CN119396051A

CN119396051A - A wheel-foot coordinated control method and system for a wheel-foot robot

Info

Publication number: CN119396051A
Application number: CN202411525819.XA
Authority: CN
Inventors: 刘逸群; 冯家亮; 周祎洋; 王剑锋; 丁亮; 高海波
Original assignee: Harbin Institute of Technology Weihai
Current assignee: Harbin Institute of Technology Weihai
Priority date: 2024-10-30
Filing date: 2024-10-30
Publication date: 2025-02-07

Abstract

The invention discloses a wheel foot robot wheel foot cooperative control method and a system, which particularly relate to the technical field of wheel foot robots, wherein the wheel foot robot is divided into a plurality of monitoring areas, temperature distribution and motion load data of each area are obtained in real time through temperature sensors and acceleration sensors arranged in a wheel type and foot type motion control system, the data are preprocessed and analyzed to determine a load abnormal area, a fuzzy logic algorithm is applied to the monitoring area with abnormal load, load distribution is adjusted in real time according to the abnormal degree and the motion speed of the wheel foot load, the energy utilization efficiency is optimized, on the basis, the adjusted load state is predicted, the motion control strategy is adjusted in advance according to the prediction result, the stability and the safety of the system are ensured, the problems of uneven energy consumption and local overheating under the wheel foot separation control are solved through intelligent monitoring and dynamic adjustment, the service life of the system is prolonged, and the running efficiency is improved.

Description

Wheel-foot robot wheel-foot cooperative control method and system

Technical Field

The invention relates to the technical field of wheel-foot robots, in particular to a wheel-foot cooperative control method and a wheel-foot cooperative control system for a wheel-foot robot.

Background

With the rapid development of robotics, wheeled robots and foot robots are widely used in different fields as a common type of mobile robot. The wheel type robot is suitable for a fast moving scene of a flat ground by the characteristics of simple structure, high movement efficiency and low energy consumption, and the foot type robot has stronger terrain adaptability, can keep stable in a rugged environment and is suitable for moving of complex terrains. The wheel foot robot combines the advantages of wheel type and foot type movement modes, can move at high speed on flat ground, and can flexibly walk under complex terrain. Although the wheel foot robot has remarkable advantages in the aspects of terrain adaptability and movement efficiency, the control difficulty is correspondingly increased, and particularly, how to realize wheel foot cooperative control ensures that the robot stably transits and operates efficiently in different modes, and becomes the key point and the difficulty of current research.

The existing wheel-foot robot control method mostly adopts a mode of respectively controlling wheel type movement and foot type movement. This approach typically treats wheeled and foot-type movements as separate modules, each processed using a different control algorithm. However, the control requirements for wheeled and foot motion are different, which may result in uneven energy consumption and heat dissipation. During long operation, if the system frequently switches motion modes, and the thermal management system design is not fine enough, localized overheating of a certain system may result under wheel-foot separation control. For example, the motor heats up severely during high speed operation, but if switching to foot-type motion is followed, the servo motor of the foot-type system and the like also start high load operation, resulting in failure of overall thermal management. Eventually, this may cause the motor or circuit to overheat, even causing a fire or permanent damage to the system.

Disclosure of Invention

The invention aims to provide a wheel-foot robot wheel-foot cooperative control method and system, which solve the defects in the background technology.

In order to achieve the purpose, the invention provides the following technical scheme that the wheel-foot robot wheel-foot cooperative control method comprises the following steps:

s1, dividing a wheel-foot robot into a plurality of monitoring areas, and respectively acquiring temperature distribution data and motion load data in each monitoring area through a temperature sensor and an acceleration sensor which are arranged in a wheel type motion control system and a foot type motion control system;

S2, preprocessing the monitored temperature distribution data and the motion load data, analyzing the preprocessed temperature distribution data and the preprocessed motion load data, and determining a monitoring area with abnormal wheel foot load according to an analysis result;

S3, for a monitoring area with abnormal wheel foot load, adjusting the load distribution of the wheel foot system in real time according to the abnormal degree and the movement speed of the wheel foot load by a fuzzy logic algorithm, and optimizing the energy utilization rate of the system;

And S4, predicting the wheel foot load change condition in the wheel foot load abnormality monitoring area in a fixed time period after adjustment, and adjusting the motion control strategy in advance according to the prediction result to prevent local overheating.

