Multi-target torque distribution method and system for multi-wheel drive robot

Abstract

The invention relates to the technical field of driving robots, in particular to a multi-target torque distribution method and system for a multi-wheel driving robot. According to the method, by introducing the dynamic constraint set, it is ensured that constraint conditions of the motor and the tire are considered in real time in the torque distribution solving process, constraint violation is avoided, and by constructing the initial multi-target optimization function and adjusting the target weight coefficient according to the real-time running state data and the torque requirement, the real-time running state data of the motor and the tire is obtained. According to the method, the torque distribution strategy can optimize the multi-target requirement to the maximum extent while meeting the rigid constraint, and the continuous evaluation index and the torque deviation value are considered in the torque distribution process, so that the stable operation performance can be kept in the running process of the multi-wheel drive robot; a feedback adjustment mechanism based on a real-time constraint violation proportion enables a torque distribution strategy to be more refined, so that the defects of a traditional method under complex constraints can be effectively avoided, and the performance and stability of the multi-wheel drive robot in practical application are improved.

Description

Technical Field

[0001] This invention relates to the field of drive robot technology, and in particular to a multi-target torque distribution method and system for a multi-wheel drive robot. Background Technology

[0002] When performing tasks, multi-wheel drive robots need to rationally distribute the torque of each wheel according to multiple objectives. These objectives typically include moving the robot along a predetermined path, ensuring that the force on each wheel is balanced, avoiding excessive wear or uneven load, and dynamically adjusting the torque of each wheel according to factors such as the friction of different wheels, the robot's current posture, and ground conditions.

[0003] Current technologies mainly use weighted least squares to achieve multi-objective torque distribution in multi-wheel drive robots. However, multi-wheel drive robots have hard constraints such as maximum motor torque and tire adhesion limits. The core logic of weighted least squares is to minimize the total deviation first, and then verify the hard constraints such as maximum motor torque and tire adhesion limits. This means that the unconstrained optimal solution is found first, and then the constraint violation is forcibly corrected. However, weighted least squares does not re-optimize the multi-objective balance and can cause abrupt changes in torque at the constraint boundaries, thereby destroying the optimality and continuity of torque distribution and affecting the robot's control accuracy and driving stability. Summary of the Invention

[0004] The main objective of this invention is to provide a multi-objective torque distribution method for multi-wheel drive robots, aiming to solve the technical problems in the prior art.

[0005] This invention proposes a multi-objective torque distribution method for a multi-wheel drive robot, comprising: Acquire real-time operating status data and multiple target torque allocation requirements of the multi-wheel drive robot, and construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; The maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and the real-time operating status data of each drive motor in the multi-wheel drive robot are obtained to construct a dynamic constraint set, and the initial weight coefficients corresponding to the torque distribution requirements of each objective are determined according to the initial multi-objective optimization function and the dynamic constraint set. The initial multi-objective optimization function, dynamic constraint set, and multiple initial weight coefficients are input into a preset optimization solver to obtain an initial torque allocation strategy, and it is determined whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. If the conditions are met, the initial torque allocation strategy will be used as the final torque allocation strategy. If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. The intermediate torque allocation strategy is obtained by resolving the updated multi-objective optimization function and updated constraint set, and the torque response data and operation stability parameters of the multi-wheel drive robot when executing the intermediate torque allocation strategy are collected in real time. Based on the torque response data and operational stability parameters, torque distribution deviation values ​​and continuity evaluation indicators are obtained, and a final torque distribution strategy is generated based on the torque distribution deviation values ​​and continuity evaluation indicators to achieve multi-objective torque distribution for the multi-wheel drive robot.

[0006] This application also provides a multi-objective torque distribution system for a multi-wheel drive robot, including: A construction module is used to acquire real-time operating status data and multiple target torque allocation requirements of a multi-wheel drive robot, and to construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; The determination module is used to acquire the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and real-time operating status data of each drive motor in the multi-wheel drive robot to construct a dynamic constraint set, and to determine the initial weight coefficients corresponding to each objective torque allocation requirement based on the initial multi-objective optimization function and the dynamic constraint set. The input module is used to input the initial multi-objective optimization function, the dynamic constraint set, and multiple initial weight coefficients into a preset optimization solver to obtain an initial torque allocation strategy, and to determine whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. An adjustment module is used to, if satisfied, adopt the initial torque allocation strategy as the final torque allocation strategy; If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. The re-solution module is used to re-solve the problem based on the updated multi-objective optimization function and the updated constraint set, obtain the intermediate torque distribution strategy, and collect torque response data and operational stability parameters of the multi-wheel drive robot when executing the intermediate torque distribution strategy in real time. The generation module is used to obtain torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generate a final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index to achieve multi-objective torque distribution of the multi-wheel drive robot.

[0007] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described multi-objective torque distribution method for a multi-wheel drive robot.

[0008] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described multi-objective torque distribution method for a multi-wheel drive robot.

[0009] The beneficial effects of this invention are as follows: By introducing a dynamic constraint set, this invention ensures that the constraints of the motor and tires are considered in real time during the torque distribution solution process, thus avoiding constraint violations. By constructing an initial multi-objective optimization function and adjusting the target weight coefficients according to real-time operating data and torque requirements, the torque distribution strategy can maximize the optimization of multi-objective requirements while satisfying hard constraints. By considering continuity evaluation indicators and torque deviation values ​​during the torque distribution process, the invention can maintain stable operational performance during the operation of the multi-wheel drive robot. During the solution process, the feedback adjustment mechanism based on the real-time constraint violation ratio makes the torque distribution strategy more refined, and timely corrects solutions that do not meet the constraints, avoiding the disadvantage of over-reliance on post-processing methods. This effectively avoids the shortcomings of traditional methods under complex constraints and improves the performance and stability of multi-wheel drive robots in practical applications. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of a method flow according to an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention.

[0012] Figure 3 This is a schematic diagram of the internal structure of a computer device according to an embodiment of this application.

