Intelligent analysis method and system for use state of forklift

Abstract

The invention relates to the technical field of state monitoring, in particular to an intelligent analysis method and system for the use state of a forklift, and the method comprises the following steps: based on the operation data of the forklift, extracting characteristic parameters including the vehicle speed, the fork load and the vehicle body attitude angle, and carrying out the interval division of the fluctuation range of the characteristic parameters; according to the method, through fluctuation range interval division and normalization processing, the comparison and stability of characteristic parameters are optimized, weighted distance calculation and dynamic weight adjustment are adopted, the sample classification precision is improved, the coupling interference among parameters in operation data is effectively weakened, and through combination of merging clustering of density distribution values and density peak values, the classification accuracy is improved. According to the method, the diversity of operation modes is accurately identified, the resolution capability for complex operation behaviors is enhanced, accurate division of multi-stage states is realized by extracting weighted feature values of operation stages and calculating the volatility, the systematization capability of abnormal operation mode identification and operation behavior feature classification is enhanced, and multi-dimensional decision support is provided for equipment management.

Description

Technical Field

[0001] The present invention relates to the technical field of state monitoring, and in particular to a method and system for intelligently analyzing the use state of a forklift. Background Art

[0002] The field of condition monitoring technology includes monitoring and data analysis of the operating status of various types of equipment or systems in order to obtain their operating parameters and behavior patterns. The core content of this technical field is to obtain information on equipment status, fault warning, operating efficiency, etc. through the collection, analysis and processing of equipment or system operating data. Condition monitoring technology usually covers sensor data collection, signal transmission, status data analysis, information storage and processing, forming a systematic data monitoring and management method. By systematically building a monitoring platform that integrates hardware and software, real-time grasp and analysis of equipment status information can be achieved, providing technical support for the operation and management of various types of equipment.

[0003] Among them, the forklift uses an intelligent state analysis method to obtain the forklift's operating parameters (position, speed, load, etc.) in real time through data acquisition equipment, and uses data processing methods to make state judgments, identify operating modes, and identify abnormalities to achieve multi-dimensional state analysis. This technology is based on system data from forklift controllers, ECUs and other modules to obtain equipment status, operating efficiency, fault information, etc. Through sensor data acquisition, signal transmission, state analysis and storage, a hardware and software integrated monitoring platform is built to achieve real-time monitoring, analysis, module testing, updating, and adjustment of the forklift status, providing data support for vehicle maintenance and controller diagnosis.

[0004] The existing technology does not fully consider the impact of the fluctuation range when processing feature parameters, which easily leads to insufficient accuracy in the expression of parameter features. The sample classification method lacks the ability to dynamically adjust weights, so that the importance of different features in the operating parameters cannot be reasonably distributed as the scene changes, affecting the recognition effect of complex scenes. In the analysis of operating modes, there is a lack of dynamic clustering methods based on density distribution, which makes it difficult to capture the characteristic distribution laws of diversified operating modes. The division of operating stages mainly relies on static thresholds, which cannot effectively analyze the changing trends between related features, limit the early detection and accurate judgment of abnormal conditions, and reduce the accuracy and reliability of condition monitoring and management. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a forklift use status intelligent analysis method and system.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for intelligent analysis of forklift usage status, comprising the following steps: S1: Based on the forklift operation data, feature parameters including vehicle speed, fork load, and vehicle body posture angle are extracted, the fluctuation range of the feature parameters is divided into intervals, the fluctuation mean and variance of multiple intervals are calculated, all feature parameters are normalized, and the initial feature weight value is generated; S2: Based on the initial feature weight value, calculate the weighted distance of each forklift operation sample point on the vehicle speed sensor and the position change rate, calculate the weighted average distance of samples within the class and the weighted average distance of samples between classes, normalize the difference between the two and adjust the feature weight to generate a dynamic feature weight value; S3: Based on the dynamic feature weight value, the weighted density distribution value of the forklift operation sample point is calculated, the density peak value is determined by combining the vehicle speed and the fork load feature value, the density peak value is merged and clustered, the cluster samples are re-divided, the weighted density value of the cluster center is calculated, and the operation mode clustering result is generated; S4: Based on the operation mode clustering result, the sample groups are divided according to the operation stages of acceleration, turning, and stable operation in the forklift operation data, and the weighted feature values ​​of the vehicle body posture angle and position change rate in each stage are extracted. The fluctuation rate of the weighted feature values ​​is calculated and the weights are readjusted to generate the staged operation mode classification results; S5: Based on the classification results of the staged operation modes, the fluctuation value ranges of the fork load and the vehicle body posture angle in the differentiated operation stages are extracted, the fluctuation value distribution of each type of sample is classified and counted, the distribution mean and the fluctuation range are calculated, and a forklift usage status classification indicator set is generated.

