A Crane Intelligent Load Control Method Based on Working Condition Adaptation

Abstract

This invention relates to the field of crane control technology, and more particularly to an intelligent load control method for cranes based on adaptive operating conditions. The method, based on an AI adaptive control model, executes the following steps: Step 1, building an operating condition parameter acquisition system; Step 2, designing an AI adaptive operating condition recognition model; Step 3, establishing an operating condition-control parameter correlation model; Step 4, building a load closed-loop control module; Step 5, setting a safety early warning mechanism. This invention achieves real-time and accurate identification of no-load, light-load, heavy-load, and high-wind operating conditions, and also realizes dynamic optimization of control parameters, balancing safety and energy efficiency. It can also provide real-time feedback to correct control parameters and, combined with Miner's linear damage accumulation theory, extends equipment lifespan. Furthermore, it can provide timely early warnings and adjust control strategies, improving operational safety. It is adaptable to various types of cranes, effectively improving operational efficiency and intelligence.

Description

Technical Field

[0001] This invention relates to the field of crane control technology, and in particular to a crane intelligent load control method based on working condition adaptation. Background Technology

[0002] Cranes, as core equipment for heavy material handling, are widely used in ports, construction, metallurgy, warehousing and other fields. Their operating conditions are complex and varied, mainly covering various scenarios such as no-load, light-load, heavy-load, and high-wind conditions. The load characteristics, stress state and environmental interference of cranes vary significantly under different operating conditions, which places high demands on the real-time adaptation of control parameters.

[0003] Currently, traditional cranes mostly adopt a fixed parameter control mode, with preset fixed lifting speed, luffing speed, and braking parameters, which do not adjust with changes in working conditions, resulting in the following drawbacks:

[0004] First, poor adaptability: High-speed lifting under heavy load easily leads to overload and instability, while using heavy-load braking parameters under no-load conditions results in wasted energy. Second, high safety risks: Fixed parameters cannot respond promptly to sudden changes in operating conditions, easily causing accidents such as swaying and overturning. Third, high energy consumption: Maintaining heavy-load energy consumption levels under no-load and light-load conditions results in low energy utilization. Fourth, low control precision: It is impossible to accurately match load changes with control parameters, affecting operational efficiency and equipment lifespan.

[0005] Furthermore, a search of existing technologies reveals that the technical solutions related to crane load control can be mainly categorized into the following three types:

[0006] The first category is simple operating condition differentiation control technology (such as the technology disclosed in application numbers CN202410567890.1 and CN202310876543.2), which can only roughly distinguish between no-load and heavy-load conditions, does not cover complex operating conditions such as light-load and strong wind, relies on manually preset thresholds, cannot achieve real-time accurate identification and dynamic adaptive adjustment, and does not involve energy consumption quantification optimization.

[0007] The second category is single AI algorithm operating condition identification and control technology (such as the technologies disclosed in application numbers CN202410345678.9 and CN202310987654.8), which adopts a single CNN or LSTM algorithm, can only extract spatial features or time series features, and the identification accuracy is less than 95%. It does not establish a quantitative correlation model between operating conditions and energy consumption and safety parameters.

[0008] The third category is crane energy-saving control technology (such as the technologies disclosed in application numbers CN202410123456.7 and CN202310765432.1), which only optimizes no-load and heavy-load working conditions, does not take into account the working conditions of strong winds, and has no closed-loop control and safety early warning mechanism, and does not introduce a model self-learning mechanism.

[0009] It is evident that existing technologies have not yet solved the technical problems of accurate identification of multiple operating conditions, dynamic adaptive adjustment of control parameters, and synergistic optimization of safety and energy efficiency. Summary of the Invention

[0010] The main objective of this invention is to overcome the problems of fixed control parameters, poor adaptability to operating conditions, high safety risks, and high energy consumption in existing crane technologies. It provides a crane intelligent load control method based on adaptive operating conditions. This method designs an AI adaptive control model to accurately identify the crane's operating conditions in real time, automatically and dynamically adjusts the hoisting speed, luffing speed, and braking parameters, achieving adaptive load control. This effectively avoids safety risks such as overload and instability, while reducing energy consumption by more than 15%, thus improving the safety, stability, efficiency, and energy saving of crane operations.

[0011] The technical solution adopted by this invention to achieve its technical objective is: a crane intelligent load control method based on adaptive operating conditions, characterized in that the method performs the following steps based on an AI adaptive control model:

[0012] Step 1: Build a working condition parameter acquisition system to collect various working condition parameters during crane operation in real time and construct a working condition parameter dataset.

[0013] Step 2: Based on the collected operating parameters, design an AI adaptive operating condition recognition model to achieve real-time and accurate recognition of no-load, light-load, heavy-load, and high-wind operating conditions.

[0014] Step 3: Establish a working condition-control parameter correlation model, and combine it with the energy consumption optimization target to construct an adaptive adjustment formula for control parameters, so as to realize the dynamic optimization adjustment of hoisting speed, luffing speed and braking parameters;

[0015] Step 4: Build a load closed-loop control module to provide real-time feedback on the load control effect, dynamically correct control parameters, and ensure control accuracy and stability.