[0013] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0014] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0015] like Figure 1 As shown, this application provides a multi-objective torque distribution method for a multi-wheel drive robot, including: S1. Obtain real-time operating status data and multiple target torque allocation requirements of the multi-wheel drive robot, and construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; S2. Obtain the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and the real-time operating status data in the multi-wheel drive robot to construct a dynamic constraint set, and determine the initial weight coefficients corresponding to the torque distribution requirements of each objective based on the initial multi-objective optimization function and the dynamic constraint set. S3. Input the initial multi-objective optimization function, dynamic constraint set and multiple initial weight coefficients into a preset optimization solver to obtain the initial torque allocation strategy, and determine whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. S4. If satisfied, the initial torque allocation strategy shall be used as the final torque allocation strategy. If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. S5. Solve the problem again based on the updated multi-objective optimization function and the updated constraint set to obtain the intermediate torque distribution strategy, and collect the torque response data and operation stability parameters of the multi-wheel drive robot when executing the intermediate torque distribution strategy in real time. S6. Obtain the torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generate the final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index to achieve multi-objective torque distribution for the multi-wheel drive robot.

[0016] As described in steps S1-S6 above, the initial torque allocation strategy is obtained by selecting a sequential quadratic programming algorithm as the core solution algorithm to build a preset optimization solver. The initial weight coefficients, the initial multi-objective optimization function, and the dynamic constraint set are input as input parameters to the preset optimization solver. The preset optimization solver iteratively solves the initial multi-objective optimization function to obtain the initial torque allocation value of each drive wheel. The unique wheel position identifier information and the corresponding initial torque allocation value of each drive wheel are obtained. A wheel position-torque mapping relationship table is established based on the unique wheel position identifier information and the initial torque allocation value. The layout parameters of the drive wheels of the multi-wheel drive robot are obtained. The wheel position-torque mapping relationship table is sorted according to the layout parameters. The sorted wheel position-torque mapping relationship table is converted into a structured document, and the scheme generation time and constraint satisfaction status are marked. The initial torque allocation strategy is then formed by combining these elements.

[0017] The steps to obtain the constraint violation ratio are as follows: obtain the torque distribution value of each drive wheel in the initial torque distribution strategy and the hard constraint threshold in the dynamic constraint set; calculate the constraint violation amount by the difference between each torque distribution value and the corresponding hard constraint threshold; and obtain the constraint violation ratio based on the ratio of the constraint violation amount to the hard constraint threshold.

[0018] This invention acquires real-time data on the robot's operating status and the torque requirements of multiple objectives. This not only enables torque allocation based on a static theoretical model but also allows it to adapt to real-time changes in different environments. By constructing an initial multi-objective optimization function, it achieves the balance and optimization of multiple objectives, rather than simply pursuing the optimization of a single objective. This solves the problem in existing technologies that only focus on a single objective (such as minimizing a specific deviation). Multi-objective optimization effectively balances the needs of different objectives, avoiding the problem of infeasible optimization schemes due to unreasonable weight allocation. By acquiring real-time constraints such as the maximum motor torque and tire adhesion limits, the practicality and rationality of the torque allocation strategy are significantly improved. By dynamically adjusting the weight coefficients of the objective torque allocation in real time, the optimization scheme can adaptively adjust according to different operating states, thereby optimizing the balance between different objectives. By introducing motor torque thresholds and tire adhesion limits, the safety and stability of the torque allocation strategy are ensured, avoiding problems such as overloading or tire slippage that may occur in existing technologies due to insufficient consideration of physical constraints, thus improving the operational reliability of the robot.

[0019] Since the initial multi-objective optimization function is constructed based on real-time operating status data and multiple objective torque allocation requirements, its weights are set based on static parameters, such as thermal safety weights based on real-time motor temperature rise data. However, the tire adhesion limit and maximum motor torque in the dynamic constraint set are hard boundaries. When the solver prioritizes the balance of multiple objectives, such as allocating higher torque for dynamic performance, it may passively exceed constraint boundaries, such as exceeding the tire adhesion limit. This is because the initial weights did not fully consider the rigidity priority of the constraints, causing the solution direction to be biased towards multi-objective balance rather than constraint compliance. The dynamic constraint set changes in real-time with the robot's operating state (such as changes in road friction coefficient and increased motor temperature rise), but the initial weights... During the solution process, the priority of the dynamic constraint set is the initial static default order. For example, the default motor torque limit has a higher priority than the tire slip ratio constraint. When the operating state changes abruptly, such as suddenly entering a wet and slippery road surface, the tire adhesion limit drops sharply. The constraints in the initial static default order will become out of touch with the actual requirements. The torque strategy calculated by the solver according to the initial constraint priority will then violate the new constraint boundary. This is because the initial constraint conditions have not kept up with the dynamic changes in the state. Subsequent steps such as judging constraint violations, adjusting weights and constraint priorities, and resolving are to make the optimization function and constraint set more closely match the solver's capability boundary through three key corrections, thereby reducing the impact of limitations. Specific corrections include adjusting the initial weights to tilt the optimization objective towards constraint compliance, and so on. By adjusting the constraint set to match the real-time state, the constraint priorities are matched. This is because the initial constraint priority is static, and subsequent adjustments are made dynamically based on the constraint violation ratio and real-time state. This allows for finding a constraint-compliant, multi-objective balanced, and practically executable torque strategy within the solver's limitations, avoiding large-scale constraint violations due to condition mismatch during the initial solution. Therefore, the solver needs to determine whether the initial torque allocation strategy satisfies all dynamic constraints. This includes verifying one by one whether the torque allocation value of each drive wheel in the initial torque allocation scheme is less than the maximum torque threshold of the corresponding motor, whether it is within the tire adhesion limit parameter range (hard constraint condition), and whether it is within the safe operating range (soft constraint condition), ensuring that from the beginning... This approach avoids torque solutions that violate constraints, improving solution efficiency and effectively preventing repeated adjustments during later optimization processes. It ensures the superiority and stability of the torque allocation strategy. By monitoring constraint satisfaction in real-time during the solution process, it avoids the delayed handling of constraints in traditional methods, identifying non-compliant solutions in advance and making corresponding adjustments, thus reducing the complexity of later adjustments. Dynamically adjusting the weight coefficients and constraint priorities in the optimization scheme using the constraint violation ratio allows for more precise control of torque allocation, avoiding suboptimal results caused by over-correction. The dynamic adjustment strategy ensures that the optimization process avoids local optima, and intelligent priority adjustment ensures the search for the global optimum, thereby improving overall performance.Dynamically updating the optimization function and constraint set enables the system to adapt to different operating states, ensuring that an effective torque distribution strategy can still be generated even under extreme conditions.