Preferably, in S2, the temperature distribution abnormality index is generated after analyzing the preprocessed temperature distribution data, and the method for obtaining the temperature distribution abnormality index is as follows:

setting a priori distribution P (θ), if the temperature data conforms to the normal distribution, using a priori normal distribution representation therefor Where μ is the initial mean of the temperature and σ ² is the initial variance of the temperature, and for a new temperature data point T, a likelihood function is defined from its actual distribution, i.e., the probability that the current temperature data is observed, the likelihood function P (T|θ) is a probability density function based on the new temperature data T, expressed as: according to the bayesian theorem, the posterior distribution P (θ|t) combines the prior distribution with the likelihood function of the current observed data, and the expression is: P (θ|t) is a posterior distribution, i.e., probability of temperature distribution parameter θ after observing data T, P (θ) is a priori distribution, representing temperature distribution before observation, P (T) is edge likelihood of observed temperature, and temperature distribution abnormality index is calculated with the expression: Wherein KM is an abnormality index of temperature distribution, and [ T _min,T_max ] represents a temperature threshold range.

Preferably, in S2, the motion load drift index is generated after analysis according to the preprocessed motion load data, and the method for obtaining the motion load drift index is as follows:

The motion load data in the Q time period is obtained in real time and is expressed as L ₁,L₂,…,L_n, wherein each L _n is load data at the moment of time n, distance measurement between load data points is defined, the difference between the two load data points is measured, and the expression is as follows: Wherein, L _i,k and L _j,k respectively represent the kth eigenvalue of load data L _i and L _j, m is the characteristic dimension, a neighborhood radius of a point is defined in the load data space, if the distance between two points is smaller than or equal to epsilon, the two points are regarded as neighbors of each other, within the given radius epsilon, minPts neighbors are at least needed to treat the data point as a core point and classify the core point into a cluster, the load data is divided into a plurality of clusters, and the points which cannot be classified into any cluster are marked as noise points, namely abnormal load points: Wherein, |N _∈(L_i) | represents the number of data points within a radius ε, V _∈ represents a neighborhood volume with ε as the radius, for a two-dimensional space of Euclidean distance, the neighborhood volume is V _∈＝π∈², and the motion load drift index is calculated with the expression: Where ρ (L _i) is the local density of data point L _i, ρ _mean (C) is the average local density of cluster C where point L _i is located, and SD is the moving load drift index.

Preferably, in S2, the temperature distribution abnormal index and the motion load drift index are converted into first feature vectors, the first feature vectors are used as inputs of a machine learning model, the machine learning model predicts the abnormal degree value labels of the wheel foot load of each monitoring area by using each group of the first feature vectors as a prediction target, the sum of prediction errors of the abnormal degree value labels of the wheel foot load of all monitoring areas is minimized as a training target, the machine learning model is trained, the model training is stopped until the sum of the prediction errors reaches convergence, and the abnormal degree value of the wheel foot load of each monitoring area is determined according to a model output result, wherein the machine learning model is a polynomial regression model.

Preferably, the obtained abnormal wheel foot load degree value of each monitoring area is compared with a preset reference threshold value of the abnormal wheel foot load degree value, if the abnormal wheel foot load degree value is larger than or equal to the reference threshold value of the abnormal wheel foot load degree value, the abnormal wheel foot load degree in the monitoring area is indicated to be high, an early warning signal is generated at the moment and is marked as an abnormal wheel foot load monitoring area, and if the abnormal wheel foot load degree value is smaller than the reference threshold value of the abnormal wheel foot load degree value, the abnormal wheel foot load degree in the monitoring area is indicated to be low, the early warning signal is not generated at the moment and is marked as an normal wheel foot load monitoring area.

Preferably, in S3, for a monitoring area of abnormal wheel foot load, load distribution of the wheel foot system is adjusted in real time according to abnormal degree and movement speed of the wheel foot load by a fuzzy logic algorithm, so as to optimize energy utilization rate of the system, specifically:

taking the abnormal degree value CP of the wheel foot load and the movement speed E of the wheel foot as input items of a fuzzy logic, and taking the load distribution H of the wheel foot system as output items of the fuzzy logic;

fuzzifying the input variable and the output variable, namely converting the accurate numerical values into fuzzy sets;

constructing fuzzy rules, and generating corresponding output based on the input fuzzy set and a rule base;

the result of fuzzy reasoning is a fuzzy set, which is converted into an accurate output value, namely defuzzification is carried out, and the centroid of fuzzy output is calculated as a final load distribution value H;

a specific load distribution value H is obtained through a centroid method, and represents the load quantity distributed by the system under the current condition.

The invention also provides a wheel-foot robot wheel-foot cooperative control system, which comprises a data acquisition module, a data analysis module, a load adjustment module and a motion control module;

the data acquisition module is used for dividing the wheel-foot robot into a plurality of monitoring areas and respectively acquiring temperature distribution data and motion load data in each monitoring area through a temperature sensor and an acceleration sensor which are arranged in the wheel-type motion control system and the foot-type motion control system;

The data analysis module is used for preprocessing the monitored temperature distribution data and the motion load data, analyzing the preprocessed temperature distribution data and the preprocessed motion load data and determining a monitoring area with abnormal wheel foot load according to an analysis result;

the load adjustment module is used for adjusting the load distribution of the wheel foot system in real time according to the abnormal degree and the movement speed of the wheel foot load through a fuzzy logic algorithm for the monitoring area of the wheel foot load abnormality, so as to optimize the energy utilization rate of the system;

And the motion control module predicts the wheel foot load change condition in the wheel foot load abnormality monitoring area in a fixed time period after adjustment, and adjusts the motion control strategy in advance according to the prediction result to prevent local overheating.