[0007] The initial feature weight value specifically includes the feature parameter fluctuation range, multi-interval fluctuation mean, variance, and normalized feature parameters. The dynamic feature weight value includes weighted distance, average distance between samples within and outside the class, and difference normalization. The operation mode clustering result specifically refers to density peak, merged clustering, and weighted density of cluster center. The staged operation mode classification result includes stage division, volatility calculation, and weight adjustment. The forklift usage status classification indicator set specifically includes the fork load fluctuation range, body posture angle fluctuation range, fluctuation value distribution statistics, distribution mean, and fluctuation interval.

[0008] As a further solution of the present invention, the step of obtaining the initial feature weight value is specifically as follows: S111: extracting characteristic parameters of vehicle speed, fork load and vehicle body posture angle from the forklift operation data, analyzing the fluctuation range of the characteristic parameters item by item according to the discrete interval rule, dividing the data set in each interval into separate areas, and generating fluctuation mean and fluctuation variance by performing item by item mean operation and variance calculation on sample points in the area; S112: Normalize the fluctuation mean and fluctuation variance using the formula: ; Calculate and generate a normalized feature parameter matrix; in, Represents the normalized feature matrix Line The eigenvalues ​​of the columns, Represents the original feature matrix Line The eigenvalues ​​of the columns, Representative The mean of the column features, Representative The variance of the column features, represents the smoothing coefficient used to avoid the denominator being zero; S113: Calculate the inter-column mean of the normalized feature parameter matrix, analyze its initial weight value column by column, use the matrix column mean vector calculation as the initial weight vector, and generate the initial feature weight value based on the column mean and the deviation superposition processing between the matrix column multiple eigenvalues.

[0009] As a further solution of the present invention, the step of obtaining the dynamic feature weight value is specifically as follows: S211: performing parameter weighting processing on the running sample points of the vehicle speed sensor and the position change rate according to the initial feature weight value, calculating the weight distribution of the vehicle speed and position change rate parameters on the sample points by calling the initial weight value, and calculating the accumulated square difference value of each sample point on the weighted distance, to obtain a weighted sample point distance set; S212: separating the weighted distance of sample points within a class and the weighted distance of sample points between classes from the weighted sample point distance set, and calculating the weighted mean values ​​of sample points within a class and between classes respectively according to the classes to which the sample points belong by calling the weighted distance information of the sample points within the set, and gradually aggregating the sample class information to obtain the weighted average distance of samples within a class and the weighted average distance of samples between classes, and generating a sample weighted average distance parameter set; S213: Based on the sample weighted average distance parameter set, the difference between the weighted average distances of samples between classes and within classes is calculated point by point, using the formula: ; Calculate and generate dynamic feature weight values; in, represents the dynamic feature weight value, Indicates The weighted average distance within the class of samples, Indicates The weighted average distance between classes of samples, Indicates The initial feature weight value of the sample point, Dynamic adjustment coefficients are used for smooth weight calculations. represents the total number of samples, Used to adjust the influence of sample point weight on difference normalization. Represents the normalized sum of weight differences.

