Method and apparatus for predicting energy consumption of industrial robot, and device and storage medium

Abstract

Disclosed in the present invention are a method and apparatus for predicting the energy consumption of an industrial robot, and a device and a storage medium. The method comprises: associating motion process data of an industrial robot with energy consumption data thereof, so as to generate a motion-energy-consumption time-series dataset; on the basis of a density clustering algorithm, dividing the motion-energy-consumption time-series dataset into industrial robot datasets under different operation conditions; constructing industrial robot energy consumption prediction models under the different operation conditions on the basis of a LightGBM algorithm, performing model training on the industrial robot energy consumption prediction models on the basis of the industrial robot datasets, and determining optimal industrial robot energy consumption prediction models under the different operation conditions; and inputting acquired real-time motion process data into a target optimal industrial robot energy consumption prediction model corresponding to a target operation condition, such that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data. Compared with the prior art, the technical solution of the present invention can improve the measurement accuracy of the energy consumption of an industrial robot and reduce the cost of energy consumption measurement.

Description

A method, device, equipment and storage medium for predicting energy consumption of industrial robots Technical Field

[0001] The present invention relates to the technical field of energy consumption measurement, and in particular to a method, device, equipment and storage medium for predicting the energy consumption of an industrial robot. Background Art

[0002] At present, the automotive manufacturing industry uses a large number of automated industrial robots, and the real-time energy consumption of industrial robots is mostly measured and collected directly using industrial intelligent sensors. These industrial intelligent sensors are expensive. If each industrial robot is equipped with an industrial intelligent sensor to collect real-time energy consumption data, the cost will be high.

[0003] In the existing technology, there are also methods that use machine learning to measure the real-time energy consumption of industrial robots, but the existing method generally adopts the method of splitting the different components of the industrial robot for measurement, which cannot well reflect the energy consumption of the industrial robot in the actual process; and the existing energy consumption measurement model may use fewer data variables and cannot fully consider the impact of factors such as the motion characteristics of the industrial robot on energy consumption, so the accuracy may also be low.

[0004] Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, device, equipment and storage medium for predicting the energy consumption of an industrial robot, which can improve the measurement accuracy of the energy consumption of the industrial robot and reduce the cost of energy consumption measurement.

[0006] In order to solve the above technical problems, the present invention provides a method for predicting the energy consumption of an industrial robot, comprising:

[0007] Acquiring motion process data and energy consumption data of the industrial robot, correlating the motion process data and the energy consumption data, and generating a motion-energy consumption time series data set;

[0008] Based on a density clustering algorithm, the motion-energy consumption time series dataset is subjected to working condition identification, and the motion-energy consumption time series dataset is divided into industrial robot datasets under different working conditions;

[0009] Building an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, training the industrial robot energy consumption prediction model based on the industrial robot dataset, and determining the optimal industrial robot energy consumption prediction model under different working conditions;

[0010] Acquire real-time motion process data of the industrial robot under target working conditions, and input the real-time motion process data into a target optimal industrial robot energy consumption prediction model corresponding to the target working conditions, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

[0011] In one possible implementation, motion process data and energy consumption data of an industrial robot are obtained, wherein the motion process data include the name, first model, first time, axis joint angle, axis motor current, axis joint torque, and axis temperature of a first industrial robot; and the energy consumption data include the name, second model, second time, and energy consumption of a second industrial robot.

[0012] In a possible implementation, the motion process data and the energy consumption data are correlated to generate a motion-energy consumption time series data set, specifically including:

[0013] Obtaining the name, the first model, and the first time of the first industrial robot from the motion process data, and simultaneously obtaining the name, the second model, and the second time of the second industrial robot from the energy consumption data;

[0014] Performing a similarity judgment on the name, the first model, and the first time of the first industrial robot with the name, the second model, and the second time of the second industrial robot, respectively; if the name of the first industrial robot is the same as the name of the second industrial robot, the first model is the same as the second model, and the first time is the same as the second time, obtaining the axis joint angle, the axis motor current, the axis joint torque, and the axis temperature from the motion process data, and obtaining the energy consumption from the energy consumption data;

[0015] The axis joint angle, the axis motor current, the axis joint torque, the axis temperature and the energy consumption are associated to generate a motion-energy consumption time series data set.

[0016] In one possible implementation, working condition identification is performed on the motion-energy consumption time series dataset based on a density clustering algorithm, and the motion-energy consumption time series dataset is divided into industrial robot datasets under different working conditions, specifically including:

[0017] Extracting motion time series data from the motion-energy consumption time series data set to obtain a motion time series data set;

[0018] Calculating a kernel matrix of the motion time series data set based on a preset Gaussian kernel function, obtaining eigenvalues ​​and eigenvectors of the kernel matrix, and calculating a feature contribution rate based on the eigenvalues ​​and the eigenvectors;

[0019] Comparing the feature contribution rate with a preset feature contribution rate threshold, and if the feature contribution rate is not less than the feature contribution rate threshold, obtaining a principal component data set;

[0020] Inputting the principal component data set into a density clustering algorithm to obtain a working condition label data set corresponding to the principal component data set;

[0021] The motion-energy consumption time series dataset is classified based on the working condition label dataset to obtain industrial robot datasets under different working conditions.