In the technical scheme, the invention has the technical effects and advantages that:

1. The intelligent cooperative control method for the wheel foot robot solves the problems of uneven energy consumption and local overheating in the traditional wheel foot separation control method. By dividing the robot into a plurality of monitoring areas and collecting the temperature and motion load data of each area in real time, the system can accurately monitor the working state of each area. On the basis, the data are deeply analyzed by using the temperature distribution abnormality index and the motion load drift index, and the load abnormality area is identified by a machine learning model. The system can find potential abnormal points in time and provide accurate basis for subsequent real-time adjustment.

2. According to the invention, a fuzzy logic algorithm is adopted for the identified load abnormal region, and the load distribution of the wheel foot system is dynamically adjusted according to the real-time load abnormal degree and the motion speed. The self-adaptive control method enables the system to balance energy consumption under different working conditions, and effectively avoids the situations of local overheating and overload. After load adjustment, the system can also predict future load change, and make and adjust a motion control strategy in advance, so that possible overheat risks are further prevented, the energy utilization efficiency of the wheel-foot robot is improved, and the safety and stability of system operation are obviously enhanced.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments described in the present application, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.

FIG. 1 is a flow chart of the method of the present invention.

FIG. 2 is a block diagram of a system according to the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

An embodiment 1, referring to fig. 1 and 2, is a wheel-foot robot wheel-foot cooperative control method according to the embodiment, which includes the following steps:

In S1, the wheel-foot robot is divided into a plurality of monitoring areas, and temperature distribution data and motion load data in each monitoring area are respectively obtained through a temperature sensor and an acceleration sensor installed in the wheel-type motion control system and the foot-type motion control system, which specifically comprises:

Firstly, the wheel foot robot is divided into a plurality of functional areas according to the physical structure of the wheel foot robot. The typical wheel-foot robot mainly comprises a wheel type motion control system, a foot type motion control system, a central control unit, a power transmission module and the like. Thus, the monitoring area may be divided according to these critical components.

Dividing a wheeled motion monitoring area for a wheeled system typically has multiple wheels or motor drives. Each motor can be considered as an independent monitoring area. The method is concretely divided into:

in the wheel type motor area, the motor of each wheel is a key monitoring point, and a temperature sensor and a power sensor are required to be installed.

The wheel axle and the power transmission area, i.e. the wheels are connected by the wheel axle or the transmission mechanism, which transmission parts may also be affected by temperature and load, so that these areas may be monitored.

The wheel type motor control module area is the position of the motor controller, and the energy consumption and heat dissipation conditions of the motor controller need to be monitored.

The divided foot motion monitoring area foot motion is typically comprised of a plurality of joints, each of which may generate heat and bear the motion load by the motor and drive components. Divided into the following areas:

in the foot joint area, motors and servo drivers corresponding to each foot joint are key monitoring points, and temperature and acceleration sensors are arranged for monitoring the motion load.

Leg structure area-the connection of the legs is also subject to stress, especially when the foot-type movement is loaded with a large load, the temperature change of the transmission member needs to be monitored.

The foot motor control module area, the control unit for driving the foot joints needs to monitor its energy consumption and temperature.

The central control and heat dissipation management areas are divided into, in addition to the wheel-foot exercise system, the central control unit and the heat dissipation management module, which areas are generally responsible for managing the coordination and heat distribution of the overall system. The method is divided into:

the central processing unit area monitors the temperature and load of the main controller, as it generates heat when operated for a long time and is critical to the system stability.

And the heat dissipation management module area comprises a heat dissipation fan, a battery, a heat dissipation fin and the like and is used for monitoring heat dissipation efficiency.

And installing temperature sensors at the motor, the motor control module and the joints of each monitoring area, and using the temperature sensors for monitoring temperature data in real time. Specific sensor mounting points include:

And the wheel type motor and the foot type joint motor are directly provided with temperature sensors for monitoring the working temperature of the motor. Temperature sensors are mounted on the critical circuit boards of the respective control modules to prevent overheating of the control units. These sensors send the collected temperature data to the central control unit at a fixed frequency (e.g., once per second), and the central system stores the data for further analysis.

And the acceleration sensor is installed at the joint of the robot and used for acquiring the motion load data in each monitoring area. The method comprises the steps of installing an acceleration sensor in each foot joint area, and monitoring the acceleration change of the joint in the motion process so as to calculate the actual motion load condition. An acceleration sensor is mounted in the axle area on the wheel for monitoring vibration and load conditions of the wheel during running. The output data of the acceleration sensor can be compared with the planning speed of the robot motion, and whether overload operation or abnormal vibration exists is analyzed.