[0010] As a further solution of the present invention, the step of obtaining the operation mode clustering result is specifically: S311: calling the dynamic feature weight value, calculating the weighted density distribution value of the forklift operation sample point, performing weighted density calculation on each sample point through the vehicle speed and fork load feature value, accumulating the sample point density value point by point, and performing density classification processing in combination with the feature distribution parameter to generate a weighted density distribution value set; S312: Combine the weighted density distribution value set, compare the distribution values ​​of the vehicle speed and fork load characteristic values ​​with the overall density distribution point by point, and use the formula: ; Determine the local density peak of each sample point and generate a density peak set; in, represents the density peak, Indicates The weighted density value of sample points, represents the dynamic weight of vehicle speed characteristics, The dynamic weight representing the fork load characteristics, To adjust the parameters used to adjust the smoothness of the weight to density calculation, Indicates the dynamic smoothing effect of the fork load weight on the density value, The operation selects the local density maximum as the density peak; S313: performing merging clustering on the density peak set, determining the weighted distance distribution between the sample points by calculating the weighted density difference between the density peak of each sample point and other sample points, and reclassifying the clustering results by gradually aggregating the sample point attribution information to generate cluster sample division; S314: Call the cluster sample division, calculate the weighted density value of each cluster center, perform weighted summation of the vehicle speed and fork load characteristic values ​​of the sample points belonging to the cluster center in sequence, and dynamically adjust the center characteristics in combination with the mean distribution of the sample density values ​​within the class to generate the operation mode clustering result.

[0011] As a further solution of the present invention, the step of obtaining the classification result of the staged operation mode is specifically as follows: S411: calling the operation mode clustering result, dividing the sample groups item by item according to the acceleration, turning and stable operation stages in the forklift operation data, extracting the sample point position parameters and attitude angle feature values ​​of the operation stage, performing operation feature grouping processing on the sample groups in turn, and generating stage-by-stage sample group division; S412: extracting the characteristic values ​​of the vehicle body posture angle and position change rate in each stage of the phased sample group division, weighting the characteristic values ​​of the sample points point by point by combining the dynamic characteristic weight values, and performing weighted accumulation processing on the characteristic data of the running stage to generate a phased weighted characteristic value set; S413: Calculate the volatility of the phased weighted eigenvalue set, and readjust the weights using the formula: ; Calculate and generate the re-adjusted weight parameters; in, represents the re-adjusted weight parameter, represents the volatility of the weighted eigenvalue set, represents the mean of the weighted eigenvalue set, Represents the minimum fluctuation threshold, which is used to adjust the flexibility and sensitivity of the weights and increase the formula's ability to respond to small fluctuations; S414: Call the readjusted weight parameters, re-evaluate and classify the weighted feature values ​​of each stage in combination with the phased sample group division information, calculate the comprehensive feature parameters of the operation stage by classification, classify the readjusted classification features of the operation stage in turn, and generate the phased operation mode classification results.

[0012] As a further solution of the present invention, the steps of obtaining the forklift use status classification index set are specifically as follows: S511: calling the classification result of the staged operation mode, extracting the fluctuation value range of the fork load and the body posture angle in the differentiated operation stage, and gradually summarizing and classifying the fluctuation value range of the sample points by sequentially screening the dynamic change values ​​of the fork load and the body posture angle parameters for the sample points in the operation stage, and generating a sample fluctuation value set for the operation stage; S512: extracting the fluctuation value distribution of each type of samples in the sample fluctuation value set of the operation stage, calculating the fluctuation value interval characteristics of the samples in turn and combining the distribution range of each type of samples for classification statistics, and gradually classifying and summarizing the characteristic parameters of the fluctuation range of the sample points to generate the classification statistics results of the sample fluctuation value distribution; S513: Call the sample fluctuation value distribution classification statistics result, calculate the fluctuation mean and fluctuation range of each type of sample distribution, and perform weighted operation on the fluctuation value features point by point and use the formula: ; Calculate and generate sample distribution mean and fluctuation range; in, represents the weighted fluctuation value of the sample distribution, Indicates The fluctuation value of the samples, Represents the mean of the fluctuation values ​​of all sample points, Indicates The dynamic weight of the sample points, The nonlinear effect of adjusting the weights for the smoothing coefficients, represents the smoothing effect of dynamically adjusting weights on distribution characteristics, Indicates the total number of samples; S514: calling the sample distribution mean and fluctuation range, and gradually classifying and summarizing the fluctuation value range and mean characteristic parameters of each type of sample in combination with the classification statistics results in the operation stage, recalculating the classification index for the dynamic change parameter characteristics of each type of operation stage sample, and generating a forklift usage status classification index set.

