsequent attitude control actions.

[0118] Embodiment 5

[0119] Please refer to Figure 1 , specifically: S4 includes S41 and S42;

[0120] S41. After the preliminary comparative evaluation triggers the dynamic attitude adjustment, start the three-axis gyroscope, acceleration IMU module and anti-seismic IMU of the pipeline cleaning robot to collect the attitude data of the robot in real time, and after normalizing the attitude data, calculate and extract the structural stability index Estab;

[0121] The attitude data includes the rotational inertia Irot of the robot around the center, the current angular velocity w and the vibration impact force Fvib;

[0122] The structural stability index Estab is calculated and output through the following algorithm formula;

[0123] ;

[0124] In the formula, Mbase represents the self-weight of the robot, and g represents the acceleration of gravity, both of which are initially set by the user and are dimensionless;

[0125] The calculation logic and physical meaning of the formula, the peak value represents the disturbance input energy, where, It represents the dynamic rotational kinetic energy. The vibration impact force Fvib is used to analyze the instantaneous force disturbance, and the denominator represents the basic ability of the structure to resist interference. It is equivalent to the structural gravity stabilizing moment. The ratio of the numerator to the denominator is calculated to measure the disturbance energy and the basic ability of the robot structure to resist interference, reflecting the self-attitude stability of the robot.

[0126] S42. Based on the obtained structural stability index Estab and the thermal disturbance aggregation trend index Rtrend, calculate and output the attitude offset angle Oproj, and analyze the maximum attitude offset angle of the pipeline cleaning robot due to the influence of disturbances.

[0127] The attitude offset angle Oproj is calculated and output through the following algorithm formula;

[0128] ;

[0129] In the formula, represents the disturbance amplification coefficient, which is dimensionless and is set by the user according to the sensitivity of the robot structure to the disturbance trend. represents the positive correlation term, which is the "absolute influence degree" of the disturbance. The disturbance amplification coefficient converts the disturbance index into an amplification factor of "actual angle influence", representing that the softer the robot structure and the higher the center of gravity, the larger the disturbance amplification coefficient the larger; the more stable and rigid the robot, the smaller the disturbance amplification coefficient;

[0130] Specific example:

[0131] Suppose at a certain moment, the thermal disturbance aggregation trend index Rtrend = 1.5, the disturbance is large, the structural stability index Estab = 0.6, the structure is in a medium-stable state, and the disturbance amplification coefficient = 3.5;

[0132] .

[0133] In this embodiment, after it is evaluated and confirmed through the previous steps that the thermal disturbance aggregation trend index Rtrend has reached the condition to be adjusted, the method automatically invokes the built-in three-axis gyroscope, acceleration IMU module, and anti-seismic IMU device of the pipeline cleaning robot to perform real-time sampling of the current attitude state. By obtaining the moment of inertia Irot of the robot around the central axis, the current angular velocity w, and the instantaneous vibration impact force Fvib, and combining the preset self-weight Mbase of the robot and the standard gravitational acceleration g, the structural stability index Estab is calculated and output. This index takes the ratio of the dynamic disturbance input energy, rotational kinetic energy, and impact energy to the structural gravity stabilizing moment of the robot as the core, and truly reflects the current attitude stability ability and anti-interference level of the robot. Based on the real-time obtained structural stability index Estab and thermal disturbance aggregation trend index Rtrend, the attitude offset angle Oproj is calculated and output as the maximum attitude angle offset value that the robot may generate in the short term in the future under the influence of disturbances. A disturbance amplification factor is introduced in this calculation. The user can set it according to factors such as the structural stiffness of the robot and the layout of the center of gravity. The larger the value, the higher the sensitivity of the robot to disturbances. This angle prediction model not only has adjustability and individual adaptability, but also deeply couples external disturbances and internal structural states to form an integrated prediction mechanism of disturbance and response. In summary, through the implementation of step S4, on the basis of completing the thermal disturbance trend evaluation, the present invention further establishes a fusion path between the structural mechanics model and the disturbance trend model. By constructing the structural stability index Estab and the attitude offset angle Oproj, for the first time, a quantitative prediction of the attitude offset trend of the robot is realized, providing clear, continuous, and physically interpretable prediction indicators for the subsequent execution of control strategies, effectively improving the response accuracy of the system to disturbance risks, the dynamic adaptation ability of control parameters, and the forward-looking and rationality of the overall attitude adjustment

[0134] Embodiment 6

[0135] Please refer to Figure 1 , specifically: S5 includes S51 and S52;