[0020] By collecting and analyzing intermediate torque distribution strategies in real time, the system's response can be fed back promptly, ensuring that the obtained solution meets the expected results in actual operation. It not only collects torque response data but also provides a solid basis for subsequent optimization by monitoring operational stability parameters in real time. More comprehensive parameter collection and analysis allows for more accurate evaluation of optimization effects and provides a more robust torque distribution strategy. By comprehensively optimizing the torque distribution strategy using torque distribution deviation values ​​and continuity evaluation indicators, the strategy can be evaluated and adjusted more comprehensively, ensuring optimal performance across multiple dimensions. The introduction of continuity evaluation indicators ensures the stability and smoothness of the torque distribution strategy in long-term operation, effectively avoiding potential torque mutations or instability in existing technologies. By comprehensively considering multiple optimization indicators and evaluation criteria, this invention ensures that the final torque distribution strategy achieves optimal balance across all objectives, significantly improving the overall performance of multi-wheel drive robots and resolving potential conflicts or irreconcilable differences between objectives in existing technologies.

[0021] In one embodiment, step S1, which involves constructing an initial multi-objective optimization function based on the real-time operating status data and multiple target torque distribution requirements, includes: S11. Obtain the real-time motor speed, real-time motor temperature rise data, and vehicle load data from the real-time operating status data, as well as the power requirements, motor thermal safety requirements, and stability requirements from the target torque distribution requirements, and obtain the power target weight based on the real-time motor speed and power requirements. S12. Obtain thermal safety target weights based on the real-time temperature rise data of the motor and the thermal safety requirements of the motor, and obtain stability target weights based on the vehicle load data and stability requirements. S13. Obtain the target torque requirement value and actual output torque value of each drive wheel in the multi-wheel drive robot, and construct a dynamic sub-objective function based on the target torque requirement value, actual output torque value, and dynamic target weight, wherein the dynamic sub-objective function is: Where A represents the dynamic deviation value, a represents the dynamic target weight, and Z... i Y represents the target torque requirement value for the i-th drive wheel. i This represents the actual output torque value of the i-th drive wheel, where i represents the sequence number of the drive wheel and I represents the number of drive wheels. S14. Obtain the real-time temperature rise data and maximum allowable temperature rise threshold of each drive motor in the multi-wheel drive robot, and construct a motor thermal safety sub-objective function based on the real-time temperature rise data, the maximum allowable temperature rise threshold, and the thermal safety target weight. The motor thermal safety sub-objective function is as follows: Where B represents the temperature rise deviation value, b represents the thermal safety target weight, M represents the number of drive motors, m represents the serial number of the drive motor, and X m This represents the real-time temperature rise data of the m-th drive motor, W. m This represents the maximum allowable temperature rise threshold for the m-th drive motor; S15. Obtain the real-time yaw rate and ideal yaw rate of the multi-wheel drive robot as a whole, and construct a stability sub-objective function based on the real-time yaw rate, ideal yaw rate, and stability target weights, wherein the stability sub-objective function is: Where C represents the stability deviation value, c represents the stability target weight, n represents the calculation time window, and U t V represents the real-time yaw rate at time t. t Represents the ideal yaw rate at time t; S16. The stability sub-objective function, the motor thermal safety sub-objective function, and the dynamic performance sub-objective function are weighted and summed to generate the initial multi-objective optimization function.

[0022] As described in steps S11-S16 above, the following steps are performed: A speed weight adjustment coefficient is calculated by comparing the real-time motor speed with the rated motor speed; a power requirement level coefficient is determined based on power requirements; and the power requirement level coefficient and the speed weight adjustment coefficient are multiplied to obtain the power target weight. Similarly, a temperature rise weight adjustment coefficient is calculated by comparing the real-time motor temperature rise data with the maximum allowable temperature rise threshold; a thermal safety requirement level coefficient is determined based on motor thermal safety requirements; and the thermal safety target weight is multiplied by the thermal safety requirement level coefficient and the temperature rise weight adjustment coefficient. Finally, a load weight adjustment coefficient is calculated by comparing the vehicle load data with the maximum design load value; a stability requirement level coefficient is determined based on stability requirements; and the stability target weight is multiplied by the stability requirement level coefficient and the load weight adjustment coefficient.

[0023] This invention accurately reflects the robot's current operating status by acquiring real-time data on motor speed, motor temperature rise, and vehicle load. This allows for dynamic adjustment of the objective function weights. Power weights are generated based on real-time speed and power requirements, ensuring that power targets are prioritized when high power output is needed, thereby improving acceleration performance and response speed. Thermal safety weights are generated based on real-time temperature rise and safety thresholds, automatically reducing torque distribution in high-load or high-temperature environments to prevent excessive temperature rise. Stability weights are generated based on the vehicle load, automatically adjusting torque distribution under different loads or turning conditions to maintain yaw rate close to the ideal value. Dynamic weight adjustment ensures that the three objectives (power performance, thermal safety, and stability) are considered simultaneously in the same optimization process, rather than being verified post-processed, improving the safety and reliability of the optimization results. By weighting the sub-objective functions, the torque output of each wheel is more accurately aligned with the requirements, improving overall power response and avoiding potential problems of insufficient local power or uneven torque distribution between wheels in existing methods. This ensures that the optimization results balance safety and stability while maintaining power performance.

[0024] The motor thermal safety sub-objective function actively limits the output torque of high-temperature motors during the optimization process to avoid overheating. By dynamically adjusting the output of each motor, it ensures motor safety under different loads or long-term operation. By actively reducing yaw rate deviation during the optimization process, it enables the vehicle to maintain an ideal posture during high-speed cornering or on uneven roads. Through joint optimization with the dynamics and thermal safety sub-objective functions, it achieves a balance between power, safety, and stability, improving the driving experience. By integrating the three sub-objective functions through weighted summation, with weights dynamically generated based on real-time status, it achieves dynamic multi-objective optimization. This ensures that the optimization algorithm considers dynamics, thermal safety, and stability simultaneously during the solution process, avoiding over-limit or insufficient performance issues. Dynamic weight adjustment prioritizes the most critical objectives based on current operating conditions, improving the overall system adaptability.