[0013] A forklift use status intelligent analysis system, the forklift use status intelligent analysis system is used to execute the forklift use status intelligent analysis method, the system comprises: The feature extraction module extracts feature values ​​from vehicle speed, fork load, and vehicle body posture angle based on forklift operation data, calculates the fluctuation mean and variance, normalizes the feature values, and obtains the initial feature weight value; The dynamic weight adjustment module calculates the weighted distance of the sample points based on the initial feature weight value, obtains the weighted average distance of samples inside and outside the class, performs normalization adjustment through the difference, updates the weight of each feature, and generates a dynamic feature weight value; The cluster analysis module calculates the weighted density distribution of the forklift operation sample points based on the dynamic feature weight values, determines the density peaks, performs merging clustering of the density peaks, calculates the weighted density of the cluster centers, and generates the operation mode clustering results; The operation mode classification module divides the forklift operation data into acceleration, turning, and stable operation stages based on the operation mode clustering results, extracts weighted features of the vehicle body posture angle and position change rate, calculates the fluctuation rate and adjusts the weight, and generates a staged operation mode classification result; The state classification index generation module extracts the fluctuation range of the fork load and the body posture angle in the differentiated operation stage based on the staged operation mode classification result, performs statistics on the fluctuation value distribution, calculates the distribution mean and interval, and generates a forklift use state classification index set.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by extracting characteristic parameters such as vehicle speed, fork load and vehicle body posture angle, dividing the fluctuation range interval and normalizing the process, the contrast and stability of the characteristic parameters are optimized. Weighted distance calculation and dynamic weight adjustment are used to improve the sample classification accuracy and effectively weaken the coupling interference between parameters in the operation data. Combined with the merging and clustering of density distribution values ​​and density peaks, the diversity of operation modes can be accurately identified and the ability to distinguish complex operating behaviors can be enhanced. By extracting the weighted characteristic values ​​of the operation stage and calculating the volatility, the accurate division of multi-stage states is achieved, which enhances the systematic ability to identify abnormal operation modes and classify operation behavior characteristics, providing multi-dimensional decision support for equipment management. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 This is a flow chart of the steps for obtaining the initial feature weight value of the present invention; Figure 3 This is a flow chart of the steps for obtaining the dynamic feature weight value of the present invention; Figure 4 A flow chart of the steps for obtaining the clustering results of the operation modes of the present invention; Figure 5 A flow chart of the steps for obtaining the classification results of the phased operation mode of the present invention; Figure 6 This is a flow chart of the steps for obtaining the forklift usage status classification indicator set of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] Embodiment 1 See also Figure 1 The present invention provides a technical solution: a method for intelligently analyzing the use status of a forklift, comprising the following steps: S1: Based on the forklift operation data, feature parameters including vehicle speed, fork load, and vehicle body posture angle are extracted, the fluctuation range of the feature parameters is divided into intervals, the fluctuation mean and variance of multiple intervals are calculated, all feature parameters are normalized, and the initial feature weight value is generated; S2: Based on the initial feature weight value, calculate the weighted distance of each forklift operation sample point on the speed sensor and the position change rate, calculate the weighted average distance of samples within the class and the weighted average distance of samples between classes, normalize the difference between the two and adjust the feature weight to generate a dynamic feature weight value; S3: Based on the dynamic feature weight value, the weighted density distribution value of the forklift operation sample point is calculated, and the density peak value is determined by combining the vehicle speed and fork load feature value. The density peak value is merged and clustered, and the cluster samples are re-divided. The weighted density value of the cluster center is calculated to generate the operation mode clustering result; S4: Based on the operation mode clustering results, the sample groups are divided according to the operation stages of acceleration, turning, and stable operation in the forklift operation data, and the weighted eigenvalues ​​of the vehicle body posture angle and position change rate in each stage are extracted. The volatility of the weighted eigenvalues ​​is calculated and the weights are readjusted to generate the staged operation mode classification results; S5: Based on the classification results of the staged operation mode, the fluctuation value range of the fork load and the vehicle body posture angle in the differentiated operation stage is extracted, the fluctuation value distribution of each type of sample is classified and counted, the distribution mean and fluctuation range are calculated, and the forklift usage status classification indicator set is generated.