[0022] In one possible implementation, a prediction model for the energy consumption of industrial robots under different working conditions is constructed based on the LightGBM algorithm, specifically including:

[0023] An industrial robot energy consumption prediction model under different working conditions is constructed based on the LightGBM algorithm, wherein the industrial robot energy consumption prediction model under each working condition is set with LightGBM hyperparameters, and the LightGBM hyperparameters include the maximum depth of the tree, the minimum leaf node sample weight, the learning rate, and the proportion of subsamples in the entire sample set.

[0024] In one possible implementation, training the industrial robot energy consumption prediction model based on the industrial robot dataset to determine the optimal industrial robot energy consumption prediction model under different working conditions specifically includes:

[0025] Initializing the LightGBM hyperparameters in the industrial robot energy consumption prediction model under different working conditions to obtain an initialized industrial robot energy consumption prediction model;

[0026] An industrial robot training dataset is extracted from the industrial robot dataset, the industrial robot training dataset is input into the initialized industrial robot energy consumption prediction model, and the initialized industrial robot energy consumption prediction model is iteratively optimized. In the iterative optimization process, an adaptive random search algorithm is used to perform parameter optimization on the LightGBM hyperparameters in the initialized industrial robot energy consumption prediction model, and the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process is calculated until a preset maximum number of iterations is reached;

[0027] Obtain the fitness value corresponding to each iterative optimization process, select the first LightGBM hyperparameter corresponding to the maximum fitness value, and determine the optimal industrial robot energy consumption prediction model under different working conditions based on the first LightGBM hyperparameter.

[0028] The present invention also provides an industrial robot energy consumption prediction device, comprising: an industrial robot data association module, a data set partitioning module, a model training module and an energy consumption data prediction module;

[0029] The industrial robot data association module is used to obtain the motion process data and energy consumption data of the industrial robot, perform association processing on the motion process data and the energy consumption data, and generate a motion-energy consumption time series data set;

[0030] The data set division module is used to perform working condition identification on the motion-energy consumption time series data set based on a density clustering algorithm, and divide the motion-energy consumption time series data set into industrial robot data sets under different working conditions;

[0031] The model training module is used to build an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, perform model training on the industrial robot energy consumption prediction model based on the industrial robot dataset, and determine the optimal industrial robot energy consumption prediction model under different working conditions;

[0032] The energy consumption data prediction module is used to obtain the real-time motion process data of the industrial robot under the target working conditions, and input the real-time motion process data into the target optimal industrial robot energy consumption prediction model corresponding to the target working conditions, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

[0033] In one possible implementation, the industrial robot data association module is used to obtain the motion process data and energy consumption data of the industrial robot, wherein the motion process data includes the name, first model, first time, axis joint angle, axis motor current, axis joint torque, and axis temperature of the first industrial robot; and the energy consumption data includes the name, second model, second time, and energy consumption of the second industrial robot.

[0034] In one possible implementation, the data set division module is configured to perform working condition identification on the motion-energy consumption time series data set based on a density clustering algorithm, and divide the motion-energy consumption time series data set into industrial robot data sets under different working conditions, specifically including:

[0035] Extracting motion time series data from the motion-energy consumption time series data set to obtain a motion time series data set;

[0036] Calculating a kernel matrix of the motion time series data set based on a preset Gaussian kernel function, obtaining eigenvalues ​​and eigenvectors of the kernel matrix, and calculating a feature contribution rate based on the eigenvalues ​​and the eigenvectors;

[0037] Comparing the feature contribution rate with a preset feature contribution rate threshold, and if the feature contribution rate is not less than the feature contribution rate threshold, obtaining a principal component data set;

[0038] Inputting the principal component data set into a density clustering algorithm to obtain a working condition label data set corresponding to the principal component data set;

[0039] The motion-energy consumption time series dataset is classified based on the working condition label dataset to obtain industrial robot datasets under different working conditions.

[0040] In one possible implementation, the model training module is used to build an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, specifically including:

[0041] An industrial robot energy consumption prediction model under different working conditions is constructed based on the LightGBM algorithm, wherein the industrial robot energy consumption prediction model under each working condition is set with LightGBM hyperparameters, and the LightGBM hyperparameters include the maximum depth of the tree, the minimum leaf node sample weight, the learning rate, and the proportion of subsamples in the entire sample set.

[0042] In one possible implementation, the model training module is configured to perform model training on the industrial robot energy consumption prediction model based on the industrial robot dataset to determine the optimal industrial robot energy consumption prediction model under different working conditions, specifically including:

[0043] Initializing the LightGBM hyperparameters in the industrial robot energy consumption prediction model under different working conditions to obtain an initialized industrial robot energy consumption prediction model;

[0044] An industrial robot training dataset is extracted from the industrial robot dataset, the industrial robot training dataset is input into the initialized industrial robot energy consumption prediction model, and the initialized industrial robot energy consumption prediction model is iteratively optimized. In the iterative optimization process, an adaptive random search algorithm is used to perform parameter optimization on the LightGBM hyperparameters in the initialized industrial robot energy consumption prediction model, and the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process is calculated until a preset maximum number of iterations is reached;

[0045] Obtain the fitness value corresponding to each iterative optimization process, select the first LightGBM hyperparameter corresponding to the maximum fitness value, and determine the optimal industrial robot energy consumption prediction model under different working conditions based on the first LightGBM hyperparameter.