[0025] In one embodiment, step S2, which involves acquiring the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and real-time operating status data in a multi-wheel drive robot to construct a dynamic constraint set, and determining the initial weight coefficients corresponding to each objective torque allocation requirement based on the initial multi-objective optimization function and the dynamic constraint set, includes: S21. Obtain the real-time motor speed and real-time tire slip ratio from the real-time operating status data, and determine the operating status constraints based on the real-time motor speed and real-time tire slip ratio; S22. Obtain the vertical load and adhesion coefficient in each tire adhesion limit parameter, and obtain the maximum adhesion torque of the tire based on the vertical load and adhesion coefficient. S23. Determine the upper limit constraint of tire torque based on the maximum tire adhesion torque, and determine the upper limit constraint of output motor torque of the corresponding drive motor based on the maximum torque threshold. S24. The tire torque upper limit constraint and the output motor torque upper limit constraint are used as hard constraints, and the operating state constraints are used as soft constraints. They are sorted and integrated to obtain a dynamic constraint set. S25. Obtain the corresponding hard constraint correlation coefficient based on the degree of correlation between each sub-objective function in the initial multi-objective optimization function and the hard constraint condition, and obtain the corresponding soft constraint correlation coefficient based on the degree of correlation between each sub-objective function in the initial multi-objective optimization function and the soft constraint condition. S26. Obtain the constraint fit value of the corresponding sub-objective function by performing a weighted summation based on each of the hard constraint correlation coefficients and soft constraint correlation coefficients; S27. Obtain the constraint adaptation coefficients according to the soft constraint conditions, and perform preliminary weight allocation on each target torque distribution requirement according to the constraint adaptation coefficients and multiple constraint adaptation values ​​to obtain the corresponding initial weight coefficients.

[0026] As described in steps S21-S27 above, the step of obtaining the dynamic constraint set is to prioritize hard constraints over soft constraints. Hard constraints are constraints that cannot be broken, as breaking them would lead to robot failure or safety risks. These include the maximum torque threshold constraint for the motor and the maximum adhesion torque constraint for the tires. The upper limit constraint for the tire torque is prioritized over the upper limit constraint for the output motor torque because the risk of tire slippage is higher than the risk of short-term motor overload. Soft constraints are constraints that can be relaxed to a certain extent. Therefore, the initial order of soft constraints is placed after hard constraints. This includes the safe range constraint for motor temperature and the safe range constraint for tire slip rate. The initial order of soft constraints is placed after hard constraints.

[0027] The steps for obtaining the hard constraint correlation coefficient are as follows: obtain the core influencing parameters of each sub-objective function and the constraint parameter set of the hard constraint condition; obtain the correlation parameter proportion based on the ratio of the number of parameters in the constraint parameter set to the total number of core influencing parameters; obtain the optimal torque value based on the closeness between the constraint boundary of the hard constraint condition and the optimization objective value of the sub-objective function; obtain the hard constraint torque limit based on the constraint boundary of the hard constraint component; obtain the constraint difference based on the difference between the optimal torque value and the hard constraint torque limit; obtain the closeness between the constraint boundary of the hard constraint condition and the optimization objective value of the sub-objective function based on the ratio of the constraint difference to the hard constraint torque limit; and finally, obtain the coefficient by linear weighted summation using the closeness and the correlation parameter proportion.

[0028] The steps for obtaining the soft constraint correlation coefficient are as follows: First, obtain the number of control variables that overlap with the core control variables of the optimization objective of each sub-objective function (e.g., yaw rate to be controlled for the stability sub-objective, motor output torque to be controlled for the dynamic sub-objective). Second, obtain the control variable overlap ratio by comparing the number of overlapping control variables with the total number of core control variables of the sub-objective function. Third, obtain the relaxation ratio of the soft constraint during the optimization process of the sub-objective function (i.e., the ratio of the allowable fluctuation range of the soft constraint to the actual fluctuation). Finally, calculate the soft constraint correlation coefficient by performing a linear weighted summation of the control variable overlap ratio and the relaxation ratio.

[0029] This invention optimizes constraints in real time by acquiring motor speed and tire slip ratio, ensuring that torque distribution always meets constraints during operation and avoiding system instability caused by exceeding hard limits. By accurately acquiring tire adhesion limit parameters, it ensures that torque distribution does not exceed tire adhesion capacity, preventing slippage or loss of control. By directly using the upper limit of torque as a hard constraint, it maintains the optimality and continuity of torque distribution, thus directly generating solutions that satisfy all hard constraints, greatly improving computational efficiency and accuracy. Separating hard and soft constraints and sorting and integrating them according to actual conditions helps to more rationally handle various constraints. Through the reasonable differentiation and sorting of hard and soft constraints, the torque distribution strategy can be flexibly adjusted, ensuring the most stringent hard constraints are met. In addition, other soft objectives should be optimized as much as possible. By accurately calculating the correlation weights of multiple objectives, a more detailed balance of objectives can be achieved, avoiding bias towards a single objective during the optimization process. The weight values ​​are dynamically adjusted according to hard constraints to ensure that the priority of each objective can be dynamically adjusted according to the actual situation under different working states, thereby achieving a more balanced and optimized torque distribution. By quantitatively evaluating the importance of each objective and adjusting it according to the adaptation coefficient of soft constraints, a more reasonable weight allocation can be obtained. Using the analytic hierarchy process (AHP) combined with the adaptation coefficient of soft constraints for weight allocation makes the weight of each objective more scientific and reasonable, and can be dynamically adjusted. This ensures that when the robot performs complex tasks, it not only meets hard constraints but also achieves optimal balance in soft objectives, thereby improving the overall system performance and stability.

[0030] In one embodiment, step S4, which adjusts the initial weight coefficients and the priority of the dynamic constraint set based on the constraint violation ratio and real-time operating status data to generate an updated multi-objective optimization function and an updated constraint set, includes: S41. According to the constraint violation ratio, match and obtain the corresponding constraint level from the preset violation ratio-level table, and obtain the corresponding basic weight adjustment range according to the constraint level; S42. Obtain multiple weight adjustment coefficients based on the real-time operating status data, and correct the basic weight adjustment range based on each weight adjustment coefficient to obtain the corresponding corrected weight adjustment range; S43. The corresponding initial weight coefficients are added and updated according to the adjustment range of each corrected weight to obtain the corresponding updated weight coefficients, and the initial multi-objective optimization function is corrected according to the multiple updated weight coefficients to generate the updated multi-objective optimization function; S44. Adjust the constraint priority according to the constraint violation ratio and the real-time tire slip rate to obtain the updated constraint priority, and reorder the hard constraint conditions and soft constraint conditions in the dynamic constraint set according to the updated constraint priority to generate the updated constraint set.