[0016] Step 5: Set up a safety early warning mechanism. Based on the operating condition identification results and load control parameters, assess safety risks in real time, issue early warnings in a timely manner, and adjust control strategies accordingly.

[0017] Preferably, in step 1, the operating condition parameters collected in real time by the operating condition parameter acquisition system include: the actual load mass of the crane. Operating environment wind speed Lifting speed Amplitude variation speed boom angle Drive motor current Braking time ;

[0018] Define the rated load capacity of the crane as The load quantization formula is as follows:

[0019] ;

[0020] in, For load factor, For actual load quality, The rated load mass of the crane is 1.1, which is the dynamic load factor.

[0021] Preliminary load condition classification criteria:

[0022] when At that time, it was in no-load condition;

[0023] when At that time, it was a light load condition;

[0024] when At that time, it was a heavy-load operating condition;

[0025] when When this occurs, it indicates an overload state (triggering a safety warning).

[0026] At the same time, define the wind speed level coefficient. The formula used to quantify strong wind conditions is as follows:

[0027] ;

[0028] in, This refers to the actual ambient wind speed. The critical wind speed for safe operation of cranes;

[0029] when When the wind condition is determined to be a strong wind condition, a special control strategy for strong wind conditions needs to be activated.

[0030] Preferably, in step 2, the AI ​​adaptive working condition recognition model adopts a CNN-LSTM fusion neural network model, combined with the working condition parameters collected in real time in step 1, to accurately identify the working condition, specifically as follows:

[0031] Step 2.1, Dataset Preprocessing: Normalize the collected operating parameters to eliminate the influence of dimensions. The normalization formula is as follows:

[0032] ;

[0033] in, These are the normalized parameter values. These are the original parameter values. This is the maximum value of the parameter. This is the minimum value of the parameter.

[0034] Step 2.2, Model Structure Design: The CNN module is used to extract the spatial features of the working condition parameters, and the LSTM module is used to extract the time series features of the working condition parameters. The two are combined to achieve comprehensive extraction of multi-dimensional working condition features.

[0035] Step 2.3, Model Training and Output: Using the preprocessed operating condition parameters as input and the no-load, light-load, heavy-load, and high-wind operating conditions as output labels, train the CNN-LSTM model. The cross-entropy loss function is used during training. The cross-entropy loss function formula for optimizing model parameters is as follows:

[0036] ;

[0037] in, The number of training samples. For the first The first sample Labels for various working conditions For the model to predict the first The sample is the first The probability of a certain working condition;

[0038] Step 2.4, Working Condition Identification and Judgment: The model outputs the working condition identification probability. When the identification probability is ≥95%, it is determined to be the current working condition; when the identification probability is <95%, the fuzzy identification supplementary strategy is activated, combined with the load coefficient. and wind speed rating coefficient The working conditions are further determined by fuzzy membership functions, which are as follows:

[0039] Membership in no-load condition: ;

[0040] Membership in light load condition: ;

[0041] Membership in heavy-duty conditions: ;

[0042] Membership of strong wind conditions: ;

[0043] The working condition corresponding to the maximum membership degree is taken as the current working condition to ensure the accuracy and reliability of working condition identification. The identification response time is ≤0.5s to meet the real-time control requirements, and the identification accuracy is ≥98%.

[0044] Preferably, in step 3, the working condition-control parameter correlation model constructs adaptive adjustment formulas for the hoisting speed, luffing speed, and braking parameters, as follows:

[0045] Step 3.1, Adaptive adjustment formula for lifting speed: combined with load factor Wind speed rating coefficient Lifting speed The formula is adjusted as follows:

[0046] ;

[0047] in, The rated lifting speed of the crane. For load factor, This is the wind speed rating coefficient. Energy consumption optimization coefficient;

[0048] And the following constraints are met: ,in This is the minimum lifting speed.

[0049] Step 3.2, Adaptive adjustment formula for luffing speed: combined with boom angle Load factor Amplitude variation speed The formula is adjusted as follows:

[0050] ;

[0051] in, For the rated variable speed, For the boom angle, Used to correct the effect of boom angle on amplitude stability;

[0052] And the following constraints are met: ,in This is the minimum amplitude speed.

[0053] Step 3.3, Adaptive Adjustment Formula for Braking Parameters: Braking parameters include braking pressure. and braking time Combined with load factor Lifting speed Wind speed rating coefficient To achieve precise matching of braking parameters, the formula is as follows:

[0054] Brake pressure adjustment formula: ;

[0055] Braking time adjustment formula: ;

[0056] in, Rated braking pressure, This is the rated braking time;

[0057] And the following constraints are met: , ,in , These are the minimum and maximum braking pressures, respectively. , These are the minimum and maximum braking times, respectively.

[0058] 5. The intelligent load control method for cranes based on adaptive operating conditions according to claim 4, characterized in that step 3 further includes:

[0059] Step 3.4, Energy Consumption Optimization Verification Formula: Construct an energy consumption comparison formula, as follows:

[0060] ;

[0061] in, For energy consumption reduction rate, Energy consumption under traditional fixed parameter control mode This refers to the energy consumption under the AI ​​adaptive control mode of this invention.