[0136] S51. Based on the obtained attitude offset angle Oproj and the thermal disturbance aggregation trend index Rtrend, couple the self-adjusting ability with the trend development speed of the current external disturbance, and comprehensively calculate and output the adjustment trigger decision function Jctrl to measure the attitude of the current pipeline cleaning robot during operation;

[0137] The adjustment trigger decision function Jctrl is calculated and output through the following algorithm formula;

[0138] ;

[0139] Wherein, wcomp represents the compensation rate, which is dimensionless and obtained from the dynamic response of the actuator of the pipeline robot, indicating the maximum achievable compensation rate at present; d represents the integral function, and dt represents the time integral function. Represents a very small number to prevent division by zero, dimensionless, and the specific value is 0.000001. Represents the growth rate of the disturbance trend. Represents the weight factor of the growth rate of the disturbance trend, dimensionless, and is set by the user.

[0140] Calculation logic of the formula: The attitude offset angle Oproj in the numerator represents how many angles the external disturbance will cause the offset; the larger it is, the stronger the disturbance and the worse the stability.

[0141] The denominator term Represents the maximum executable adjustment ability; how fast the robot control system can correct the offset angle at most; if the control ability is low and the response is slow, then the denominator is small, resulting in a larger ratio and easier to trigger adjustment.

[0142] Growth rate of the disturbance trend , represents the growth rate of the disturbance index, that is, how fast the external disturbance develops; the faster the trend, the greater the risk, and the system should respond more actively. Represents the sensitivity to the trend speed, related to the structure. Set it high for light and sensitive robots; set it low for heavy and stable robots.

[0143] S52. By extracting the maximum compensation rate wcomp of the differential speed control of the robot track max , and based on the attitude offset angle Oproj under the maximum controllable disturbance max , the maximum compensation rate wcomp max and the attitude offset angle Oproj max are used for ratio calculation, and a disturbance development speed superposition factor is added to set the trigger threshold Cth. Then, the real-time obtained adjustment trigger decision function Jctrl is compared with the trigger threshold Cth for secondary comparison and evaluation to analyze the current attitude operation of the pipeline cleaning robot, and trigger control logic is carried out according to the results of the secondary comparison and evaluation. The specific evaluation content is as follows;

[0144] When the adjustment trigger decision function Jctrl ≤ trigger threshold Cth, it indicates that the current pipeline cleaning robot has the ability of self-stabilization, and at this time, the first control logic is triggered;

[0145] When the adjustment trigger decision function Jctrl > trigger threshold Cth, it indicates that the disturbance development of the current Hiroshima cleaning robot exceeds the upper limit of the control ability, and at this time, the second control logic is triggered;

[0146] The first control logic adjusts the wheel speed of the track of the pipeline cleaning robot symmetrically to maintain forward stability, activates the lateral micro-propulsion state, and does not activate high-energy consumption devices;

[0147] The second control logic automatically turns on the self-stabilizing support of the pipe cleaning robot and slows it down by 50%;

[0148] Specific example of setting the trigger threshold Cth: Assuming the maximum compensation rate wcomp max =10, attitude deviation angle Oproj under maximum controllable disturbance max =25, the disturbance development speed superposition factor is about 1.0, and the trigger threshold Cth=25 / 10+1.0=3.5.

[0149] In this embodiment, the method constructs the adjustment trigger decision function Jctrl based on the posture deviation angle Oproj and the disturbance aggregation trend index Rtrend calculated in S4, and jointly considers the maximum adjustment rate wcomp that the current actuator of the robot can achieve, as well as the growth rate of the disturbance trend over time. This function not only reflects the proportional relationship between the disturbance impact amplitude and the control response capability, but also integrates the urgency factor of the disturbance evolution to form a highly comprehensive posture risk judgment indicator. By setting the disturbance trend sensitive weight factor , which can realize the adaptive adjustment of robots of different structural types, such as light, highly sensitive or heavy and high inertia. By further obtaining the upper limit value wcomp of the robot track differential control capability max , and the maximum controllable attitude deviation angle Oproj calibrated in the actual operating environment max , combined with the disturbance speed development factor, the trigger threshold Cth is calculated. By comparing and evaluating the real-time updated adjustment trigger decision function Jctrl with the trigger threshold Cth for a second time, it is possible to accurately determine whether the current posture state exceeds the controllable boundary, thereby triggering the corresponding control logic: if the adjustment trigger decision function Jctrl is less than or equal to the trigger threshold Cth, the first control logic is triggered, and only the wheel speed is adjusted symmetrically and the lateral micro-pushing action is performed to maintain stable operation and avoid energy waste; if the adjustment trigger decision function Jctrl exceeds the trigger threshold Cth, it is determined that the robot has entered a high-disturbance uncontrollable state, and the second control logic is immediately triggered to activate the self-stabilizing bracket and actively reduce the speed by 50% to quickly restore the posture and suppress the disturbance effect. Through the implementation of step S5, the present invention successfully realizes the dynamic coupling and hierarchical response control mechanism among disturbance trend modeling, structural stability assessment and real-time control capability matching judgment. This mechanism not only has adaptive capabilities, predictive adjustment capabilities and energy consumption optimization capabilities, but also ensures the robot's continuous operation safety and path stability in an environment with severe thermal fluctuations.