The temperature distribution data and the motion load data acquired by the sensor are synchronously transmitted to a central control system in a wired or wireless mode. The system can synchronously process the data according to the acquisition frequencies of the sensors in different areas. The data is preprocessed and then input into an advanced algorithm module, and the historical data, the real-time state and the thermal management model are analyzed to identify the temperature rising trend or the overload condition. If a temperature or load exceeds a standard for a certain monitored area, the system will send out an alarm and execute a corresponding adjustment strategy, such as reducing the workload of the area or increasing the working intensity of other areas. And (3) comprehensively analyzing the sensor data of each area through data fusion and decision, and judging whether the current heat dissipation and load distribution are reasonable or not through the system. For example, if multiple motor temperatures are elevated, which may indicate excessive overall energy consumption, the system may reduce high speed motion or reduce total load by algorithms. If the acceleration of a certain joint is obviously abnormal, the system can judge that the joint is possibly overloaded or damaged and send out maintenance reminding.

According to the application, by dividing different monitoring areas, the system can monitor the temperature and the motion load state of the wheel foot robot in a targeted and real-time manner, and the energy consumption and the heat dissipation of each area are ensured to be balanced. The combination of the acquisition of the sensor data and the advanced control algorithm ensures that the robot can intelligently cope with the heat dissipation problem in a complex environment and prevent local overheating or system failure.

S2, preprocessing the monitored temperature distribution data and the motion load data, analyzing the preprocessed temperature distribution data and the preprocessed motion load data, and determining a monitoring area with abnormal wheel foot load according to an analysis result.

The motor temperature and motion load data may be affected by noise, sensor failure or external interference during the acquisition process, resulting in incomplete or erroneous acquired data. In the event of an anomaly in the sensor, the temperature data or the moving load data may be invalid (e.g., zero, negative, or exceeding a physical upper limit). These invalid data need to be removed to avoid affecting subsequent analysis. If part of the data is lost, interpolation methods (such as linear interpolation and spline interpolation) can be adopted to fill in the missing value, or the missing data segment can be deleted directly.

Since the sensors in the wheeled and foot motion control systems may collect data at different sampling rates or different time stamps, synchronization of the data of the different sensors must be ensured during preprocessing for subsequent joint analysis.

Ensuring that the data of all sensors has the same time reference. The sampled data of different sensors may be linearly interpolated or nearest-neighbor interpolated, mapping all data onto the same timeline. If the sensor sampling rates are different, it is often necessary to sample the high frequency sampled data to a lower frequency to ensure that the data of the different sensors are aligned in time and to ensure consistency of subsequent analysis.

Since the motor temperature and motion load data may have different ranges of values (e.g., a temperature range of 0 ℃ to 100 ℃ and an acceleration range of 0to 10m/s ²), normalization is necessary to analyze the data on a uniform scale.

Analyzing the preprocessed temperature distribution data to generate a temperature distribution abnormality index, wherein the temperature distribution abnormality index is obtained by the following steps:

setting a priori distribution P (θ), wherein θ represents the mean and variance of the temperature, and if the temperature data conforms to the normal distribution, using a priori normal distribution representation therefor Where μ is the initial mean of the temperature and σ ² is the initial variance of the temperature, and for a new temperature data point T, a likelihood function is defined from its actual distribution, i.e., the probability that the current temperature data is observed, the likelihood function P (T|θ) is a probability density function based on the new temperature data T, expressed as: according to the bayesian theorem, the posterior distribution P (θ|t) combines the prior distribution with the likelihood function of the current observed data, and the expression is: P (θ|T) is a posterior distribution, i.e., the probability of the temperature distribution parameter θ after the observation of the data T, P (θ) is an a priori distribution, representing the temperature distribution before the observation, P (T) is the edge likelihood of the observed temperature, which can be obtained by integrating all possible parameter values θ, and the temperature distribution abnormality index is calculated by: Wherein KM is an abnormality index of temperature distribution, and [ T _min,T_max ] represents a temperature threshold range.

The greater the temperature profile abnormality index, the more the temperature rise in the monitored zone deviates significantly from the normal range, indicating that the zone is being subjected to a greater load. When the motor or apparatus is in a high load condition, the operating intensity increases, more heat is generated, and the temperature rises rapidly. An abnormal exponential increase generally means that the area may be in an overload operating condition, not in time with heat dissipation or with excessive energy consumption. If the temperature continues to rise without effective control, it may cause the equipment to overheat or even fail. Thus, a large anomaly index generally indicates that the load is out of specification, requiring timely adjustment or load shedding to avoid failure.