[0062] when Automatically adjust the energy consumption optimization coefficient. Optimize the control parameters again until they meet the requirements. ;

[0063] Crane operation energy consumption The calculation formula is as follows:

[0064] ;

[0065] in, For homework time, For the real-time power of the drive motor, It is the acceleration due to gravity. For frictional resistance, For motor efficiency.

[0066] 6. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, in step 4, the load closed-loop control module adopts a PID control algorithm, and dynamically corrects the control parameters in combination with real-time feedback load parameters. The PID control algorithm formula is as follows:

[0067] ;

[0068] in, This is the control correction value output by the PID controller. This is the proportionality coefficient. The integral coefficient is... The differential coefficients are... To control deviation, Control deviations at different times;

[0069] By using a PID control algorithm, the hoisting speed, luffing speed, and braking parameters are corrected in real time to reduce control deviation. Based on Miner's linear damage accumulation theory, the damage accumulation formula is as follows:

[0070] ;

[0071] in, This represents the cumulative damage value. For the first Number of control adjustments under various operating conditions For the first The maximum number of adjustments allowed for the equipment under certain operating conditions, when At that time, reduce the frequency of control parameter adjustments.

[0072] Preferably, in step 5, the safety early warning mechanism is based on the operating condition identification result and control parameters, and the determination mechanism of the safety early warning mechanism is as follows:

[0073] Overload warning determination: When At that time, an overload warning is triggered, and the lifting speed is automatically adjusted to... Braking pressure adjusted to Stop lifting operations until the load is restored to a safe range;

[0074] Instability Early Warning Judgment: Based on the force analysis of the boom, an instability judgment formula is constructed. When... At that time, an instability warning was triggered, among which For the actual force on the boom, The maximum allowable force on the boom is given; the actual force calculation formula for the boom is as follows:

[0075] ;

[0076] in, For load suspension length, This refers to the effective length of the boom;

[0077] Gale instability warning: When When the strong wind instability warning is triggered, the luffing operation is automatically stopped, the boom is adjusted to a safe angle, and the lifting speed is reduced to [missing information]. .

[0078] Preferably, the AI ​​adaptive control model further includes a model self-learning module, which collects operating condition identification data and control effect data in real time, and updates the model parameters through a gradient descent algorithm. The gradient descent algorithm update formula is as follows:

[0079] ;

[0080] in, For the updated model parameters, These are the current model parameters. For learning rate, This is the partial derivative of the loss function with respect to the current model parameters.

[0081] The working principle of this intelligent load control method for cranes based on adaptive operating conditions is as follows: Crane operating condition parameters are collected in real time by multiple sensors. A CNN-LSTM fusion neural network model is used to identify the current operating condition (no load / light load / heavy load / strong wind), establishing a correlation model between operating condition parameters and control parameters. Energy consumption optimization coefficients, boom angle correction, and constraints are introduced to achieve adaptive dynamic adjustment of control parameters. Simultaneously, PID closed-loop control is used to correct deviations in real time. Miner's linear damage accumulation theory is combined to optimize equipment fatigue life. Furthermore, a multi-level safety early warning mechanism is constructed based on load coefficients, boom stress analysis, and wind speed levels, achieving intelligent control and safety protection across all operating conditions during crane operation.

[0082] Compared with the prior art, the beneficial effects of the present invention are:

[0083] This invention employs a CNN-LSTM fusion AI model, combined with multi-dimensional operating condition parameters and quantization formulas, to achieve real-time and accurate identification of no-load, light-load, heavy-load, and high-wind operating conditions, with an accuracy rate of ≥98%. This solves the problems of single-dimensional feature extraction and low identification accuracy of existing technologies.

[0084] This invention constructs a complete working condition-control parameter correlation model, designs adaptive adjustment formulas for hoisting speed, luffing speed, and braking parameters, and introduces corrections based on boom angle and wind speed level coefficients to achieve dynamic optimization of control parameters, balancing safety and energy efficiency, and reducing energy consumption by more than 15%.

[0085] This invention introduces PID closed-loop control and a model self-learning mechanism to provide real-time feedback and correct control parameters, achieving a control deviation of ≤±5%. It also optimizes the adjustment frequency by combining Miner's linear damage accumulation theory, thereby extending the service life of the equipment.

[0086] This invention establishes a safety early warning mechanism, uses quantitative formulas to determine overload and instability risks in real time, and combines boom stress analysis to construct an instability early warning formula, thereby providing timely warnings and adjusting control strategies to improve operational safety.

[0087] The method of the present invention is highly versatile and can be adapted to various types of cranes. It does not require large-scale modification, is easy to promote and apply, and can effectively improve work efficiency and intelligence level. Attached Figure Description

[0088] Figure 1 This is a flowchart of the steps in a crane intelligent load control method based on adaptive operating conditions. Detailed Implementation

[0089] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0090] In the description of this invention, it should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to or indirectly connected to the other element.