[0046] The present invention also provides a terminal device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for predicting the energy consumption of an industrial robot as described in any one of the above items is implemented.

[0047] The present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the industrial robot energy consumption prediction method as described in any one of the above.

[0048] The present invention provides a method, device, equipment, and storage medium for predicting the energy consumption of an industrial robot. Compared with the prior art, the present invention has the following advantages:

[0049] By acquiring the motion process data and energy consumption data of the industrial robot, correlating the motion process data and the energy consumption data to generate a motion-energy consumption time series data set, we can better understand the relationship between motion and energy consumption and lay the foundation for subsequent energy consumption prediction; based on the density clustering algorithm, the motion-energy consumption time series data set is used to identify the working conditions and divide the motion-energy consumption time series data set into industrial robot data sets under different working conditions; the data set can be effectively divided into industrial robot data sets under different working conditions, which is helpful for the subsequent analysis and modeling of motion process data and energy consumption data under different working conditions; based on the LightGBM algorithm, an industrial robot energy consumption prediction model under different working conditions is constructed, and the industrial robot data set is used to identify the working conditions. The industrial robot energy consumption prediction model is trained to determine the optimal industrial robot energy consumption prediction model under different working conditions, which can accurately predict the energy consumption of the industrial robot under different working conditions; the real-time motion process data of the industrial robot under the target working condition is obtained, and the real-time motion process data is input into the target optimal industrial robot energy consumption prediction model corresponding to the target working condition, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data, and realizes the real-time energy consumption prediction of the industrial robot under the target working condition, avoiding the high measurement cost problem of the industrial robot brought by directly using industrial intelligent sensors, and can reduce the energy consumption measurement cost; compared with the prior art, the technical solution of the present invention can improve the measurement accuracy of the energy consumption of the industrial robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] FIG1 is a flow chart of an embodiment of a method for predicting energy consumption of an industrial robot provided by the present invention;

[0051] FIG2 is a schematic structural diagram of an embodiment of a device for predicting energy consumption of an industrial robot provided by the present invention;

[0052] FIG3 is a schematic diagram of energy consumption prediction effect according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0054] Example 1, referring to FIG1 , is a flow chart of an embodiment of a method for predicting energy consumption of an industrial robot provided by the present invention. As shown in FIG1 , the method includes steps 101 to 104, which are specifically as follows:

[0055] Step 101: Acquire motion process data and energy consumption data of an industrial robot, perform correlation processing on the motion process data and the energy consumption data, and generate a motion-energy consumption time series data set.

[0056] In one embodiment, the motion process data and energy consumption data of the industrial robot are extracted from the intelligent manufacturing data service platform.

[0057] Specifically, the energy consumption data is obtained by collecting energy consumption data of industrial robots through industrial intelligent sensors.

[0058] Specifically, the energy consumption data of the industrial robot is repeatedly sampled within a preset time period based on the industrial intelligent sensor to obtain multiple energy consumption data, and the motion process data and the energy consumption data are stored in the intelligent manufacturing data service platform.

[0059] Preferably, the energy consumption data is historical data stored in the intelligent manufacturing data service platform.

[0060] In one embodiment, the motion process data includes the name of the first industrial robot, the first model, the first time, the axis joint angle, the axis motor current, the axis joint torque, and the axis temperature.

[0061] Specifically, the axis joint angle, the axis motor current, the axis joint torque, and the shaft temperature include a first axis joint angle, a first axis motor current, a first axis joint torque, a first axis shaft temperature, a second axis joint angle, a second axis motor current, a second axis joint torque, a second axis shaft temperature, a third axis joint angle, a third axis motor current, a third axis joint torque, a third axis shaft temperature, a fourth axis joint angle, a fourth axis motor current, a fourth axis joint torque, a fourth axis shaft temperature, a fifth axis joint angle, a fifth axis motor current, a fifth axis joint torque, a fifth axis shaft temperature, a sixth axis joint angle, a sixth axis motor current, a sixth axis joint torque, and a sixth axis shaft temperature. By arranging the motion process data of the industrial robot to include multiple types of data, the data source for energy consumption measurement can be enriched, and the accuracy of subsequent energy consumption measurement can be improved.

[0062] Preferably, the motion process data format is {first industrial robot name, first model, first time, one-axis joint angle, one-axis motor current, one-axis joint torque, one-axis shaft temperature, ..., six-axis torque, six-axis shaft temperature}.

[0063] In one embodiment, the energy consumption data includes a second industrial robot name, a second model, a second time, and energy consumption; wherein the data format of the energy consumption data is {second industrial robot name, second model, second time, energy consumption}.