[0150] Example 7

[0151] Please refer to Figure 1 and Figure 2 , an intelligent control system for an industrial robot, including a sensing and acquisition module, a sensing and processing module, a disturbance analysis module, an attitude analysis module, and an integrated control module;

[0152] The sensing and acquisition module sets acquisition points on the pipeline cleaning robot, constructs a heat flux disturbance sensor array, and real-time collects heat disturbance data through the heat capacity density sensor array, and then transmits the heat disturbance data to the central control server through a wired direct connection data line;

[0153] The sensing and processing module extracts features from the heat disturbance data in the central control server to obtain a heat disturbance feature set, and preprocesses the heat disturbance feature set to obtain a standardized digital set;

[0154] The disturbance analysis module calculates and outputs the heat disturbance aggregation trend index Rtrend based on the standardized digital set, sets a trend interval threshold for a preliminary comparison and evaluation with the heat disturbance aggregation trend index Rtrend, and triggers a self-stabilizing control mechanism based on the preliminary comparison and evaluation results;

[0155] The attitude analysis module triggers dynamic attitude adjustment based on the preliminary comparison and evaluation, and calculates and outputs the attitude offset angle Oproj by extracting the structural stability index Estab of the current robot in combination with the heat disturbance aggregation trend index Rtrend;

[0156] The integrated control module comprehensively calculates the heat disturbance aggregation trend index Rtrend and the attitude offset angle Oproj to output an adjustment trigger decision function Jctrl, sets a trigger threshold Cth for a secondary comparison and evaluation, and triggers the control logic.

[0157] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

Claims

1. An intelligent control method for an industrial robot, characterized in that: The following steps are involved: S1. Set up collection points on the pipeline cleaning robot, build a thermal flow disturbance sensor array, and collect thermal disturbance data in real time through a thermal capacity density sensor array, and then transmit the thermal disturbance data to the central control server through a wired direct data cable; S2. Extracting features of the thermal disturbance data in the central control server to obtain a thermal disturbance feature set, and preprocessing the thermal disturbance feature set to obtain a standardized digital set; S3. Calculate and output the thermal disturbance aggregation trend index Rtrend based on the standardized digital set, set the trend interval threshold and the thermal disturbance aggregation trend index Rtrend for preliminary comparative evaluation, and trigger the self-stabilizing control mechanism based on the preliminary comparative evaluation results; S4, triggering dynamic posture adjustment based on preliminary comparative evaluation, by extracting the structural stability index Estab of the current robot and combining it with the thermal disturbance aggregation trend index Rtrend to calculate and output the posture deviation angle Oproj; S5. Comprehensively calculate the thermal disturbance aggregation trend index Rtrend and the attitude deviation angle Oproj to output the adjustment trigger decision function Jctrl, and set the trigger threshold Cth for secondary comparison and evaluation to trigger the control logic.

2. The intelligent control method for an industrial robot according to claim 1, wherein: Said S1 includes S11 and S12; S11, setting collection points at symmetrical positions of the front, back, left and right of the pipeline cleaning robot, and installing a thermal flow disturbance sensor array in each collection point to collect thermal disturbance data of each collection point in real time, wherein the number of thermal flow disturbance sensor arrays set is greater than or equal to 4 groups; The thermal flow disturbance sensor array includes an array-type MEMS thermal sensor, a TOF ultrasonic microwave transmitter, a TOF ultrasonic microwave receiver and an infrared thermal imaging sensor; The thermal disturbance data includes the temperature T at the i-th acquisition point i , the return signal, the main direction vector vflow of heat convection, and the forward direction vector vmove of the robot; S12. The thermal disturbance data collected in real time is transmitted to the central control server of the pipeline cleaning robot through a data transmission line directly connected to the back end of the pipeline cleaning robot.