[0091] In the description of this invention, it should be noted that the terms "center," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0092] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0093] Example 1:

[0094] Please see Figure 1A crane intelligent load control method based on adaptive operating conditions is disclosed. The method executes steps 1 to 5 based on an AI adaptive control model. Specifically:

[0095] Step 1: Build a working condition parameter acquisition system to collect various working condition parameters during crane operation in real time and construct a working condition parameter dataset.

[0096] In step 1, the operating condition parameter acquisition system includes a load sensor, a wind speed sensor, a speed sensor, an angle sensor, and a current sensor. The operating condition parameters acquired in real time include: the actual load mass of the crane. Operating environment wind speed Lifting speed Amplitude variation speed boom angle Drive motor current Braking time The sampling frequency is 10-50Hz to ensure the real-time and completeness of parameter acquisition.

[0097] Define the rated load capacity of the crane as Based on the dynamic load coefficient requirements in GB / T3811-2008 "Code for Design of Cranes" and combined with the actual operating load, the load conditions are first preliminarily quantified and classified to provide a basis for AI condition recognition. The load quantification formula is as follows:

[0098] ;

[0099] in, For load factor, The actual load mass (kg). The rated load capacity of the crane is 1.1, which is the dynamic load factor to accommodate the dynamic load impact during lifting.

[0100] Preliminary load condition classification criteria:

[0101] when At that time, it was in no-load condition;

[0102] when At that time, it was a light load condition;

[0103] when At that time, it was a heavy-load operating condition;

[0104] when When this occurs, it indicates an overload state (triggering a safety warning).

[0105] At the same time, define the wind speed level coefficient. The formula used to quantify strong wind conditions is as follows:

[0106] ;

[0107] in, The actual ambient wind speed (m / s) Critical wind speed for safe operation of cranes ( The speed is typically set to 8-13.8 m / s, depending on the crane model.

[0108] when When the wind condition is determined to be a strong wind condition, a special control strategy for strong wind conditions needs to be activated.

[0109] Step 2: Based on the collected operating parameters, design an AI adaptive operating condition recognition model to achieve real-time and accurate recognition of no-load, light-load, heavy-load, and high-wind operating conditions.

[0110] In step 2, the AI ​​adaptive working condition recognition model uses a CNN-LSTM fusion neural network model, combined with the working condition parameters collected in real time in step 1, to accurately identify the working condition (distinguishing itself from the shortcomings of existing single-algorithm technologies), specifically as follows:

[0111] Step 2.1, Dataset Preprocessing: Normalize the collected operating parameters to eliminate the influence of dimensions. The normalization formula is as follows:

[0112] ;

[0113] in, These are the normalized parameter values. These are the original parameter values. This is the maximum value of the parameter. This is the minimum value of the parameter.

[0114] Step 2.2, Model Structure Design: The CNN module is used to extract the spatial features of the operating parameters (such as the correlation features between load and wind speed, and the coupling features between current and speed), and the LSTM module is used to extract the time series features of the operating parameters (such as the load change trend and wind speed fluctuation features). The two are integrated to achieve comprehensive extraction of multi-dimensional operating features (which cannot be achieved by a single algorithm in the current technology), thereby improving the recognition accuracy.

[0115] Step 2.3, Model Training and Output: Using the preprocessed operating condition parameters as input and the no-load, light-load, heavy-load, and high-wind operating conditions as output labels, train the CNN-LSTM model. The cross-entropy loss function is used during training. The cross-entropy loss function formula for optimizing model parameters is as follows:

[0116] ;

[0117] in, The number of training samples. For the first The first sample Labels for different operating conditions (0 or 1, one-hot encoding). For the model to predict the first The sample is the first The probability of a certain working condition;

[0118] Step 2.4, Working Condition Identification and Judgment: The model outputs the working condition identification probability. When the identification probability of a certain working condition is ≥95%, it is determined to be the current working condition; when the identification probability is <95%, the fuzzy identification supplementary strategy is activated, combined with the load coefficient. and wind speed rating coefficient The working conditions are further determined by fuzzy membership functions, which are as follows:

[0119] Membership in no-load condition: ;

[0120] Membership in light load condition: ;

[0121] Membership in heavy-duty conditions: ;

[0122] Membership of strong wind conditions: ;

[0123] The working condition corresponding to the maximum membership degree is taken as the current working condition to ensure the accuracy and reliability of working condition identification. The identification response time is ≤0.5s, which meets the real-time control requirements, and the identification accuracy is ≥98% (higher than the identification accuracy of existing single algorithms).

[0124] Step 3: Establish a working condition-control parameter correlation model, and combine it with the energy consumption optimization target to construct an adaptive adjustment formula for control parameters, so as to realize the dynamic optimization adjustment of lifting speed, luffing speed and braking parameters.