[0064] In one embodiment, since both the motion process data and the energy consumption data contain the name, model and time of the industrial robot, the motion process data and the energy consumption data are associated with each other through the name, model and time of the industrial robot to construct a motion-energy consumption time series data set.

[0065] Specifically, obtain the name, the first model and the first time of the first industrial robot in the motion process data, and simultaneously obtain the name, the second model and the second time of the second industrial robot in the energy consumption data; perform similarity judgment on the name, the first model and the first time of the first industrial robot with the name, the second model and the second time of the second industrial robot respectively; if the name of the first industrial robot is the same as the name of the second industrial robot, the first model is the same as the second model, and the first time is the same as the second time, then obtain the axis joint angle, the axis motor current, the axis joint torque and the axis temperature in the motion process data, and obtain the energy consumption in the energy consumption data; associate the axis joint angle, the axis motor current, the axis joint torque, the axis temperature and the energy consumption to generate a motion-energy consumption time series data set.

[0066] In one embodiment, after obtaining the motion-energy consumption time series data set, the motion-energy consumption time series data set is also traversed to determine whether there are data rows with missing data in the motion-energy consumption time series data set. If so, the data rows are deleted, and the motion-energy consumption time series data set is updated to obtain an updated motion-energy consumption time series data set; as shown in Table 1, Table 1 is a schematic table of the motion-energy consumption time series data set.

[0067] Table 1:

[0068] Step 102: performing working condition identification on the motion-energy consumption time series dataset based on a density clustering algorithm, and dividing the motion-energy consumption time series dataset into industrial robot datasets under different working conditions.

[0069] In one embodiment, motion time series data extraction is performed on the motion-energy consumption time series data set to obtain a motion time series data set.

[0070] Specifically, the motion state information of the industrial robot is extracted from the motion-energy consumption time series data set, wherein the motion state information includes one-axis joint angle, one-axis motor current, one-axis joint torque, one-axis shaft temperature, two-axis joint angle, two-axis motor current, two-axis joint torque, two-axis shaft temperature, three-axis joint angle, three-axis motor current, three-axis joint torque, three-axis shaft temperature, four-axis joint angle, four-axis motor current, four-axis joint torque, four-axis shaft temperature, five-axis joint angle, five-axis motor current, five-axis joint torque, five-axis shaft temperature, six-axis joint angle, six-axis motor current, six-axis joint torque and six-axis shaft temperature. Based on the motion state information, a motion time series data set is obtained.

[0071] Specifically, the expression of the motion time series data set is set to Y s×i , where s is the number of samples, i is the number of motion state variables, i.e., the independent variable dimension; preferably, the number of samples s is set to 10,000, and the number of motion state variables i is set to 24, i.e., the expression of the motion time series data set is Y 10000×24 .

[0072] In one embodiment, a kernel matrix of the motion time series data set is calculated based on a preset Gaussian kernel function, eigenvalues ​​and eigenvectors of the kernel matrix are obtained, and a feature contribution rate is calculated based on the eigenvalues ​​and the eigenvectors.

[0073] Specifically, the Gaussian kernel function is defined as: Wherein, x1 and x2 are any two motion state variables, K is a square matrix, and σ is a parameter that controls the width of the Gaussian kernel function; preferably, σ is set to 15.

[0074] Specifically, based on the preset Gaussian kernel function, when calculating the kernel matrix of the motion time series data set, the motion state variables in the motion time series data set are combined in pairs to obtain multiple motion state variable pairs, and the Gaussian kernel function is applied to each motion state variable pair to obtain the similarity between each motion state variable, and based on the similarity, the kernel matrix K of the motion time series data set is constructed. 24×24 .

[0075] Specifically, for the kernel matrix K 24×24 Perform eigenvalue decomposition to obtain eigenvalues ​​λ1,λ2,...,λ 24 and the corresponding eigenvectors v1,v2,...,v 24 ; eigenvector v i corresponds to the eigenvalue λ.

[0076] Specifically, based on the eigenvalues ​​and eigenvectors, the contribution rate of each eigenvalue is calculated, and the contribution rate corresponding to each eigenvalue is integrated to determine the characteristic contribution rate; wherein the contribution rate can be calculated by dividing each eigenvalue by the sum of all eigenvalues; for example, the contribution rate of the i-th eigenvalue is λ i / (λ1+λ2+...+λ 24 ).

[0077] In one embodiment, the feature contribution rate is compared with a preset feature contribution rate threshold, and if the feature contribution rate is not less than the feature contribution rate threshold, a principal component data set is obtained.

[0078] Specifically, the feature contribution rate threshold is set to 90%.

[0079] Specifically, the feature contribution rate represents the percentage of the total variance explained by the selected principal components; when the feature contribution rate reaches or exceeds a preset threshold, such as 90%, it can be considered that these principal components have captured most of the data information and can be used as the result of dimensionality reduction.

[0080] Specifically, when obtaining the principal component data set, the contribution rate of each eigenvalue is sorted in ascending order, and the sum of the contribution rates of the first k eigenvalues ​​is calculated as P = p1 + p2 + ... + p k When the cumulative contribution rate is greater than or equal to 90%, the number of principal components after dimensionality reduction is k, and the principal component data set obtained is Y′ 10000×k .