3. The intelligent control method for an industrial robot according to claim 2, wherein: The S2 includes S21 and S22; S21. Extracting features of thermal disturbance data in the central control server to obtain a thermal disturbance feature set; The thermal disturbance feature set includes the thermal gradient Rtgrad(i) of the i-th acquisition point, the disturbance vortex density Pvortex(i) of the i-th acquisition point, and the convection direction deviation Ddiv(i) of the i-th acquisition point; The thermal gradient Rtgrad is calculated and extracted by dividing the temperature T difference between each pair of arrayed MEMS thermal sensors by the distance between the sensor pairs; The disturbance vortex density Pvortex is obtained by transmitting ultrasonic pulses through a TOF ultrasonic microwave transmitter, and then receiving the return signal through a TOF ultrasonic microwave receiver. The disturbance waveform spectrum of the return signal is extracted using a fast Fourier transform (FFT). Based on the disturbance waveform spectrum and by setting a disturbance peak determination rule, the proportion of the number of disturbance events in the statistical unit time window in the spatial region is derived. The convection direction deviation Ddiv is extracted by performing arccos calculation based on the main direction vector vflow of the thermal convection and the forward direction vector vmove of the robot; S22, preprocessing the thermal disturbance feature set to obtain a standardized digital set; The preprocessing includes outlier detection and normalization processing; The outlier detection uses the IQR interquartile range method to remove extreme fluctuations in the thermal disturbance feature set, and then uses the mean vector method to complement the data of the thermal disturbance feature set at the removed positions; The normalization processing uses the Min-Max normalization method for the thermal disturbance feature set after outlier detection to normalize physical quantities with different dimensions and scales into the interval [0, 1], so as to eliminate the influence of dimensions.

4. An intelligent control method for an industrial robot according to claim 3, characterized in that: The S3 includes S31 and S32; S31: Based on the standardized digital set obtained from all acquisition points, calculate and output the disturbance aggregation trend index Rtrend to share the degree of thermal disturbance focusing, the degree of direction deflection, and the spatial complexity within the acquisition points; The disturbance aggregation trend index Rtrend is calculated and output through the following algorithm formula; ； In the formula, N represents the total number of acquisition points, and sin represents the sine function.

5. The intelligent control method for an industrial robot according to claim 4, wherein: S32: Set the trend interval threshold through the experimental calibration method. The trend interval threshold includes the first trend threshold F1 and the second trend threshold F2. Compare and evaluate the disturbance aggregation trend index Rtrend obtained in real time with the trend interval threshold to judge the temperature disturbance, vortex density, and deflection direction of the current pipeline cleaning robot, and perform regional division based on the preliminary comparison and evaluation results. The specific evaluation content is as follows; When the disturbance aggregation trend index Rtrend < the first trend threshold F1, it indicates that the disturbance is dispersed. At this time, it is divided into the stable flow area and the current operating state is continued; When the first trend threshold F1 ≤ the disturbance aggregation trend index Rtrend < the second trend threshold F2, it indicates that the disturbance in the current area is aggregated. At this time, it is divided into the controllable disturbance area and the attitude warning compensation mode is started; In the attitude warning compensation mode, the standardized digital set of the next cycle is extracted, and the disturbance aggregation trend index Rtrend is calculated and output after secondary calculation, and a secondary preliminary evaluation is performed. When the first trend threshold F1 ≤ the disturbance aggregation trend index Rtrend < the second trend threshold F2 still holds, the current area is divided into the disturbance abnormal area at this time, and dynamic attitude adjustment is triggered; When the disturbance aggregation trend index Rtrend ≥ the second trend threshold F2, it indicates that the thermal disturbance vortices are concentrated and the direction skew is obvious. At this time, it is divided into the disturbance abnormal area and dynamic attitude adjustment is triggered.

6. The intelligent control method for an industrial robot according to claim 1, characterized in that: The S4 includes S41 and S42; S41: After the dynamic attitude adjustment is triggered in the preliminary comparison and evaluation, start the three-axis gyroscope, acceleration IMU module, and anti-seismic IMU of the pipeline cleaning robot to collect the attitude data of the robot in real time. After normalizing the attitude data, calculate and extract the structural stability index Estab; The attitude data includes the rotational inertia Irot of the robot around the center, the current angular velocity w, and the vibration impact force Fvib; The structural stability index Estab is calculated and output through the following algorithm formula; ； In the formula, Mbase represents the self-weight of the robot, and g represents the acceleration due to gravity, both of which are initially set by the user and are dimensionless.