[0125] In step 3, the working condition-control parameter correlation model, combined with the energy consumption optimization objective, constructs adaptive adjustment formulas for the hoisting speed, luffing speed, and braking parameters respectively (distinguishing itself from the shortcomings of existing technologies that lack quantification formulas and do not incorporate multi-parameter collaborative optimization), thereby achieving dynamic optimization of the control parameters, as detailed in the following steps:

[0126] Step 3.1, Adaptive adjustment formula for lifting speed: combined with load factor Wind speed rating coefficient Balancing lifting efficiency, safety, and energy consumption, lifting speed The formula is adjusted as follows:

[0127] ;

[0128] in, The rated lifting speed of the crane (m / min). For load factor, This is the wind speed rating coefficient. This is the energy consumption optimization coefficient (range 0.85-0.95), used to further reduce energy consumption;

[0129] And the following constraints are met: ,in Minimum lifting speed ( The speed is set to a preset value, typically 0.5-1 m / min, to avoid the speed being too low and affecting work efficiency, while also preventing the speed from being too high and causing safety risks.

[0130] Step 3.2, Adaptive adjustment formula for luffing speed: combined with boom angle Load factor Amplitude variation speed The formula is adjusted as follows:

[0131] ;

[0132] in, The rated variable speed (m / min). The boom angle (range 30°-80°). This is used to correct the effect of the boom angle on the luffing stability. The smaller the boom angle, the slower the luffing speed, thus avoiding boom swaying (current technology does not introduce boom angle correction).

[0133] And the following constraints are met: ,in Minimum amplitude speed ( The preset speed is typically set to 0.3-0.8 m / min.

[0134] Step 3.3, Adaptive Adjustment Formula for Braking Parameters: Braking parameters include braking pressure. and braking time Combined with load factor Lifting speed Wind speed rating coefficient To achieve precise matching of braking parameters, the formula is as follows:

[0135] Brake pressure adjustment formula: ;

[0136] Braking time adjustment formula: ;

[0137] in, Rated braking pressure (MPa). The rated braking time (s) is determined by the load, speed, and wind speed. The greater the load, the faster the speed, the greater the wind speed, the greater the braking pressure, and the longer the braking time. This ensures smooth braking and avoids load swaying or equipment damage caused by braking shock.

[0138] And the following constraints are met: , ,in, , These are the minimum and maximum braking pressures, respectively. , These are the minimum and maximum braking times, respectively. , , ).

[0139] Step 3.4, Energy Consumption Optimization Verification Formula: To ensure that energy consumption is reduced by more than 15%, an energy consumption comparison formula is constructed to verify the energy consumption optimization effect in real time. The formula is as follows:

[0140] ;

[0141] in, Energy consumption reduction rate (%) Energy consumption (kWh) under traditional fixed parameter control mode. Energy consumption (kWh) under the AI ​​adaptive control mode of this invention;

[0142] when Automatically adjust the energy consumption optimization coefficient. Optimize the control parameters again until they meet the requirements. (The existing technology does not clearly define the energy consumption optimization target and the quantitative verification method).

[0143] Crane operation energy consumption The calculation formula is as follows (considering the drive motor power and operating time):

[0144] ;

[0145] in, The time for the task is in hours (h). This represents the real-time power (kW) of the drive motor. The acceleration due to gravity is taken as 9.8 m / s². Frictional resistance (N, (Based on crane model preset) The value is the motor efficiency (range 0.8-0.9).

[0146] Step 4: Build a load closed-loop control module to provide real-time feedback on the load control effect, dynamically correct control parameters, and ensure control accuracy and stability.

[0147] In step 4, the load closed-loop control module uses a PID control algorithm, combined with real-time feedback load parameters, to dynamically correct control parameters and ensure control accuracy. The PID control algorithm formula is as follows:

[0148] ;

[0149] in, This is the control correction value output by the PID controller. This is the proportionality coefficient. The integral coefficient is... The differential coefficients are... To control the deviation (i.e., the difference between the target load control parameters and the actual load control parameters). Control deviations at different times;

[0150] By using a PID control algorithm, the hoisting speed, luffing speed, and braking parameters are corrected in real time to reduce control deviation. This ensures the stability and accuracy of load control, while also incorporating Miner's linear damage accumulation theory to consider equipment fatigue life and avoid damage caused by frequent adjustments (current technologies do not incorporate equipment fatigue life optimization). The damage accumulation formula is as follows:

[0151] ;

[0152] in, This represents the cumulative damage value. For the first Number of control adjustments under various operating conditions For the first The maximum number of adjustments allowed for the equipment under certain operating conditions, when In this case, reduce the frequency of control parameter adjustments to extend the service life of the equipment.

[0153] Step 5: Set up a safety early warning mechanism. Based on the operating condition identification results and load control parameters, assess safety risks in real time, issue early warnings in a timely manner, and adjust control strategies accordingly.

[0154] In step 5, the safety early warning mechanism, based on the operating condition identification results and control parameters, combined with safety thresholds, judges safety risks such as overload and instability in real time. The judgment mechanism of the safety early warning mechanism is as follows (different from the shortcomings of existing technologies, such as lack of quantitative judgment of instability and untimely early warning response):

[0155] Overload warning determination: When At that time, an overload warning is triggered, and the lifting speed is automatically adjusted to... Braking pressure adjusted to Stop lifting operations until the load is restored to a safe range;

[0156] Instability Early Warning Judgment: Based on the force analysis of the boom, an instability judgment formula is constructed. When... At that time, an instability warning was triggered, among which For the actual force on the boom, The maximum allowable force on the boom is given; the actual force calculation formula for the boom is as follows:

[0157] ;

[0158] in, The load suspension length is (m). The effective length of the boom (m);

[0159] Gale instability warning: When When a strong wind instability warning is triggered, the boom operation is automatically stopped, the boom is adjusted to a safe angle (≥60°), and the lifting speed is reduced to [missing information]. To ensure equipment safety.