[0081] In one embodiment, the principal component dataset is input into a density clustering algorithm to obtain a working condition label dataset corresponding to the principal component dataset.

[0082] Specifically, the density clustering algorithm DBSCAN is used to cluster the working conditions of the principal component data set, and two parameters of the density clustering algorithm DBSCAN are set: radius eps=0.88 and minimum sample number MinPts=5.

[0083] Specifically, the clustering algorithm is based on the principal component data set Y′ 10000×k As input, the output is the working condition label dataset corresponding to each principal component dataset.

[0084] Example description of the working condition label dataset L 10000×1 =[1,1,1,2,3,3,2,―1,....,2,1,―1], where 1, 2, and 3 respectively indicate that the sample belongs to the first, second, and third working conditions, and -1 indicates that the sample is abnormal data.

[0085] Preferably, abnormal data is identified for the operating condition label data set corresponding to each principal component data set, and the identified abnormal data is deleted.

[0086] In one embodiment, the motion-energy consumption time series dataset is classified based on the working condition label dataset to obtain industrial robot datasets under different working conditions.

[0087] Specifically, when extracting the industrial robot dataset under the first working condition, we extract the labeled dataset L 10000×1 The median value is 1, which is the sample index position of the first working condition. The number of values ​​1 is q, and the industrial robot dataset X′ under the first working condition is obtained. q×25 ; For the remaining working conditions, the classification process of the corresponding industrial robot dataset is the same as that described for the first working condition.

[0088] Step 103: construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, train the industrial robot energy consumption prediction model based on the industrial robot dataset, and determine the optimal industrial robot energy consumption prediction model under different working conditions.

[0089] In one embodiment, an industrial robot energy consumption prediction model under different working conditions is constructed based on the LightGBM algorithm, wherein the industrial robot energy consumption prediction model under each working condition is provided with LightGBM hyperparameters, and the LightGBM hyperparameters include the maximum depth of the tree, the minimum leaf node sample weight, the learning rate, and the proportion of subsamples in the entire sample set.

[0090] Specifically, the maximum depth of the tree in the LightGBM hyperparameters is set to max_depth = 2, the minimum leaf node sample weight and min_child_weight = 0.5, the learning rate learning_rate = 0.005, and the proportion of subsamples in the entire sample set subsample = 0.7.

[0091] In one embodiment, the LightGBM hyperparameters in the industrial robot energy consumption prediction model under different working conditions are initialized to obtain an initialized industrial robot energy consumption prediction model.

[0092] In one embodiment, an industrial robot training dataset is extracted from the industrial robot dataset.

[0093] Specifically, the industrial robot dataset X′ q×25 The industrial robot training dataset Train is divided into different working conditions according to the ratio of 8:2. a×25 Industrial robot test dataset Test corresponding to different working conditions b×25 ; Where a = int (q × 0.8), b = q - a.

[0094] Specifically, when the initialized industrial robot energy consumption prediction model is trained, in the model training phase, the input and output industrial robot training data sets are Train and a×24 and Train a×1 ; In the model testing phase, the input and output industrial robot test data sets are Test b×24 and Test b×1 .

[0095] In one embodiment, the industrial robot training data set is input into the initialized industrial robot energy consumption prediction model, and the initialized industrial robot energy consumption prediction model is iteratively optimized. During the iterative optimization process, an adaptive random search algorithm is used to perform parameter optimization on the LightGBM hyperparameters in the initialized industrial robot energy consumption prediction model, and the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process is calculated until the preset maximum number of iterations is reached.

[0096] Specifically, the upper and lower limits of the four hyperparameters are set, where max_depth∈[2,100], min_child_weight∈[0.05,24], learning_rate∈[0.005,0.1], and subsample∈[0.7,1); and the maximum number of iterations of the model is set to iterations=300.

[0097] Specifically, the adaptive random search algorithm is used to optimize the parameters of the four LightGBM hyperparameters. The mean absolute percentage error of the model iteration i is MAPE i ,in,

[0098] Among them, pre b×1 is the predicted energy consumption dataset.

[0099] Specifically, set the fitness value to fitness = 1 / (1+MAPE i ), based on the mean absolute percentage error, calculate the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process.

[0100] Specifically, as the number of iterations increases, the optimization model searches for four LightGBM hyperparameters in the direction of increasing fitness value (that is, decreasing mean absolute percentage error); if the maximum number of iterations = 300 is reached, the parameter optimization is stopped.

[0101] In one embodiment, the fitness value corresponding to each iterative optimization process is obtained, the first LightGBM hyperparameter corresponding to the maximum fitness value is selected, and based on the first LightGBM hyperparameter, the optimal industrial robot energy consumption prediction model under different working conditions is determined.

[0102] Specifically, the four LightGBM hyperparameters corresponding to the maximum fitness value in 300 iterations, that is, the four LightGBM hyperparameters corresponding to the minimum mean absolute percentage error, are obtained as the first LightGBM hyperparameters; the format of setting the first LightGBM hyperparameters is {max_depth j ,min_child_weight j ,learning_rate j ,subsample j}, where j represents the jth operating condition.