[0019] The initial feature weight values ​​specifically include the feature parameter fluctuation range, multi-interval fluctuation mean, variance, and normalized feature parameters. The dynamic feature weight values ​​include weighted distance, average distance between samples within and outside the class, and difference normalization. The operation mode clustering results specifically refer to density peak, merged clustering, and weighted density of cluster center. The staged operation mode classification results include stage division, volatility calculation, and weight adjustment. The forklift usage status classification indicator set specifically includes the fork load fluctuation range, body posture angle fluctuation range, fluctuation value distribution statistics, distribution mean, and fluctuation interval.

[0020] See also Figure 2 , the specific steps for obtaining the initial feature weight value are: S111: extracting characteristic parameters of vehicle speed, fork load and vehicle body posture angle from the forklift operation data, analyzing the fluctuation range of the characteristic parameters item by item according to the discrete interval rule, dividing the data set in each interval into separate areas, and generating fluctuation mean and fluctuation variance by performing item by item mean operation and variance calculation on sample points in the area; The vehicle speed, fork load and vehicle body posture angle parameters recorded during the operation of the forklift are sampled and extracted according to the time series. The extreme value, mean and fluctuation range in the time series sampling data are divided into intervals. For the data in each interval after the division, the fluctuation mean and fluctuation variance in each interval are calculated. Combining the fluctuation mean and fluctuation variance in each interval, a fluctuation parameter set with interval as the unit is gradually constructed to generate the fluctuation mean and fluctuation variance. More specifically, for the characteristic parameters of the forklift, such as vehicle speed, fork load, vehicle body posture angle, steering angle torque, etc., it is necessary to divide the fluctuation range of the characteristic parameters into intervals. When collecting data, first define the basic value range of each parameter. For example, the vehicle speed is generally between 0-25km / h, and the fork load can vary between 0-5000kg. The value of the vehicle attitude angle is usually between -10° and 10°, while the steering angle torque is between 0-200Nm. Then, the time series data of each parameter is analyzed to calculate the maximum, minimum and mean values ​​in each time segment. For example, within a minute, the vehicle speed may vary between 12km / h and 18km / h, then its mean is (12+18) / 2=15km / h. The variance calculation needs to sum the deviations of each data point relative to the mean and divide it by the total number of data points. For example, if the vehicle speed data points in this time period are 12, 14, 16, and 18km / h respectively, then its variance is [((12-15)²+(14-15)²+(16-15)²+( 18-15)²) / 4]=4.5; for the fork load, if the data in a certain time period is 500kg, 800kg, 1200kg, 1000kg, the mean is (500+800+1200+1000) / 4=875kg, and the variance is calculated in the same way; for the vehicle attitude angle, assuming that the values ​​in a certain time period are -3°, 0°, 2°, 5°, the mean is (-3+0+2+5) / 4=1°, and the variance is [((-3-1)²+(0-1)²+(2-1)²+(5-1)²) / 4]=6; through the above calculations, the fluctuation of each characteristic parameter can be quantified to ensure the accuracy of the subsequent normalization processing.

[0021] S112: Normalize the volatility mean and volatility variance using the formula: ; Calculate and generate a normalized feature parameter matrix; in, Represents the normalized feature matrix Line The eigenvalues ​​of the columns, Represents the original feature matrix Line The eigenvalues ​​of the columns, Representative The mean of the column features, Representative The variance of the column features, represents the smoothing coefficient used to avoid the denominator being zero; formula: ; The formula is useful in that by introducing the smoothing parameter It avoids calculation errors caused by a denominator of zero, and combines the mean and variance of each feature to achieve normalization, so that different features can be calculated at the same scale, improving the comparability of data.

[0022] Detailed explanation of the formula and the process of formula calculation: For the characteristic parameter matrix Each element of , the sampling data of each second in the forklift operation record is used as the input data, and the mean is calculated for the same column of data (i.e. the same characteristic parameter column) and variance ,in is the total number of data samples, and the normalized value of each element is calculated using the mean and variance , such as , , , , substituting it into the formula: ; Calculate the normalized values ​​of all elements in the matrix in turn to generate a normalized feature parameter matrix.