[0160] Furthermore, in this embodiment, the AI ​​adaptive control model also includes a model self-learning module, which collects operating condition identification data and control effect data in real time, and updates the model parameters through the gradient descent algorithm to improve the accuracy of operating condition identification and the optimization effect of control parameters (existing technologies do not introduce a self-learning mechanism and cannot continuously optimize). The gradient descent algorithm update formula is as follows:

[0161] ;

[0162] in, For the updated model parameters, These are the current model parameters. The learning rate (range 0.001-0.01). This is the partial derivative of the loss function with respect to the current model parameters.

[0163] The specific usage process of this crane intelligent load control method based on working condition adaptation is as follows: First, the crane load mass, wind speed, lifting / luffing speed, boom angle, motor current and braking time are collected in real time at a frequency of 10-50Hz. The load coefficient and wind speed level coefficient are calculated to perform preliminary working condition classification.

[0164] The collected data is then input into a CNN-LSTM fusion neural network model for normalization, spatial and temporal feature extraction, and working condition identification. When the identification probability is lower than 95%, a fuzzy membership function is activated to supplement the determination.

[0165] Next, based on the identification results, the hoisting speed, luffing speed, braking pressure and braking time are dynamically adjusted through the working condition-control parameter association model, and it is verified whether the energy consumption reduction rate reaches more than 15%.

[0166] Simultaneously, the PID closed-loop control module provides real-time feedback and corrects control parameter deviations, and combines the Miner damage accumulation value to monitor the equipment fatigue status.

[0167] Finally, based on the load factor, boom stress and wind speed level, the risk of overload, instability and strong wind instability is judged in real time, triggering corresponding early warnings and automatically adjusting the control strategy. At the same time, the model parameters are continuously optimized through gradient descent algorithm.

[0168] Example 2:

[0169] Based on the above embodiments, this embodiment of the intelligent load control method for cranes based on adaptive operating conditions takes a tower crane (rated load) as an example. Taking (e.g., ) as an example, the specific implementation steps are as follows:

[0170] Step 1: Set up a working condition parameter acquisition system, selecting a load sensor (measurement range 0-15t, accuracy ±0.5%), anemometer (measurement range 0-30m / s, accuracy ±0.1m / s), velocity sensor (measurement range 0-10m / min, accuracy ±0.05m / min), angle sensor (measurement range 0-90°, accuracy ±0.1°), and current sensor (measurement range 0-100A, accuracy ±0.5A). Set the acquisition frequency to 30Hz and collect the following working condition parameters in real time: actual load mass. Ambient wind speed Lifting speed Amplitude variation speed boom angle Drive motor current Braking time .

[0171] Preset parameters: Rated lifting speed Rated luffing speed Rated braking pressure Rated braking time Critical wind speed for safe operation Minimum lifting speed Minimum amplitude speed Energy consumption optimization coefficient motor efficiency Frictional resistance Effective length of boom Maximum allowable force on the boom .

[0172] Step 2: Building and training the AI ​​adaptive working condition recognition model:

[0173] Step 2.1 Dataset Preprocessing: Collect 10,000 sets of parameter data under different operating conditions, covering no-load, light-load, heavy-load, and high-wind conditions, and use a normalization formula. Preprocess the data to eliminate the influence of dimensions;

[0174] Step 2.2 Model Training: The CNN module is configured with 3 convolutional layers and 2 pooling layers, and the LSTM module is configured with 2 hidden layers with 64 neurons in each hidden layer. The cross-entropy loss function is used. Learning rate The training iterations were 1000 times, and the model's recognition accuracy reached 98.5% after training (higher than the recognition accuracy of a single algorithm in existing technologies).

[0175] Step 2.3 Example of working condition identification:

[0176] Example 1: Parameter Acquisition , Calculate the load factor Wind speed rating coefficient Model recognition probability: 99.2% for no load, 0.5% for light load, 0.2% for heavy load, and 0.1% for strong wind, which is determined to be a no-load condition;

[0177] Example 2: Parameter Acquisition , Calculate the load factor Wind speed rating coefficient Model recognition probability: 98.7% for heavy load, indicating a heavy load condition;

[0178] Example 3: Parameter Acquisition , Calculate the load factor (0.1 < 0.33 ≤ 0.5), wind speed rating coefficient Model recognition probability: 98.3% for strong winds, indicating a strong wind condition.

[0179] Step 3: Adaptive adjustment of control parameters. Based on the above operating condition example, substitute the parameters into the adjustment formula to calculate the control parameters. Compared with existing single-operating-condition control technologies, the advantages of this invention are highlighted:

[0180] Step 3.1 No-load condition (Example 1):

[0181] Lifting speed: (satisfy );

[0182] Luffing speed (assuming boom angle) ): (satisfy );

[0183] Braking pressure: ;

[0184] Braking time: ;

[0185] Energy consumption calculation: Assuming working time Load suspension length ,but Traditional fixed parameter control of energy consumption Existing technologies control energy consumption under single operating conditions. The energy consumption reduction rate of this invention It is superior to existing technologies (energy consumption reduction rate is only about 8%) and meets the energy consumption optimization target.