[0031] As described in steps S41-S44 above, the initial weighting coefficients include the initial speed weighting coefficient, the initial temperature rise weighting coefficient, and the initial load weighting coefficient. The updated weighting coefficients include the updated speed weighting coefficient, the updated temperature rise weighting coefficient, and the updated load weighting coefficient. The basic weighting adjustment range is corrected according to the speed weighting adjustment coefficient, the temperature rise weighting adjustment coefficient, and the load weighting adjustment coefficient to obtain the corresponding corrected speed weighting adjustment range, the corrected temperature rise weighting adjustment range, and the corrected load weighting adjustment range. The corresponding initial speed weighting coefficient, the initial temperature rise weighting coefficient, and the initial load weighting coefficient are added together and updated according to the corrected speed weighting adjustment range, the corrected temperature rise weighting adjustment range, and the corrected load weighting coefficient to obtain the corresponding updated speed weighting coefficient, the updated temperature rise weighting coefficient, and the updated load weighting coefficient.

[0032] The total weight coefficient is obtained by summing the updated speed weight coefficient, updated temperature rise weight coefficient, and updated load weight coefficient. Then, the corresponding corrected speed weight ratio, corrected temperature rise weight ratio, and corrected load weight ratio are obtained by calculating the ratios of the updated speed weight coefficient, updated temperature rise weight coefficient, and updated load weight coefficient to the total weight coefficient. The updated multi-objective optimization function is obtained by replacing the initial weight coefficients of the stability sub-objective function, the motor thermal safety sub-objective function, and the dynamic performance sub-objective function in the initial multi-objective optimization function with the corrected speed weight ratio, corrected temperature rise weight ratio, and corrected load weight ratio.

[0033] The steps for adjusting constraint priority include determining the constraint level corresponding to the constraint violation ratio and whether the real-time tire slip rate exceeds the safety threshold. If the constraint level is high and the real-time tire slip rate exceeds the threshold, the upper limit constraint of tire torque is set to the highest priority. If the violation ratio is moderate and the real-time tire slip rate does not exceed the threshold, the priority of hard constraints is maintained and the priority of tire slip rate constraints is increased. If the violation ratio is low, only the internal order of hard constraints is slightly adjusted, and the priority of soft constraints remains unchanged, thus completing the constraint priority adjustment.

[0034] The steps for generating the updated constraint set are as follows: based on the priority of the updated constraints (e.g., when the violation rate of constraints is high, the violated hard constraints are set to the highest priority), the sorting order of hard constraints and soft constraints is adjusted. For example, if the motor torque exceeds the constraint and the temperature exceeds the threshold, the motor maximum torque threshold constraint is raised to the highest priority, the motor temperature safety range constraint is raised to the second highest priority, and the remaining constraints are sorted in order. The hard constraints and soft constraints are then integrated according to the new sorting to form the updated constraint set, thereby ensuring that the solver prioritizes the satisfaction of high-priority constraints.

[0035] This invention introduces a preset violation ratio-gear table, enabling the system to dynamically adjust constraint handling based on specific circumstances for different violation ratios. This allows for earlier prediction and dynamic adjustment of relevant constraints, reducing the need for later corrections. By associating violation ratios with specific gears, the constraint handling strategy can be finely adjusted according to different scenarios, effectively avoiding instability or discontinuity in torque distribution caused by subsequent corrections. Real-time acquisition of operating status data and dynamic adjustment of weight coefficients based on this data improves the response speed and accuracy of the torque distribution system, ensuring that the optimal solution can be adaptively updated according to real-time requirements. Correcting the adjustment range through real-time operating status not only improves the system's performance under different load, speed, or environmental conditions but also effectively avoids torque unevenness or imbalance problems caused by neglecting real-time status in complex dynamic environments. By adding the corrected weight coefficients to the initial weights and updating them, the continuity of the objective function throughout the optimization process is ensured. Each updated optimization function is based on the previous stage. Based on gradual modifications, this approach avoids abrupt changes. By progressively updating the weight coefficients, the optimization function becomes more aligned with actual needs, maintaining a smooth transition and stability of the objective function throughout the process. The updated multi-objective optimization function better suits the current constraints and operating state, avoiding discrete solutions or local optima caused by insufficient consideration of dynamic changes. This helps improve the robot's overall performance in actual operation, ensuring that torque distribution maintains high optimality and stability while satisfying all constraints. By comprehensively considering the changes in constraint violation ratio and real-time slip rate, the constraint priority is dynamically adjusted, ensuring that hard constraints are prioritized during the operation of the multi-wheel drive robot, while soft constraints can be flexibly adjusted according to actual conditions. This allows the system to better cope with complex and changing working environments. By dynamically adjusting the constraint priority during operation, hard and soft constraints can be coordinated and optimized based on real-time data, thereby improving the system's flexibility and the overall effect of constraint handling, ultimately achieving a more balanced and efficient torque distribution strategy.

[0036] In one embodiment, step S5, which involves resolving the problem based on the updated multi-objective optimization function and the updated constraint set to obtain the intermediate torque allocation strategy, includes: S51. Obtain the sub-objective functions in the updated multi-objective optimization function and the priority order in the updated constraint set; S52. Based on the priority sorting of the updated constraint set, high-priority constraints are transformed into mandatory constraints of the preset optimization solver, and low-priority constraints are transformed into constraints with penalty coefficients, thus obtaining constraint expressions that can be recognized by the solver. S53. Weigh and fuse each sub-objective function of the updated multi-objective optimization function according to the corresponding updated weight coefficients to generate the overall objective function; S54. Input the total objective function and constraint expression into a preset optimization solver. The preset optimization solver performs iterative solution with the goal of minimizing the deviation of the total objective function and prioritizing the satisfaction of high-priority constraints until the iteration result meets the convergence condition. Then, it outputs the intermediate torque distribution value of each drive wheel to obtain the intermediate torque distribution strategy.