The smaller the temperature distribution abnormality index, the closer or the temperature change in the monitored area is to be kept within the normal range, meaning that the load of the apparatus is at a normal or lower level. When the motor or the component operates at a standard load, the generated heat can be effectively discharged through a heat dissipation mechanism of the system, and the temperature fluctuation is small. The lower abnormality index indicates that the region is stable in operation, light in equipment load and good in heat dissipation condition. This generally means that the device is in a healthy operating state, with neither overload nor risk of an abnormal rise in temperature.

The motion load drift index is generated after analysis is carried out according to the preprocessed motion load data, and the motion load drift index acquisition method comprises the following steps:

The motion load data in the Q time period is obtained in real time and is represented as L ₁,L₂,…,L_n, wherein each L _n is load data at time n, and includes multiple dimensions (such as acceleration, moment, speed, etc.), a distance measure between load data points is defined, and the most commonly used distance is euclidean distance, which is used for measuring the difference between two load data points, and the expression is as follows: Where L _i,k and L _j,k represent the kth eigenvalues of load data L _i and L _j, respectively, m is the eigenvector, and in the load data space, a neighborhood radius of a point is defined. If the distance between two points is less than or equal to epsilon, the two points are regarded as neighbors of each other. Within a given radius e, at least MinPts neighbors are needed to treat the data point as a core point and to classify it as a cluster. By the DBSCAN algorithm, the load data is divided into clusters, and the points in the clusters represent load data with higher density and relatively concentrated distribution. And those points that cannot fall into any cluster are marked as noise points, i.e., outlier load points.

Core point if a point L _i has at least MinPts neighbors within a radius e, the point is considered as a core point and belongs to a cluster. Boundary point if the number of neighbors of a certain point L _i is less than MinPts, but it is within the epsilon neighborhood of a certain core point, then it is considered a boundary point. Noise point-a point that fails to meet the above condition is regarded as a noise point, i.e., an outlier. Clustering is based on density, so the local density of points within each cluster should be relatively uniform. The local density can be calculated by the number of neighbors within the e radius. The specific formula is as follows: Where N _∈(L_i represents the number of data points within a radius ε and V _∈ represents the neighborhood volume with ε as the radius. For the two-dimensional space of Euclidean distance, the neighborhood volume is V _∈＝π∈², and the motion load drift index is calculated, wherein the expression is: Where ρ (L _i) is the local density of data point L _i, ρ _mean (C) is the average local density of cluster C where point L _i is located, and SD is the moving load drift index.

When the moving load drift index is larger, the load of the monitoring area is larger deviated from the normal state, and the abnormal load or the severe change of the load in the area is reflected. A large drift index generally means that the area is subjected to excessive loads or significant load fluctuations occur during operation, which may be caused by abrupt environmental changes, equipment aging, or control system failure. If the drift index continues to be at a higher level, indicating that the area may be in an overload condition, long periods of high load operation may lead to increased wear or failure of the equipment, requiring attention and corresponding load adjustment or maintenance measures.

When the moving load drift index is smaller, this indicates that the load of the monitored area is relatively stable and the load level is within a normal or expected range. The smaller drift index indicates smaller load change in the monitoring area, the running state of the equipment is stable, and the system meets the design requirement or the historical running record of the system. This generally means that the system is able to handle the current load efficiently and that the device is in a healthy operating state without immediate intervention. The continuous small drift index reflects a uniform load distribution in this region and can run at lower risk for long periods of time.

Converting the temperature distribution abnormal index and the motion load drift index into first feature vectors, taking the first feature vectors as input of a machine learning model, predicting the abnormal degree value labels of the wheel foot loads of all monitoring areas by using each group of first feature vectors as a prediction target by the machine learning model, taking the sum of the prediction errors of the abnormal degree value labels of the wheel foot loads of all monitoring areas as a training target, training the machine learning model until the sum of the prediction errors reaches convergence, stopping training the model, and determining the abnormal degree value of the wheel foot loads of all monitoring areas according to the model output result, wherein the machine learning model is a polynomial regression model.

The method for acquiring the abnormal degree value of the wheel foot load of each monitoring area comprises the steps of obtaining a corresponding function expression of CP=F (KM, SD) from first feature vector training data of a machine learning model after training, wherein F is an output function of the model, KM is a temperature distribution abnormal index, SD is a motion load drift index, and CP is the abnormal degree value of the wheel foot load of each monitoring area.

Comparing the obtained abnormal wheel foot load degree value of each monitoring area with a preset reference threshold value of the abnormal wheel foot load degree value, if the abnormal wheel foot load degree value is greater than or equal to the reference threshold value of the abnormal wheel foot load degree value, indicating that the abnormal wheel foot load degree in the monitoring area is high, generating an early warning signal at the moment, and marking the early warning signal as an abnormal wheel foot load monitoring area, and if the abnormal wheel foot load degree value is smaller than the reference threshold value of the abnormal wheel foot load degree value, indicating that the abnormal wheel foot load degree in the monitoring area is low, not generating the early warning signal at the moment, and marking the early warning signal as an normal wheel foot load monitoring area.