[0186] Step 3.2 Heavy-load condition (Example 2):

[0187] Lifting speed: ;

[0188] Luffing speed (assuming boom angle) ): ;

[0189] Braking pressure: ;

[0190] Braking time: ;

[0191] Crane boom stress calculation: There is no risk of instability; however, existing technologies do not perform stress calculations on the boom, which may lead to potential instability.

[0192] Step 3.3 High wind conditions (Example 3):

[0193] Lifting speed: ;

[0194] Luffing speed (assuming boom angle) ): ;

[0195] Braking pressure: ;

[0196] Braking time: ;

[0197] When a strong wind warning is triggered, the boom is automatically adjusted to 75° and the lifting speed is reduced to 0.8m / min to ensure operational safety. Existing technologies do not have specific control strategies designed for strong wind conditions, which can easily lead to swaying and instability.

[0198] Step 4: Load closed-loop control, using PID control algorithm, setting the proportional coefficient. Integral coefficient Differential coefficients Real-time acquisition of the deviation between actual control parameters and target parameters Substitute into the formula The control correction is calculated to keep the control deviation within ±4% to ensure control stability. At the same time, the AI ​​model parameters are updated through the gradient descent algorithm to continuously improve the recognition accuracy. Existing technologies lack a self-learning mechanism, so the recognition accuracy cannot be continuously optimized.

[0199] Step 5: Security warning verification, when collecting parameters At that time, load factor When an overload warning is triggered, the hoisting speed is automatically adjusted to 0.8 m / min, the braking pressure is adjusted to 10 MPa, and the hoisting operation is stopped until the load drops below 10,000 kg, at which point the warning is lifted. The response time of the warning in the prior art is ≥0.8s, while the response time of the present invention is ≤0.5s, making the warning more timely.

[0200] The solution in this embodiment can be selectively combined with solutions in other embodiments.

[0201] It should be noted that although the above embodiments have been described herein, this does not limit the scope of patent protection of this invention. Therefore, any changes and modifications made to the embodiments described herein based on the innovative concept of this invention, or equivalent structural, procedural, or functional transformations made using the description and drawings of this invention, directly or indirectly applying the above technical solutions to other related technical fields, are all included within the scope of protection of this invention.

Claims

1. A crane intelligent load control method based on adaptive operating conditions, characterized in that, The method performs the following steps based on an AI adaptive control model: Step 1: Build a working condition parameter acquisition system to collect various working condition parameters during crane operation in real time and construct a working condition parameter dataset. Step 2: Based on the collected operating parameters, design an AI adaptive operating condition recognition model to achieve real-time and accurate recognition of no-load, light-load, heavy-load, and high-wind operating conditions. Step 3: Establish a working condition-control parameter correlation model, and combine it with the energy consumption optimization target to construct an adaptive adjustment formula for control parameters, so as to realize the dynamic optimization adjustment of hoisting speed, luffing speed and braking parameters; Step 4: Build a load closed-loop control module to provide real-time feedback on the load control effect, dynamically correct control parameters, and ensure control accuracy and stability. Step 5: Set up a safety early warning mechanism. Based on the operating condition identification results and load control parameters, assess safety risks in real time, issue early warnings in a timely manner, and adjust control strategies accordingly.

2. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, In step 1, the operating condition parameters collected in real time by the operating condition parameter acquisition system include: the actual load mass of the crane. Operating environment wind speed Lifting speed Amplitude variation speed boom angle Drive motor current Braking time ; Define the rated load capacity of the crane as The load quantization formula is as follows: ； in, For load factor, For actual load quality, The rated load mass of the crane is 1.1, which is the dynamic load factor. Preliminary load condition classification criteria: when At that time, it was in no-load condition; when At that time, it was a light load condition; when At that time, it was a heavy-load operating condition; when When this occurs, it indicates an overload state (triggers a safety warning). At the same time, define the wind speed level coefficient. The formula used to quantify strong wind conditions is as follows: ； in, This refers to the actual ambient wind speed. The critical wind speed for safe operation of cranes; when When the wind condition is determined to be a strong wind condition, a special control strategy for strong wind conditions needs to be activated.

3. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, In step 2, the AI ​​adaptive working condition recognition model uses a CNN-LSTM fusion neural network model, combined with the working condition parameters collected in real time in step 1, to accurately identify the working condition, as detailed in the following steps: Step 2.1, Dataset Preprocessing: Normalize the collected operating parameters to eliminate the influence of dimensions. The normalization formula is as follows: ； in, These are the normalized parameter values. These are the original parameter values. This is the maximum value of the parameter. This is the minimum value of the parameter; Step 2.2, Model Structure Design: The CNN module is used to extract the spatial features of the working condition parameters, and the LSTM module is used to extract the time series features of the working condition parameters. The two are combined to achieve comprehensive extraction of multi-dimensional working condition features. Step 2.3, Model Training and Output: Using the preprocessed operating condition parameters as input and the no-load, light-load, heavy-load, and high-wind operating conditions as output labels, train the CNN-LSTM model. The cross-entropy loss function is used during training. The cross-entropy loss function formula for optimizing model parameters is as follows: ； in, The number of training samples. For the first The first sample Labels for various working conditions For the model to predict the first The sample is the first The probability of a certain working condition; Step 2.4, Working Condition Identification and Judgment: The model outputs the working condition identification probability. When the identification probability is ≥95%, it is determined to be the current working condition; when the identification probability is <95%, the fuzzy identification supplementary strategy is activated, combined with the load coefficient. and wind speed rating coefficient The working conditions are further determined by fuzzy membership functions, which are as follows: Membership in no-load condition: ; Membership in light load condition: ; Membership in heavy-load conditions: ; Membership of strong wind conditions: ; The working condition corresponding to the maximum membership degree is taken as the current working condition to ensure the accuracy and reliability of working condition identification. The identification response time is ≤0.5s to meet the real-time control requirements, and the identification accuracy is ≥98%.

4. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, In step 3, the working condition-control parameter correlation model constructs adaptive adjustment formulas for hoisting speed, luffing speed, and braking parameters, as detailed in the following steps: Step 3.1, Adaptive adjustment formula for lifting speed: combined with load factor Wind speed rating coefficient Lifting speed The formula is adjusted as follows: ； in, The rated lifting speed of the crane. For load factor, This is the wind speed rating coefficient. Energy consumption optimization coefficient; And the constraints are satisfied: ,in Minimum lifting speed; Step 3.2, Adaptive adjustment formula for luffing speed: combined with boom angle Load factor Amplitude variation speed The formula is adjusted as follows: ； in, For the rated variable speed, For the boom angle, Used to correct the effect of boom angle on amplitude stability; And the constraints are satisfied: ,in Minimum amplitude speed; Step 3.3, Adaptive Adjustment Formula for Braking Parameters: Braking parameters include braking pressure. and braking time Combined with load factor Lifting speed Wind speed rating coefficient To achieve precise matching of braking parameters, the formula is as follows: Brake pressure adjustment formula: ; Braking time adjustment formula: ; in, Rated braking pressure, This is the rated braking time; And the constraints are satisfied: , ,in , These are the minimum and maximum braking pressures, respectively. , These are the minimum and maximum braking times, respectively.

5. The intelligent load control method for cranes based on adaptive operating conditions according to claim 4, characterized in that, Step 3 also includes: Step 3.4, Energy Consumption Optimization Verification Formula: Construct an energy consumption comparison formula, as follows: ； in, For energy consumption reduction rate, Energy consumption under traditional fixed parameter control mode This refers to the energy consumption under the AI ​​adaptive control mode of this invention. when Automatically adjust the energy consumption optimization coefficient. Optimize the control parameters again until they meet the requirements. ; Crane operation energy consumption The calculation formula is as follows: ； in, For homework time, For the real-time power of the drive motor, It is the acceleration due to gravity. For frictional resistance, For motor efficiency.

6. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, In step 4, the load closed-loop control module uses a PID control algorithm, combined with real-time feedback load parameters, to dynamically correct the control parameters. The PID control algorithm formula is as follows: ； in, This is the control correction value output by the PID controller. This is the proportionality coefficient. The integral coefficient is... The differential coefficients are... To control deviation, Control deviations at different times; By using a PID control algorithm, the hoisting speed, luffing speed, and braking parameters are corrected in real time to reduce control deviation. Based on Miner's linear damage accumulation theory, the damage accumulation formula is as follows: ； in, This represents the cumulative damage value. For the first Number of control adjustments under various operating conditions For the first The maximum number of adjustments allowed for the equipment under certain operating conditions, when At that time, reduce the frequency of control parameter adjustments.

7. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, In step 5, the safety early warning mechanism is based on the operating condition identification results and control parameters, and the determination mechanism of the safety early warning mechanism is as follows: Overload warning determination: When At that time, an overload warning is triggered, and the lifting speed is automatically adjusted to... Braking pressure adjusted to Stop lifting operations until the load is restored to a safe range; Instability Early Warning Judgment: Based on the force analysis of the boom, an instability judgment formula is constructed. When... At that time, an instability warning was triggered, among which For the actual force on the boom, The maximum allowable force on the boom is given; the actual force calculation formula for the boom is as follows: ； in, For load suspension length, This refers to the effective length of the boom; Gale instability warning: When When the strong wind instability warning is triggered, the luffing operation is automatically stopped, the boom is adjusted to a safe angle, and the lifting speed is reduced to [missing information]. .

8. The intelligent load control method for cranes based on adaptive operating conditions according to claim 1, characterized in that, The AI ​​adaptive control model also includes a model self-learning module, which collects operating condition identification data and control effect data in real time, and updates the model parameters through a gradient descent algorithm. The gradient descent algorithm update formula is as follows: ； in, For the updated model parameters, These are the current model parameters. For learning rate, This is the partial derivative of the loss function with respect to the current model parameters.

Images