[0037] As described in steps S51-S54 above, the core algorithm of the preset optimization solver is a sequential quadratic programming algorithm. By obtaining the wheel position identifiers of each driving wheel (left front wheel, right front wheel, left rear wheel, right rear wheel, etc.), a mapping relationship table between wheel position and intermediate torque distribution value is established. By integrating the mapping relationship table with the solution timestamp, the intermediate torque distribution strategy can be obtained. The preset optimization solver needs to perform calculations based on standardized mathematical formulas and cannot directly recognize textual descriptions such as high-priority constraints and secondary-priority constraints. Therefore, the essence of obtaining constraint expressions that the solver can recognize is to transform the constraints ordered by priority in the dynamic constraint set into mathematical models of inequalities or equations that the solver can analyze. At the same time, the difference in constraint priority is reflected through the form of forced constraints or the fusion form of objective functions with penalty coefficients. For example, when the maximum torque threshold constraint of the motor of the multi-wheel drive robot is a hard constraint and the tire slip ratio constraint is a secondary-priority soft constraint, the forced constraint condition is: -E h ≤F h ≤E h Among them, F h E represents the actual output torque of the h-th drive motor. h Let represent the maximum torque threshold of the h-th drive motor, where h represents the drive motor number; the constraint with penalty coefficient is: Where G represents the penalty term for the tire slip ratio constraint, j represents the penalty coefficient for the slip ratio constraint, L represents the number of tires, l represents the tire number, and H... l Let P represent the rolling linear velocity of the l-th tire. l Let K represent the longitudinal velocity of the l-th tire on the robot body, and K represent the tire safety slip ratio threshold. The penalty term of the tire slip ratio constraint is incorporated into the overall objective function. The preset optimization solver will prioritize satisfying the hard constraints when minimizing the overall objective function, while minimizing the penalty value for slip ratio exceeding the threshold, thereby indirectly achieving the priority requirement of the soft constraints.

[0038] This invention prioritizes multiple objective functions and constraint sets to ensure that high-priority constraints are strictly enforced during optimization. Second-priority constraints are guided by penalty coefficients to ensure the solver satisfies high-priority constraints while appropriately considering their impact. This not only effectively guarantees constraint satisfaction but also prevents the solver from getting trapped in local optima due to overly restrictive less important constraints. By generating a weighted fusion overall objective function, the influence between objectives can be flexibly adjusted according to actual needs, ensuring the optimization results meet the multi-objective requirements of practical applications. The introduction of strict priority constraints and minimization of objective function deviations ensures that each iteration makes a reasonable trade-off between actual physical constraints and optimization objectives, avoiding unnecessary computational complexity and instability caused by post-constraint verification. Dynamically adjusting and satisfying constraints during the solution process effectively avoids exceeding hard constraints such as maximum motor torque and tire adhesion, ensuring the physical feasibility of the solution. Strict iterative process control and reasonable application of constraints ensure the continuity of the torque distribution strategy under different operating conditions, avoiding drastic torque changes that may occur in traditional methods, thus improving the stability and reliability of multi-wheel drive robots in practical use.

[0039] In one embodiment, step S6, which involves obtaining the torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generating the final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index, includes: S61. Obtain the actual torque output value and theoretical torque distribution value of each drive wheel in the torque response data, and obtain the relative deviation of the corresponding drive wheel based on the difference between each actual torque output value and theoretical torque distribution value; S62. Take the average value of the multiple relative deviations as the torque distribution deviation value; S63. Obtain the torque fluctuation value and wheel torque difference value at adjacent time moments from the operation stability parameters, and obtain the torque change rate and wheel torque difference standard deviation respectively according to the multiple torque fluctuation values ​​and wheel torque difference values; S64. Use the torque change rate and the standard deviation of the inter-wheel torque difference as continuity evaluation indicators, and determine whether the torque distribution deviation value is less than the preset deviation threshold and whether the continuity evaluation indicator is within the indicator threshold range. S65. If the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, then the intermediate torque distribution strategy shall be used as the final torque distribution strategy. S66. If the torque distribution deviation value is not less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, then obtain the torque deviation correction value based on the difference between the torque distribution deviation value and the preset deviation threshold, and input the torque deviation correction value into the preset optimization solver to solve again until the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, and obtain the final torque distribution strategy. S67. If the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is not within the index threshold range, then adjust the update weight coefficients of each sub-objective function in the multi-objective optimization function according to the continuity evaluation index and input them into the preset optimization solver to solve again until the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, and obtain the final torque distribution strategy.

[0040] As described in steps S61-S67 above, this invention accurately identifies specific deviations in torque distribution by obtaining the difference between the actual torque output value and the theoretical torque distribution value of each drive wheel. By comparing each drive wheel individually, specific deviations can be located more precisely, avoiding local problems that might be overlooked in global methods. Averaging the relative deviations of multiple wheels effectively reduces the impact of abnormal fluctuations or sudden changes in a single wheel, ensuring that optimization decisions better reflect the true performance of the overall system. By obtaining torque fluctuation values ​​and inter-wheel torque differences, the dynamic performance of the system under different operating conditions can be analyzed more comprehensively. By comprehensively considering the accuracy of torque distribution and system continuity, the optimization results ensure that they meet both the requirements for accurate torque distribution and the stability of the system during long-term operation. When both torque deviation and continuity meet the expected standards, the current optimization scheme is directly selected, avoiding unnecessary complex calculations and optimization iterations. The torque distribution deviation correction value is continuously adjusted to gradually approximate the target torque distribution. The optimization objective, through a progressive approach, avoids the local optima problem that may exist in a one-time solution, ensuring a more flexible and precise optimization process. When the torque distribution deviation is not less than the preset deviation threshold, the system is re-solved based on the real-time calculated deviation correction value. The system can dynamically adjust in the face of uncertainties, thereby ensuring continuous adaptability and accuracy in complex environments. When the continuity evaluation index is not within the index threshold range (such as the torque change rate not being within the preset threshold range, or the standard deviation of the inter-wheel torque difference not being within the preset threshold range, or both the torque change rate and the standard deviation of the inter-wheel torque difference not being within the corresponding preset threshold range), the weight coefficients of the optimization function are dynamically adjusted, which can more flexibly balance the relationship between different objectives. Through the dynamic adjustment of the multi-objective optimization function, it is ensured that even when there is a conflict between deviation and continuity, the optimization algorithm can still find a global optimal solution, avoiding the situation in traditional methods where over-reliance on one objective may lead to the failure to meet other objectives.