[0023] This result shows that by normalizing the feature parameter matrix, we can further calculate the weights between features and analyze the influence of each feature on the final target.

[0024] S113: Calculate the column mean of the normalized feature parameter matrix, analyze its initial weight value column by column, use the matrix column mean vector calculation as the initial weight vector, and generate the initial feature weight value based on the column mean and the deviation superposition processing between the multiple eigenvalues ​​of the matrix column.

[0025] For the normalized data values ​​of each column in the matrix, the normalized values ​​are added and averaged in turn, using the formula Calculate column mean weights where is the number of data points in the column, combined with the column mean weight, and the weight adjustment is calculated by the deviation between the column mean and each eigenvalue in the matrix. The adjustment is calculated by the formula Implementation, where is the matrix normalization value, The column mean weight is used to gradually generate the adjusted weight vector. After the weight adjustment is completed, the column accumulation operation is combined with the adjusted weight vector and normalized to generate the initial feature weight value.

[0026] See also Figure 3 , the specific steps for obtaining the dynamic feature weight value are: S211: performing parameter weighting processing on the running sample points of the vehicle speed sensor and the position change rate according to the initial feature weight value, calculating the weight distribution of the vehicle speed and position change rate parameters on the sample points by calling the initial weight value, and calculating the accumulated square difference value of each sample point on the weighted distance, to obtain a weighted sample point distance set; By extracting the real-time monitoring data of the vehicle speed sensor and the time series data corresponding to the position change rate, a weight mapping for each sample point is established, and the initial weight value is called to calculate the cumulative value of the square difference of each sample point on the weighted distance point by point. The extraction process of vehicle speed data is completed through continuous monitoring of the sensor, and the vehicle speed change data is recorded at intervals of 5 seconds to generate a sample point sequence, in which each sample point data is the current speed value, and the position change rate is calculated by recording the spatial coordinate change of each sample point. The formula Calculate the rate of change of position, is the spatial displacement between the current sample point and the sample point at the previous moment, The time interval between two records is obtained. After obtaining the vehicle speed and position change rate, the weighted distance is calculated according to the initial weight value and the weight distribution rule. The formula is used , is the initial weight value, is the vehicle speed or position change rate, is the overall mean of the sample, and the squared difference is accumulated point by point to form a complete sample distance set, thus obtaining a weighted sample point distance set.

[0027] S212: separating the weighted distance of sample points within a class and the weighted distance of sample points between classes from the weighted sample point distance set, and calculating the weighted mean values ​​of sample points within a class and between classes respectively according to the classes to which the sample points belong by calling the weighted distance information of the sample points within the set, and gradually aggregating the sample class information to obtain the weighted average distance of samples within a class and the weighted average distance of samples between classes, and generating a sample weighted average distance parameter set; By analyzing the category labels of the sample points, the intra-class weighted distance is calculated according to the weighted distance between sample points in the same category. The category label field of the sample points is called to extract the sample point data of each category in a grouped form. The average of the weighted distances between each pair of samples after grouping is used as the intra-class weighted average distance. The formula is expressed as ,in is the number of class sample points, For sample points and The weighted distance of Indicates category The inter-class weighted distance is calculated by the weighted distance between sample points of different categories. Perform cross-category averaging, and obtain the weighted average distance of samples within a class and the weighted average distance of samples between classes through the above calculations, and generate a sample weighted average distance parameter set.

[0028] S213: Based on the sample weighted average distance parameter set, the difference between the weighted average distances of samples between classes and within classes is calculated point by point, using the formula: ; Calculate and generate dynamic feature weight values; in, represents the dynamic feature weight value, Indicates The weighted average distance within the class of samples, Indicates The weighted average distance between classes of samples, Indicates The initial feature weight value of the sample point, Dynamic adjustment coefficients are used for smooth weight calculations. represents the total number of samples, Used to adjust the influence of sample point weight on difference normalization. Represents the normalized sum of weight differences.