[0103] Specifically, the first LightGBM hyperparameter is used as the hyperparameter after optimizing the industrial robot energy consumption prediction model under different working conditions constructed by the LightGBM algorithm by an adaptive random search algorithm to obtain the optimal industrial robot energy consumption prediction model under different working conditions.

[0104] In one embodiment, as the amount of data from different working conditions increases, the industrial robot energy consumption prediction model is continuously iterated. When the average relative deviation between the real-time measurement results of the industrial robot energy consumption prediction model and the results collected in real time by the industrial intelligent sensor is within a preset range, that is, the prediction accuracy of the industrial robot energy consumption prediction model is high, the industrial intelligent sensor can be removed and data resources for the industrial robot will no longer be collected, thereby reducing the data collection cost.

[0105] Step 104: Obtain real-time motion process data of the industrial robot under the target working condition, and input the real-time motion process data into the target optimal industrial robot energy consumption prediction model corresponding to the target working condition, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

[0106] In one embodiment, when the target working condition is the first working condition, the real-time motion process data of the industrial robot under the first working condition is obtained, and the real-time motion process data is input into the target optimal industrial robot energy consumption prediction model corresponding to the first working condition, so that the target optimal industrial robot energy consumption prediction model outputs the real-time energy consumption prediction data of the industrial robot under the first working condition; as shown in Figure 3, Figure 3 is a schematic diagram of the energy consumption prediction effect.

[0107] In one embodiment, when the target working condition is other working conditions, it is only necessary to select the target optimal industrial robot energy consumption prediction model corresponding to the current working condition to perform real-time energy consumption prediction of the industrial robot.

[0108] In one embodiment, after obtaining the optimal industrial robot energy consumption prediction model under different working conditions, the optimal industrial robot energy consumption prediction model can be deployed in different working conditions of its production site to measure the real-time energy consumption data of the industrial robot.

[0109] Example 2, referring to FIG2 , is a schematic structural diagram of an embodiment of an industrial robot energy consumption prediction device provided by the present invention. As shown in FIG2 , the device includes an industrial robot data association module 201, a data set partitioning module 202, a model training module 203, and an energy consumption data prediction module 204, specifically as follows:

[0110] The industrial robot data association module 201 is used to obtain motion process data and energy consumption data of the industrial robot, perform association processing on the motion process data and the energy consumption data, and generate a motion-energy consumption time series data set.

[0111] The data set division module 202 is configured to perform working condition identification on the motion-energy consumption time series data set based on a density clustering algorithm, and divide the motion-energy consumption time series data set into industrial robot data sets under different working conditions.

[0112] The model training module 203 is used to construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, perform model training on the industrial robot energy consumption prediction model based on the industrial robot dataset, and determine the optimal industrial robot energy consumption prediction model under different working conditions.

[0113] The energy consumption data prediction module 204 is used to obtain the real-time motion process data of the industrial robot under the target working conditions, and input the real-time motion process data into the target optimal industrial robot energy consumption prediction model corresponding to the target working conditions, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

[0114] In one embodiment, the industrial robot data association module 201 is used to obtain the motion process data and energy consumption data of the industrial robot, wherein the motion process data includes the name, first model, first time, axis joint angle, axis motor current, axis joint torque, and axis temperature of the first industrial robot; and the energy consumption data includes the name, second model, second time, and energy consumption of the second industrial robot.

[0115] In one embodiment, the industrial robot data association module 201 is used to associate the motion process data and the energy consumption data to generate a motion-energy consumption time series data set, specifically including: obtaining the first industrial robot name, the first model and the first time in the motion process data, and simultaneously obtaining the second industrial robot name, the second model and the second time in the energy consumption data; performing similarity judgment on the first industrial robot name, the first model and the first time with the second industrial robot name, the second model and the second time respectively; if the first industrial robot name is the same as the second industrial robot name, the first model is the same as the second model, and the first time is the same as the second time, then obtaining the axis joint angle, the axis motor current, the axis joint torque and the axis temperature in the motion process data, and obtaining the energy consumption in the energy consumption data; associating the axis joint angle, the axis motor current, the axis joint torque, the axis temperature and the energy consumption to generate a motion-energy consumption time series data set.

[0116] In one embodiment, the data set division module 202 is used to perform working condition identification on the motion-energy consumption time series data set based on a density clustering algorithm, and divide the motion-energy consumption time series data set into industrial robot data sets under different working conditions, specifically including: extracting motion time series data from the motion-energy consumption time series data set to obtain a motion time series data set; calculating the kernel matrix of the motion time series data set based on a preset Gaussian kernel function, obtaining the eigenvalues ​​and eigenvectors of the kernel matrix, and calculating the feature contribution rate based on the eigenvalues ​​and the eigenvectors; comparing the feature contribution rate with a preset feature contribution rate threshold, and if the feature contribution rate is not less than the feature contribution rate threshold, obtaining a principal component data set; inputting the principal component data set into a density clustering algorithm to obtain a working condition label data set corresponding to the principal component data set; and classifying the motion-energy consumption time series data set based on the working condition label data set to obtain industrial robot data sets under different working conditions.