[0041] like Figure 2 As shown, this application also provides a multi-objective torque distribution system for a multi-wheel drive robot, comprising: A construction module is used to acquire real-time operating status data and multiple target torque allocation requirements of a multi-wheel drive robot, and to construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; The determination module is used to acquire the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and real-time operating status data of each drive motor in the multi-wheel drive robot to construct a dynamic constraint set, and to determine the initial weight coefficients corresponding to each objective torque allocation requirement based on the initial multi-objective optimization function and the dynamic constraint set. The input module is used to input the initial multi-objective optimization function, the dynamic constraint set, and multiple initial weight coefficients into a preset optimization solver to obtain an initial torque allocation strategy, and to determine whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. An adjustment module is used to, if satisfied, adopt the initial torque allocation strategy as the final torque allocation strategy; If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. The re-solution module is used to re-solve the problem based on the updated multi-objective optimization function and the updated constraint set, obtain the intermediate torque distribution strategy, and collect torque response data and operational stability parameters of the multi-wheel drive robot when executing the intermediate torque distribution strategy in real time. The generation module is used to obtain torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generate a final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index to achieve multi-objective torque distribution of the multi-wheel drive robot.

[0042] In one embodiment, the re-solution module includes: The acquisition unit is used to acquire the sub-objective functions in the update multi-objective optimization function and the priority order in the update constraint set; The transformation unit is used to transform high-priority constraints into mandatory constraints of the preset optimization solver according to the priority order of the updated constraint set, and transform low-priority constraints into constraints with penalty coefficients, so as to obtain constraint expressions that can be recognized by the solver. The fusion unit is used to weight and fuse each sub-objective function of the updated multi-objective optimization function according to the corresponding updated weight coefficients to generate the overall objective function. The iterative solution unit is used to input the total objective function and constraint expression into a preset optimization solver for iterative solution until the iteration result meets the convergence condition, and output the intermediate torque distribution value of each drive wheel to obtain the intermediate torque distribution strategy.

[0043] It should be noted that each module and unit in the multi-objective torque distribution system of the multi-wheel drive robot corresponds one-to-one with the steps in the multi-objective torque distribution method of the multi-wheel drive robot.

[0044] like Figure 3 As shown, this application also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores all data required for the process of the multi-objective torque distribution method for the multi-wheel drive robot. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the multi-objective torque distribution method for the multi-wheel drive robot.

[0045] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment on which the present application is applied.

[0046] An embodiment of this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the above-described multi-objective torque distribution methods for multi-wheel drive robots.

[0047] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0048] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0049] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for multi-objective torque distribution in a multi-wheel drive robot, characterized in that, include: Acquire real-time operating status data and multiple target torque allocation requirements of the multi-wheel drive robot, and construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; The maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and the real-time operating status data of each drive motor in the multi-wheel drive robot are obtained to construct a dynamic constraint set, and the initial weight coefficients corresponding to the torque distribution requirements of each objective are determined according to the initial multi-objective optimization function and the dynamic constraint set. The initial multi-objective optimization function, dynamic constraint set, and multiple initial weight coefficients are input into a preset optimization solver to obtain an initial torque allocation strategy, and it is determined whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. If the conditions are met, the initial torque allocation strategy will be used as the final torque allocation strategy. If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. The intermediate torque allocation strategy is obtained by resolving the updated multi-objective optimization function and updated constraint set, and the torque response data and operation stability parameters of the multi-wheel drive robot when executing the intermediate torque allocation strategy are collected in real time. Based on the torque response data and operational stability parameters, torque distribution deviation values ​​and continuity evaluation indicators are obtained, and a final torque distribution strategy is generated based on the torque distribution deviation values ​​and continuity evaluation indicators to achieve multi-objective torque distribution for the multi-wheel drive robot.

2. The multi-target torque distribution method for a multi-wheel drive robot according to claim 1, characterized in that, The step of constructing an initial multi-objective optimization function based on the real-time operating status data and multiple target torque distribution requirements includes: The system acquires real-time motor speed, real-time motor temperature rise data, and vehicle load data from the real-time operating status data, as well as power requirements, motor thermal safety requirements, and stability requirements from the target torque distribution requirements, and obtains the power target weight based on the real-time motor speed and power requirements. The thermal safety target weight is obtained based on the real-time temperature rise data of the motor and the thermal safety requirements of the motor, and the stability target weight is obtained based on the vehicle load data and stability requirements. Obtain the target torque requirement value and actual output torque value of each drive wheel in the multi-wheel drive robot, and construct a dynamic sub-objective function based on the target torque requirement value, actual output torque value and dynamic target weight; The real-time temperature rise data and maximum allowable temperature rise threshold of each drive motor in the multi-wheel drive robot are obtained, and a motor thermal safety sub-objective function is constructed based on the real-time temperature rise data, the maximum allowable temperature rise threshold and the thermal safety target weight. The real-time yaw rate and ideal yaw rate of the multi-wheel drive robot are obtained, and a stability sub-objective function is constructed based on the real-time yaw rate, ideal yaw rate and stability target weight. The initial multi-objective optimization function is generated by weighted summation of the stability sub-objective function, the motor thermal safety sub-objective function, and the dynamic performance sub-objective function.

3. The multi-objective torque distribution method for a multi-wheel drive robot according to claim 1, characterized in that, The steps of acquiring the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire, and real-time operating status data of the multi-wheel drive robot to construct a dynamic constraint set, and determining the initial weight coefficients corresponding to each objective torque allocation requirement based on the initial multi-objective optimization function and the dynamic constraint set, include: The system acquires the real-time motor speed and real-time tire slip ratio from the real-time operating status data, and determines the operating status constraints based on the real-time motor speed and real-time tire slip ratio. Obtain the vertical load and adhesion coefficient from each tire adhesion limit parameter, and obtain the maximum adhesion torque of the tire based on the vertical load and adhesion coefficient; The upper limit constraint of tire torque is determined based on the maximum tire adhesion torque, and the upper limit constraint of output motor torque of the corresponding drive motor is determined based on the maximum torque threshold. The tire torque upper limit constraint and the output motor torque upper limit constraint are used as hard constraints, and the operating state constraint is used as a soft constraint. They are sorted and integrated to obtain a dynamic constraint set. The hard constraint correlation coefficient is obtained based on the degree of correlation between each sub-objective function in the initial multi-objective optimization function and the hard constraint condition, and the soft constraint correlation coefficient is obtained based on the degree of correlation between each sub-objective function in the initial multi-objective optimization function and the soft constraint condition. Obtain the constraint fit value of the corresponding sub-objective function based on each of the hard constraint correlation coefficients and soft constraint correlation coefficients; The constraint adaptation coefficients are obtained based on the soft constraint conditions, and the initial weights of each target torque distribution requirement are initially assigned based on the constraint adaptation coefficients and multiple constraint adaptation values ​​to obtain the corresponding initial weight coefficients.