[0029] formula: ; The formula is useful because by introducing the square root and the dynamic adjustment factor 1. Combined with sample weights The influence of is used to smoothly adjust the weighted differences between classes and within classes, and the sensitivity and calculation accuracy of the results are improved through point-by-point normalization calculation.

[0030] Detailed explanation of the formula and the process of formula calculation and derivation: First, extract the weighted average distance of samples within the class from the sample weighted average distance parameter set And the weighted average distance between samples of different classes , these parameters are obtained through the above calculations, assuming that the total number of initial samples is , where the initial weight value is They are distributed in the range of 0.1 to 1.0, and the formula , assuming the dynamic adjustment coefficient , the following are the parameter values: ; ; ; Substitute the formula point by point to calculate: For the first sample point: ; Repeat the above process to calculate the contribution value of each sample point and add them up item by item: at last: ; The results show that the dynamic feature weights of the sample points are adjusted to the normalized weighted results after overall smoothing, and the dynamic feature weight values ​​are obtained by calculation. , providing a standardized weight value benchmark for subsequent calculations and further analysis.

[0031] See also Figure 4 , the specific steps for obtaining the operation mode clustering results are: S311: calling the dynamic feature weight value, calculating the weighted density distribution value of the forklift operation sample point, performing weighted density calculation on each sample point through the vehicle speed and fork load feature value, accumulating the sample point density value point by point, and performing density classification processing in combination with the feature distribution parameters to generate a weighted density distribution value set; Extract the speed and fork load parameter characteristics from the forklift operation data, calculate the weighted density distribution value point by point, and normalize the speed parameter value and fork load parameter value according to the specific weight distribution rule by calling the feature weight value. After normalization, substitute the feature value of each sample point into the weighted density calculation formula ,in For sample points The density value of For sample points No. eigenvalues, The weight value of the corresponding feature is obtained by accumulating the weighted contribution values ​​of all features of the sample point to obtain the weighted density value set of each sample point. The accumulation of density distribution values ​​needs to be combined with the overall density range of the sample point. The density values ​​are distributed in a uniform interval by normalizing the density values, and then classified according to the density range of the feature distribution. The median, maximum value and minimum value of the density value distribution interval are calculated and the segmentation threshold is set to complete the classification of the density distribution set and generate a weighted density distribution value set.

[0032] S312: Combine the weighted density distribution value set, compare the distribution values ​​of the vehicle speed and fork load characteristic values ​​with the overall density distribution point by point, and use the formula: ; Determine the local density peak of each sample point and generate a density peak set; in, represents the density peak, Indicates The weighted density value of sample points, represents the dynamic weight of vehicle speed characteristics, The dynamic weight representing the fork load characteristics, To adjust the parameters used to adjust the smoothness of the weight to density calculation, Indicates the dynamic smoothing effect of the fork load weight on the density value, The operation selects the local density maximum as the density peak; formula: ; The formula is useful in that by introducing the adjustment parameter in the square root denominator , smoothed the influence of the fork load feature weight, making the calculation of the density peak more accurate, and at the same time, the local density was screened through the maximum value operation to improve the significance of the result; Detailed explanation of the formula and the process of formula calculation and derivation: Suppose in a set of sample data, The calculated value is , assuming the corresponding speed weight They are , fork load weight They are , adjustment coefficient , enter the formula to calculate: ; ; ; ; ; Maximum for ; This result shows that after the density peak is calculated, the sample point The density peak of is the highest, indicating that its feature weight contributes greatly to the weighted density of vehicle speed and fork load, and can be used as an important node in merge clustering.

[0033] S313: performing merging clustering on the density peak set, by calculating the weighted density difference between the density peak of each sample point and other sample points, determining the weighted distance distribution between the sample points, and reclassifying the clustering results by gradually aggregating the sample point attribution information to generate cluster sample divisions; Extract the weighted density peak of the sample point, and compare the weighted density value difference and spatial distribution difference between the sample points to calculate the formula Get the weighted density difference between sample points, determine whether the difference is less than the threshold, set the threshold based on 20% of the maximum density difference. For example, if the maximum density difference is 8, the threshold is 1.6. Aggregate the sample points whose difference is less than the threshold, and further calculate the weighted average distance from the cluster center to the sample point to update the density value of the cluster center. Repeat the above steps again to adjust the sample point classification and form a clustering result to generate cluster sample division.