[0117] In one embodiment, the model training module 203 is used to construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, specifically including: constructing an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, wherein the industrial robot energy consumption prediction model under each working condition is provided with LightGBM hyperparameters, and the LightGBM hyperparameters include the maximum depth of the tree, the minimum leaf node sample weight, the learning rate, and the proportion of subsamples in the entire sample set.

[0118] In one embodiment, the model training module 203 is used to perform model training on the industrial robot energy consumption prediction model based on the industrial robot data set to determine the optimal industrial robot energy consumption prediction model under different working conditions, specifically including: initializing the LightGBM hyperparameters in the industrial robot energy consumption prediction model under different working conditions to obtain an initialized industrial robot energy consumption prediction model; extracting an industrial robot training data set from the industrial robot data set, inputting the industrial robot training data set into the initialized industrial robot energy consumption prediction model, iteratively optimizing the initialized industrial robot energy consumption prediction model, and during the iterative optimization process, using an adaptive random search algorithm to perform parameter optimization on the LightGBM hyperparameters in the initialized industrial robot energy consumption prediction model, calculating the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process until a preset maximum number of iterations is reached; obtaining the fitness value corresponding to each iterative optimization process, selecting the first LightGBM hyperparameter corresponding to the maximum fitness value, and determining the optimal industrial robot energy consumption prediction model under different working conditions based on the first LightGBM hyperparameter.

[0119] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0120] It should be noted that the above-described embodiment of the device for predicting industrial robot energy consumption is merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of these modules may be selected to achieve the objectives of this embodiment as needed.

[0121] Based on the above-mentioned embodiment of the method for predicting the energy consumption of an industrial robot, another embodiment of the present invention provides a terminal device for predicting the energy consumption of an industrial robot. The terminal device for predicting the energy consumption of an industrial robot includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method for predicting the energy consumption of an industrial robot according to any embodiment of the present invention is implemented.

[0122] For example, in this embodiment, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device for predicting industrial robot energy consumption.

[0123] The terminal device for predicting the energy consumption of the industrial robot may be a computing device such as a desktop computer, a notebook computer, a palmtop computer, a cloud server, etc. The terminal device for predicting the energy consumption of the industrial robot may include, but is not limited to, a processor and a memory.

[0124] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device for predicting the energy consumption of the industrial robot, and utilizes various interfaces and lines to connect various parts of the terminal device for predicting the energy consumption of the industrial robot.

[0125] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the industrial robot energy consumption prediction terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created based on the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0126] Based on the above-mentioned embodiment of the method for predicting the energy consumption of an industrial robot, another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the method for predicting the energy consumption of an industrial robot according to any embodiment of the present invention.

[0127] In this embodiment, the storage medium is a computer-readable storage medium, and the computer program includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0128] In summary, the present invention provides a method, device, equipment and storage medium for predicting the energy consumption of an industrial robot. The method generates a motion-energy consumption time series data set by correlating the obtained motion process data and energy consumption data of the industrial robot; the motion-energy consumption time series data set is divided into industrial robot data sets under different working conditions based on the density clustering algorithm; and an industrial robot energy consumption prediction model under different working conditions is constructed based on the LightGBM algorithm. The industrial robot energy consumption prediction model is trained based on the industrial robot data set to determine the optimal industrial robot energy consumption prediction model under different working conditions; the acquired real-time motion process data is input into the target optimal industrial robot energy consumption prediction model corresponding to the target working condition, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data; compared with the prior art, the technical solution of the present invention can improve the measurement accuracy of the energy consumption of the industrial robot and reduce the cost of energy consumption measurement.

[0129] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention. These improvements and substitutions should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting the energy consumption of an industrial robot, characterized in that, Including: Obtain the motion process data and energy consumption data of an industrial robot, perform correlation processing on the motion process data and the energy consumption data to generate a motion-energy consumption time series dataset; Based on the density clustering algorithm, perform working condition identification on the motion-energy consumption time series dataset, and divide the motion-energy consumption time series dataset into industrial robot datasets under different working conditions; Based on the LightGBM algorithm, construct an industrial robot energy consumption prediction model under different working conditions, and perform model training on the industrial robot energy consumption prediction model based on the industrial robot dataset to determine the optimal industrial robot energy consumption prediction model under different working conditions; Obtain the real-time motion process data of the industrial robot under the target working condition, and input the real-time motion process data into the target optimal industrial robot energy consumption prediction model corresponding to the target working condition, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

2. The prediction method of the energy consumption of an industrial robot according to claim 1, wherein Obtain the motion process data and energy consumption data of an industrial robot, where the motion process data includes the name of the first industrial robot, the first model, the first time, the axis joint angle, the axis motor current, the axis joint torque, and the axis temperature; the energy consumption data includes the name of the second industrial robot, the second model, the second time, and the energy consumption.