7. An intelligent control method for an industrial robot according to claim 6, characterized in that: S42. Based on the obtained structure stability index Estab and the thermal disturbance aggregation trend index Rtrend, calculate and output the attitude offset angle Oproj, and analyze the maximum attitude offset angle of the pipeline cleaning robot due to disturbance effects; The attitude offset angle Oproj is calculated and output through the following algorithm formula; ； In the formula, represents the disturbance amplification factor, which is dimensionless and is set by the user according to the sensitivity of the robot structure to the disturbance trend.

8. An intelligent control method for an industrial robot according to claim 6, characterized in that: The S5 includes S51 and S52; S51. Based on the obtained attitude offset angle Oproj and the disturbance aggregation trend index Rtrend, couple the self-adjusting ability with the trend development speed of the current external disturbance, and comprehensively calculate and output the adjustment trigger decision function Jctrl to measure the attitude situation of the current pipeline cleaning robot during operation; The adjustment trigger decision function Jctrl is calculated and output through the following algorithm formula; ； where wcomp represents the compensation rate, with a dimensionless value, d represents the integral function, and dt represents the time integral function. represents an extremely small number. represents the growth rate of the disturbance trend. represents the weight factor of the growth rate of the disturbance trend, with a dimensionless value, which is set by the user.

9. An intelligent control method for an industrial robot according to claim 8, characterized in that: S52. By extracting the maximum compensation rate wcomp for the differential control of the robot crawler max , and at the same time, based on the attitude offset angle Oproj under the maximum controllable disturbance max , the maximum compensation rate wcomp max and the attitude offset angle Oproj max are used to calculate the ratio, and a disturbance development speed superposition factor is added to set the trigger threshold Cth. Then, the real-time obtained adjustment trigger decision function Jctrl is compared and evaluated with the trigger threshold Cth for the second time to analyze the current attitude operation of the pipeline cleaning robot, and the trigger control logic is carried out according to the results of the second comparison and evaluation. The specific evaluation content is as follows; When the adjustment trigger decision function Jctrl ≤ the trigger threshold Cth, it indicates that the current pipeline cleaning robot has the ability of self-stabilization, and at this time, the first control logic is triggered; When the adjustment trigger decision function Jctrl > the trigger threshold Cth, it indicates that the disturbance development of the current Hiroshima cleaning robot exceeds the upper limit of the control ability, and at this time, the second control logic is triggered; The first control logic symmetrically adjusts the wheel speeds of the crawlers of the pipeline cleaning robot to maintain forward stability, and at the same time activates the lateral micro-pushing state without activating the high-energy-consuming device; The second control logic automatically activates the self-stabilizing bracket of the pipeline cleaning robot and decelerates by 50%.

10. An intelligent control system for an industrial robot, which is applied to an intelligent control method for an industrial robot according to any one of claims 1-9, and is characterized in that: It includes a perception acquisition module, a perception processing module, a disturbance analysis module, an attitude analysis module, and a comprehensive control module; The perception acquisition module sets acquisition points on the pipeline cleaning robot to construct a thermal flow disturbance sensor array, and real-time collects thermal disturbance data through the heat capacity density sensor array, and then transmits the thermal disturbance data to the central control server through a wired direct connection data cable; The perception processing module extracts features from the thermal disturbance data in the central control server to obtain a thermal disturbance feature set, and preprocesses the thermal disturbance feature set to obtain a standardized digital set; The disturbance analysis module calculates and outputs the thermal disturbance aggregation trend index Rtrend based on the standardized digital set, sets a trend interval threshold to conduct a preliminary comparison and evaluation with the thermal disturbance aggregation trend index Rtrend, and triggers the self-stabilization control mechanism based on the preliminary comparison and evaluation results; The attitude analysis module triggers dynamic attitude adjustment based on the preliminary comparison and evaluation. By extracting the structure stability index Estab of the current robot and combining it with the thermal disturbance aggregation trend index Rtrend, calculate and output the attitude offset angle Oproj; The comprehensive control module comprehensively calculates and outputs the adjustment trigger decision function Jctrl by combining the thermal disturbance aggregation trend index Rtrend and the attitude offset angle Oproj, and sets a trigger threshold Cth to conduct a secondary comparison and evaluation to trigger the control logic.

Images