And S3, for a monitoring area with abnormal wheel foot load, adjusting the load distribution of the wheel foot system in real time according to the abnormal degree and the movement speed of the wheel foot load by a fuzzy logic algorithm, and optimizing the energy utilization rate of the system.

The input variables (CP and E) are blurred, i.e. these exact values are converted into fuzzy sets. This involves assigning consecutive numerical inputs to a set of fuzzy sets (e.g. "low", "medium", "high") to represent different states of the respective input variables.

The abnormal degree value (CP) of the wheel foot load, wherein the CP represents the abnormal degree of the wheel foot load in the monitoring area and reflects the health state of the current system. It can be calculated by the previous machine learning model, and the range is generally [0,1], wherein 0 represents that the abnormal degree of the load is low, and the system state is close to normal. 1, the abnormal degree of the load is extremely high, and the system can be in an overload or serious abnormal state.

After blurring, the CP may be divided into three fuzzy sets, low (Low), with a Low degree of load anomaly. Medium (Medium), the degree of load abnormality is moderate, and slight abnormality may exist. High (High), the degree of load abnormality is High, and severe abnormality is indicated.

The wheel foot movement speed (E) E represents the actual running speed of the wheel foot system, which reflects the current movement state. The range of motion speeds depends on the system itself, such as from 0 (stop) to maximum speed (highest travel speed of the system).

After blurring, E can be divided into three fuzzy sets, slow (Slow) low speed state, system motion is slower. Medium (Medium) Medium speed state, the system is running at normal speed. Fast (Fast), high speed state, the system runs faster.

Fuzzy inference rules are the core of fuzzy logic control, which generates corresponding outputs based on fuzzy sets of inputs and rule bases. The rule base consists of a set of "IF-THEN" rules describing the relationship between inputs and outputs.

A set of fuzzy rules may be defined that map CP (degree of load abnormality) and E (speed of movement) to output H (load distribution):

The ambiguity rule example is that if CP is low and E is slow, H should be the normal allocation. If CP is low and E is fast, H should be a decremental allocation. If CP is medium and E is slow, H should be a slightly increasing allocation. If CP is medium and E is fast, H should be a significant reduction in allocation. If CP is high and E is slow, H should be an incremental allocation. If CP is high and E is fast, H should be a significantly reduced allocation to prevent overload.

When the abnormal load degree is low and the system is close to a normal state, the load distribution can be reasonably adjusted according to the movement speed. If the system speed is fast, it is often desirable to reduce load distribution to avoid excessive energy consumption.

When the degree of load abnormality is moderate, etc., there may be slight abnormality of the system, and thus it is necessary to greatly reduce the load at a high speed to avoid further abnormality.

When the degree of load abnormality is high, the system is in a severe load state. In this case, if the speed is slow, the load distribution can be appropriately increased to avoid excessive energy waste. But if the speed is high, the load must be significantly reduced to prevent overload and damage.

The result of fuzzy reasoning is a fuzzy set that must be converted into accurate output values, i.e. defuzzified. Common defuzzification methods include centroid methods (centroids methods) which calculate the centroid of the fuzzy output as the final load distribution value H.

By centroid method, a specific load distribution value H can be obtained, which represents the amount of load that the system should distribute under the current conditions.

The output variable H represents the amount of load that the wheel foot system should distribute. Through fuzzy logic reasoning, load distribution can be dynamically adjusted so as to ensure that the system can efficiently operate under different motion speeds and load states and avoid overload.

Reduced allocation (reduction) means that the load allocation should be reduced at high load or high speed to prevent overload.

Normal distribution (Normal) means that the load is in a Normal state and the system operates at a standard load.

Increasing the distribution (incrustation) means that the load distribution is appropriately increased at low load or low speed to maintain the energy utilization efficiency.

Through the fuzzy logic control, the system can dynamically adjust the load distribution according to the abnormal degree of the wheel foot load and the motion speed which are monitored in real time. The following are specific ways to optimize the energy utilization of the system:

At high load and high speed, the system reduces energy consumption by reducing load distribution, avoids unnecessary excessive energy consumption, and reduces the risk of equipment damage.

At low load and low speed, the system ensures that energy utilization is not wasted and maintains stable load distribution by properly increasing load distribution.

And during medium load, the load distribution is adjusted according to the speed, so that the system can keep normal operation and no performance loss occurs.

In the application, the abnormal degree value of the wheel foot load and the wheel foot movement speed are used as inputs through a fuzzy logic algorithm, and a load distribution value is generated, so that the load distribution of the system can be adjusted in real time. In the process, the fuzzy inference rule provides a flexible load adjustment strategy for the system according to different load states and speed states, optimizes the energy utilization rate and ensures the stability and efficient operation of the system.