3. The prediction method for the energy consumption of an industrial robot according to claim 2, wherein, Perform correlation processing on the motion process data and the energy consumption data to generate a motion-energy consumption time series dataset, specifically including: Obtain the name of the first industrial robot, the first model, and the first time in the motion process data, and at the same time obtain the name of the second industrial robot, the second model, and the second time in the energy consumption data; Perform similarity judgment on the name of the first industrial robot, the first model, and the first time respectively with the name of the second industrial robot, the second model, and the second time; If the name of the first industrial robot is the same as the name of the second industrial robot, the first model and the second model are the same, and the first time and the second time are the same, then obtain the axis joint angle, the axis motor current, the axis joint torque, and the axis temperature in the motion process data, and obtain the energy consumption in the energy consumption data; Correlate the axis joint angle, the axis motor current, the axis joint torque, the axis temperature, and the energy consumption to generate a motion-energy consumption time series dataset.

4. The prediction method for the energy consumption of an industrial robot according to claim 1, characterized in that, Based on the density clustering algorithm, perform working condition identification on the motion-energy consumption time series dataset, and divide the motion-energy consumption time series dataset into industrial robot datasets under different working conditions, specifically including: Extract motion time series data from the motion-energy consumption time series dataset to obtain a motion time series dataset; Based on a preset Gaussian kernel function, calculate the kernel matrix of the motion time series dataset, obtain the eigenvalues and eigenvectors of the kernel matrix, and calculate the feature contribution rate based on the eigenvalues and the eigenvectors; Compare the feature contribution rate with a preset feature contribution rate threshold. If the feature contribution rate is not less than the feature contribution rate threshold, obtain the principal component dataset; Input the principal component dataset into the density clustering algorithm to obtain the working condition label dataset corresponding to the principal component dataset; Based on the working condition label dataset, classify the motion-energy consumption time series dataset to obtain industrial robot datasets under different working conditions.

5. The prediction method of the energy consumption of an industrial robot according to claim 1, wherein Construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, specifically including: Construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm. Among them, LightGBM hyperparameters are set in the industrial robot energy consumption prediction model under each working condition. The LightGBM hyperparameters include the maximum depth of the tree, the sum of the sample weights of the minimum leaf nodes, the learning rate, and the proportion of the subsample in the entire sample set.

6. The prediction method for the energy consumption of an industrial robot according to claim 5, wherein, Based on the industrial robot dataset, train the industrial robot energy consumption prediction model to determine the optimal industrial robot energy consumption prediction model under different working conditions, specifically including: Perform initialization processing on the LightGBM hyperparameters in the industrial robot energy consumption prediction model under different working conditions to obtain an initialized industrial robot energy consumption prediction model; Extract an industrial robot training dataset from the industrial robot dataset, input the industrial robot training dataset into the initialized industrial robot energy consumption prediction model, perform iterative optimization processing on the initialized industrial robot energy consumption prediction model, and during the iterative optimization process, use the adaptive random search algorithm to perform parameter optimization on the LightGBM hyperparameters in the initialized industrial robot energy consumption prediction model, and calculate the fitness value of the initialized industrial robot energy consumption prediction model corresponding to each iterative optimization process until the preset maximum number of iterations is reached; Obtain the fitness value corresponding to each iterative optimization process, select the first LightGBM hyperparameters corresponding to the maximum fitness value, and based on the first LightGBM hyperparameters, determine the optimal industrial robot energy consumption prediction model under different working conditions.

7. A prediction device for the energy consumption of an industrial robot, characterized in that, Including: An industrial robot data association module, a dataset division module, a model training module, and an energy consumption data prediction module; Among them, the industrial robot data association module is used to obtain the motion process data and energy consumption data of the industrial robot, perform association processing on the motion process data and the energy consumption data, and generate a motion-energy consumption time series dataset; The dataset division module is used to perform working condition identification on the motion-energy consumption time series dataset based on the density clustering algorithm, and divide the motion-energy consumption time series dataset into industrial robot datasets under different working conditions; The model training module is used to construct an industrial robot energy consumption prediction model under different working conditions based on the LightGBM algorithm, and train the industrial robot energy consumption prediction model based on the industrial robot dataset to determine the optimal industrial robot energy consumption prediction model under different working conditions; The energy consumption data prediction module is used to obtain the real-time motion process data of the industrial robot under the target working condition, and input the real-time motion process data into the target optimal industrial robot energy consumption prediction model corresponding to the target working condition, so that the target optimal industrial robot energy consumption prediction model outputs real-time energy consumption prediction data.

8. The prediction device for the energy consumption of an industrial robot according to claim 7, wherein The industrial robot data association module is used to obtain the motion process data and energy consumption data of the industrial robot. Among them, the motion process data includes the first industrial robot name, the first model, the first time, the axis joint angle, the axis motor current, the axis joint torque, and the axis temperature; the energy consumption data includes the second industrial robot name, the second model, the second time, and the energy consumption.

9. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the prediction method for the energy consumption of the industrial robot according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program. Among them, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the prediction method for the energy consumption of the industrial robot according to any one of claims 1 to 6.

Images