4. The multi-target torque distribution method for a multi-wheel drive robot according to claim 1, characterized in that, The step of adjusting the initial weight coefficients and the priority of the dynamic constraint set based on the constraint violation ratio and real-time operating status data, and generating an updated multi-objective optimization function and an updated constraint set, includes: According to the constraint violation ratio, the corresponding constraint level is obtained by matching from the preset violation ratio-level table, and the corresponding basic weight adjustment range is obtained according to the constraint level. Multiple weight adjustment coefficients are obtained based on the real-time operating status data, and the basic weight adjustment range is corrected based on each weight adjustment coefficient to obtain the corresponding corrected weight adjustment range. The corresponding initial weight coefficients are updated according to the adjustment magnitude of each corrected weight to obtain the corresponding updated weight coefficients. The initial multi-objective optimization function is then corrected according to the multiple updated weight coefficients to generate the updated multi-objective optimization function. The constraint priorities are adjusted based on the constraint violation ratio and the real-time tire slip rate to obtain updated constraint priorities. The hard and soft constraints in the dynamic constraint set are then reordered based on the updated constraint priorities to generate an updated constraint set.

5. The multi-target torque distribution method for a multi-wheel drive robot according to claim 1, characterized in that, The step of resolving the problem based on the updated multi-objective optimization function and the updated constraint set to obtain the intermediate torque allocation strategy includes: Obtain the sub-objective functions in the update multi-objective optimization function and the priority order in the update constraint set; Based on the priority order of the updated constraint set, high-priority constraints are transformed into mandatory constraints of the preset optimization solver, and low-priority constraints are transformed into constraints with penalty coefficients, resulting in constraint expressions that can be recognized by the solver. The sub-objective functions of the updated multi-objective optimization function are weighted and fused according to their corresponding updated weight coefficients to generate the overall objective function; The overall objective function and constraint expression are input into a preset optimization solver for iterative solution until the iteration result meets the convergence condition. The intermediate torque distribution value of each drive wheel is then output to obtain the intermediate torque distribution strategy.

6. The multi-objective torque distribution method for a multi-wheel drive robot according to claim 1, characterized in that, The step of obtaining the torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generating the final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index, includes: The actual torque output value and theoretical torque distribution value of each drive wheel in the torque response data are obtained, and the relative deviation of the corresponding drive wheel is obtained based on the difference between each actual torque output value and the theoretical torque distribution value. The average value of the multiple relative deviations is taken as the torque distribution deviation value; Obtain the torque fluctuation value and wheel torque difference value at adjacent time moments from the operational stability parameters, and obtain the torque change rate and wheel torque difference standard deviation respectively based on multiple torque fluctuation values ​​and wheel torque difference values; The torque change rate and the standard deviation of the inter-wheel torque difference are used as continuity evaluation indicators, and it is determined whether the torque distribution deviation value is less than the preset deviation threshold and whether the continuity evaluation indicator is within the indicator threshold range. If the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, then the intermediate torque distribution strategy will be used as the final torque distribution strategy. If the torque distribution deviation value is not less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, then the torque deviation correction value is obtained according to the torque distribution deviation value and the preset deviation threshold, and the torque deviation correction value is input to the preset optimization solver to solve again until the torque distribution deviation value is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, and the final torque distribution strategy is obtained. If the torque distribution deviation is less than the preset deviation threshold and the continuity evaluation index is not within the index threshold range, then the update weight coefficients of each sub-objective function in the multi-objective optimization function are adjusted and updated according to the continuity evaluation index and input into the preset optimization solver for re-solving until the torque distribution deviation is less than the preset deviation threshold and the continuity evaluation index is within the index threshold range, thus obtaining the final torque distribution strategy.

7. A multi-objective torque distribution system for a multi-wheel drive robot, characterized in that, include: A construction module is used to acquire real-time operating status data and multiple target torque allocation requirements of a multi-wheel drive robot, and to construct an initial multi-objective optimization function based on the real-time operating status data and multiple target torque allocation requirements; The determination module is used to obtain the maximum torque threshold of each drive motor, the adhesion limit parameters of each tire and the real-time operating status data of each drive motor in the multi-wheel drive robot to construct a dynamic constraint set, and to determine the initial weight coefficients corresponding to each objective torque allocation requirement based on the initial multi-objective optimization function and the dynamic constraint set. The input module is used to input the initial multi-objective optimization function, the dynamic constraint set, and multiple initial weight coefficients into a preset optimization solver to obtain an initial torque allocation strategy, and to determine whether the initial torque allocation strategy satisfies all the constraints in the dynamic constraint set. An adjustment module is used to, if satisfied, adopt the initial torque allocation strategy as the final torque allocation strategy; If not satisfied, the constraint violation ratio is obtained, and the initial weight coefficient and the priority of the dynamic constraint set are adjusted according to the constraint violation ratio and real-time running status data to generate an updated multi-objective optimization function and an updated constraint set. The re-solution module is used to re-solve the problem based on the updated multi-objective optimization function and the updated constraint set, obtain the intermediate torque distribution strategy, and collect torque response data and operational stability parameters of the multi-wheel drive robot when executing the intermediate torque distribution strategy in real time. The generation module is used to obtain torque distribution deviation value and continuity evaluation index based on the torque response data and operating stability parameters, and generate a final torque distribution strategy based on the torque distribution deviation value and continuity evaluation index to achieve multi-objective torque distribution of the multi-wheel drive robot.

8. The multi-objective torque distribution system for a multi-wheel drive robot according to claim 7, characterized in that, The re-solution module includes: The acquisition unit is used to acquire the sub-objective functions in the update multi-objective optimization function and the priority order in the update constraint set; The transformation unit is used to transform high-priority constraints into mandatory constraints of the preset optimization solver according to the priority order of the updated constraint set, and transform low-priority constraints into constraints with penalty coefficients, so as to obtain constraint expressions that can be recognized by the solver. The fusion unit is used to weight and fuse each sub-objective function of the updated multi-objective optimization function according to the corresponding updated weight coefficients to generate the overall objective function. The iterative solution unit is used to input the total objective function and constraint expression into a preset optimization solver for iterative solution until the iteration result meets the convergence condition, and output the intermediate torque distribution value of each drive wheel to obtain the intermediate torque distribution strategy.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images