Load data of the system is collected over a fixed period of time, including a temperature profile anomaly index (TDI), a moving load drift index (SD), and a real-time wheel-foot load distribution condition (H). These data can reflect the operating state and the trend of load change of the wheel foot system. Other environmental factors that may affect the load, such as terrain complexity, task type, ambient temperature, etc., are collected and these data help to improve the accuracy of the predictive model.

Common load change prediction models include time series models and machine learning models. The following are some commonly used prediction methods:

ARIMA (autoregressive integral moving average) model) is suitable for processing trends and seasonal fluctuations in time series data. The ARIMA model is able to capture short-term changes and trends in the wheel foot load. From the historical load data, the model can predict load changes over a future period of time, thereby identifying potential overload or local overheating trends.

LSTM (long short term memory network) LSTM is a recurrent neural network model that processes time series data, particularly suited to handle long-term dependencies and non-linear changes in load data. The LSTM model can accurately predict the long-term trend of the load by learning the historical change mode of the load data.

It can capture the composite effect of different factors (such as abnormal load state and movement speed) on future load.

The collected data is consolidated into input feature vectors of the model for training and prediction. Common input features include:

the abnormal Temperature Distribution Index (TDI) reflects the change condition of the temperature load of the system and provides information whether the load has overheat trend.

The motion load drift index (SD) reflects the stability of the load during system motion and helps predict load fluctuations.

Load distribution value (H) the load distribution value previously adjusted by the fuzzy logic is used as a direct reflection of the current load state of the system.

External factors such as terrain complexity, speed variation, etc.

The input features can form an input sequence of a model through time sequence processing, so that the subsequent load change prediction is facilitated.

And (3) for the ARIMA model, training the model by using historical load data, and adjusting autoregressive, moving average and differential parameters of the model so that the model can capture the trend and fluctuation of the load data.

Machine learning model if LSTM deep learning model is used, the historical load data will be used as training set, the model will learn the time correlation and change pattern of system load through multiple iterative training.

The output result is a predicted value of the foot load in the future period of time, including:

The prediction trend of abnormal temperature distribution can predict whether the system will have local overheat.

And (5) predicting load fluctuation, and identifying whether the load can enter an unstable state.

Based on the result of the prediction model, if the load prediction value shows that the temperature rises faster and the load is too high in a future period of time, local overheating can be caused, and the system sends out an early warning signal.

If the temperature profile anomaly index (TDI) prediction continues to rise and approaches or exceeds a set safety threshold, it indicates that the system load will be in an abnormally high temperature condition.

If the motion load drift index (SD) forecast fluctuates significantly, meaning that the load distribution is unstable, it may result in excessive local area load.

Once a future abnormal load trend is identified, the system will adjust the motion control strategy in advance based on the prediction to prevent local overheating. These policies may include:

And (3) dynamically adjusting the load distribution (H) by dynamically adjusting the load distribution of the wheel foot system again through a fuzzy logic algorithm. Such as:

If the load predicted value of a certain area is too high, the load distribution of the area is reduced, and the load is reduced.

At the same time, a portion of the load may be transferred to a region of lower load to balance the overall system load.

If the system predicts that the load at high speed operation will result in overheating, the control strategy may reduce the speed of movement in advance to reduce heat generation. The movement speed is directly related to the load, and the speed is reduced to help reduce the load and prolong the service life of the equipment. In complex terrain or tasks, the system can switch from wheeled motion to foot motion (or vice versa) in advance according to the prediction result so as to adapt to terrain changes and reduce overheat risks caused by load anomalies.

After prediction and adjustment, the system monitors the adjustment effect according to the real-time data. If the adjusted load distribution and motion control strategy is effective to reduce the load, the system will continue to execute the current strategy. Otherwise, the system will further optimize the adjustment until the load condition returns to the normal range.

By a predictive based control strategy, the system can identify possible load problems ahead of time and make adjustments before problems occur. Thus, not only local overheating can be effectively prevented, but also the energy utilization rate of the whole system can be optimized.

The risk of overheating is reduced, and the device overheating caused by high load is avoided by adjusting in advance, so that the risks of energy waste and device damage are reduced.

And (3) balancing load distribution, namely, by intelligently adjusting the load distribution, the system can balance energy consumption among a plurality of areas, and local overload and uneven load are avoided.

The overall operation efficiency is improved, namely the load prediction is combined with a motion control strategy, so that the system can keep a high-efficiency and stable working state in long-time operation.

According to the application, through load change prediction, local overheating in the abnormal wheel foot load monitoring area can be effectively prevented. Based on the prediction results of the temperature distribution abnormality index and the motion load drift index, the system can adjust load distribution and motion control strategies in advance, and prevent performance degradation caused by high load. Therefore, the energy utilization rate of the system is optimized, and the safety and stability of equipment operation are improved.

In this embodiment, the wheel-foot robot is first divided into a plurality of monitoring areas, and temperature distribution data and motion load data in each area are acquired respectively by a temperature sensor and an acceleration sensor installed in the wheel-type and foot-type motion control system. These data are then preprocessed and analyzed to determine the monitored areas of load anomalies. And aiming at the abnormal areas, adopting a fuzzy logic algorithm, and adjusting the load distribution of the system in real time according to the abnormal degree and the movement speed of the wheel foot load so as to optimize the energy utilization rate. And after adjustment, predicting the load change condition in a fixed time period, and adjusting the motion control strategy in advance based on a prediction result to prevent local overheating.

An embodiment 2 of the wheel-foot cooperative control system of the wheel-foot robot includes a data acquisition module, a data analysis module, a load adjustment module and a motion control module;

The above formulas are all formulas with dimensions removed and numerical values calculated, the formulas are formulas with a large amount of data collected for software simulation to obtain the latest real situation, and preset parameters in the formulas are set by those skilled in the art according to the actual situation.

It should be understood that the term "and/or" is merely an association relationship describing the associated object, and means that three relationships may exist, for example, a and/or B, and may mean that a exists alone, while a and B exist alone, and B exists alone, wherein a and B may be singular or plural. In addition, the character "/" herein generally indicates that the associated object is an "or" relationship, but may also indicate an "and/or" relationship, and may be understood by referring to the context.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application.

Claims

1. A wheel foot robot wheel foot cooperative control method is characterized by comprising the following steps of;

2. The wheel-foot robot wheel-foot cooperative control method of claim 1, wherein in S2, the temperature distribution abnormality index is generated after analyzing the preprocessed temperature distribution data, and the temperature distribution abnormality index obtaining method is as follows:

setting a priori distribution P (θ), if the temperature data conforms to the normal distribution, using a priori normal distribution representation therefor Where μ is the initial mean of the temperature and σ ² is the initial variance of the temperature, and for a new temperature data point T, a likelihood function is defined from its actual distribution, i.e., the probability that the current temperature data is observed, the likelihood function P (T|θ) is a probability density function based on the new temperature data T, expressed as: according to the bayesian theorem, the posterior distribution P (θ|t) combines the prior distribution with the likelihood function of the current observed data, and the expression is: P (θ|t) is a posterior distribution, i.e., probability of temperature distribution parameter θ after observing data T, P (θ) is a priori distribution, representing temperature distribution before observation, P (T) is edge likelihood of observed temperature, and temperature distribution abnormality index is calculated with the expression: Where kM is a temperature distribution abnormality index, and [ T _min,T_max ] represents a temperature threshold range.

3. The wheel-foot robot wheel-foot cooperative control method of claim 2, wherein in S2, the motion load drift index is generated after analysis according to the preprocessed motion load data, and the motion load drift index obtaining method comprises the following steps:

4. The wheel-foot robot wheel-foot cooperative control method according to claim 3, wherein S2, the temperature distribution abnormality index and the motion load drift index are converted into first feature vectors, the first feature vectors are used as inputs of a machine learning model, the machine learning model predicts wheel-foot load abnormality degree value labels of all monitoring areas by using each group of first feature vectors as a prediction target, the sum of prediction errors of all the wheel-foot load abnormality degree value labels of the monitoring areas is minimized as a training target, the machine learning model is trained, model training is stopped until the sum of the prediction errors reaches convergence, the wheel-foot load abnormality degree values of all the monitoring areas are determined according to model output results, and the machine learning model is a polynomial regression model.

5. The wheel-foot robot wheel-foot cooperative control method according to claim 4, wherein the obtained wheel-foot load abnormality degree value of each monitoring area is compared with a preset reference threshold value of the wheel-foot load abnormality degree value, if the wheel-foot load abnormality degree value is greater than or equal to the reference threshold value of the wheel-foot load abnormality degree value, the wheel-foot load abnormality degree in the monitoring area is indicated to be high, an early warning signal is generated at the moment, and the early warning signal is marked as the wheel-foot load abnormality monitoring area, and if the wheel-foot load abnormality degree value is smaller than the reference threshold value of the wheel-foot load abnormality degree value, the wheel-foot load abnormality degree in the monitoring area is indicated to be low, and the early warning signal is not generated at the moment, and is marked as the wheel-foot load normal monitoring area.

6. The wheel-foot robot wheel-foot cooperative control method according to claim 5, wherein in S3, for a monitoring area of wheel-foot load abnormality, load distribution of a wheel-foot system is adjusted in real time according to the abnormality degree and the movement speed of the wheel-foot load by a fuzzy logic algorithm, and the energy utilization rate of the system is optimized, specifically:

7. A wheel foot robot wheel foot cooperative control system for realizing the wheel foot robot wheel foot cooperative control method according to any one of claims 1-6, which is characterized by comprising a data acquisition module, a data analysis module, a load adjustment module and a motion control module;