[0034] S314: Call cluster sample division, calculate the weighted density value of each cluster center, perform weighted summation of the vehicle speed and fork load characteristic values ​​of the sample points belonging to the cluster center in sequence, and dynamically adjust the center characteristics in combination with the mean distribution of the sample density values ​​within the class to generate the operation mode clustering result.

[0035] A set of sample points for each cluster center is selected, and the vehicle speed characteristics and fork load characteristics of the sample points belonging to each cluster center are extracted. The weighted density value is calculated by taking a weighted average of the vehicle speed characteristic values ​​and combining the density value of the fork load characteristic with the weight parameter. The density value is then combined with the inverse of the square of the distance to the cluster center to calculate the correction value of the density center. The weighted density correction is performed on each cluster center and the operation mode clustering result is generated.

[0036] See also Figure 5 , the specific steps for obtaining the classification results of the staged operation mode are: S411: calling the operation mode clustering result, dividing the sample groups item by item according to the acceleration, turning and stable operation stages in the forklift operation data, extracting the sample point position parameters and attitude angle feature values ​​in the operation stage, performing operation feature grouping processing on the sample groups in turn, and generating stage-by-stage sample group division; The characteristic parameters of the acceleration stage are analyzed step by step, the speed change sequence recorded in the forklift operation data is called, and the position where the change value in the speed data is greater than a set threshold is extracted as the judgment standard of the acceleration stage. The sample points above the speed change threshold are classified into the acceleration stage, and the vehicle body posture angle recorded in the corresponding sample points are extracted. When extracting the turning stage, the direction change parameters and fork load change values ​​recorded in the forklift operation data are called, and the time period when the direction change angle is greater than a certain angle threshold is analyzed in combination with the time series of each sample point. The corresponding sample points are classified into the turning stage, and the characteristic values ​​of the vehicle body posture angle and position change rate in this stage are extracted at the same time. In the stable operation stage, the sample points of the acceleration and turning stages are eliminated, and the speed fluctuation rate and fork load fluctuation rate of the remaining sample points are calculated. The low-fluctuation sample points are defined as the stable stage sample group according to the volatility mean and standard deviation. The data of the above-mentioned division stages are gradually aggregated and the sample group of the operation stage is established to generate a staged sample group division.

[0037] S412: extracting the characteristic values ​​of the vehicle body posture angle and position change rate in each stage of the phased sample group division, weighting the characteristic values ​​of the sample points point by point by combining the dynamic characteristic weight values, and performing weighted accumulation processing on the characteristic data of the running stage to generate a phased weighted characteristic value set; For the weighted processing of the acceleration stage, the absolute difference between the vehicle body attitude angle and the position change rate of the sample points is calculated and extracted, and the dynamic feature weight value is allocated to the two parameters in turn. The weight ratio of the two features is adjusted according to the distribution mean of the two features, and the weighted results of the two are accumulated to obtain the weighted feature value of the acceleration stage. In the turning stage, the direction change data and fork load feature data of the sample points are called, and the weighted ratio of the dynamic feature weight value to the direction change value is calculated as the weighted feature component of the turning stage. The feature component and the fork load feature are accumulated in turn. The feature weighting of the stable operation stage adopts the weighted product of the absolute fluctuation value of the vehicle speed and position change rate of the sample points and the dynamic feature weight value, and the cumulative result is calculated in combination with the operation stage time series. The distribution of the feature values ​​of the three stages is gradually accumulated and classified to generate a stage-by-stage weighted feature value set.

[0038] S413: Calculate the volatility of the periodic weighted eigenvalue set and readjust the weights using the formula: ; Calculate and generate the re-adjusted weight parameters; in, represents the re-adjusted weight parameter, represents the volatility of the weighted eigenvalue set, represents the mean of the weighted eigenvalue set, Represents the minimum fluctuation threshold, which is used to adjust the flexibility and sensitivity of the weights and increase the formula's ability to respond to small fluctuations; formula: ; The benefit of the formula is that by introducing the minimum fluctuation threshold , increasing the sensitivity of the calculation, especially when the change of the eigenvalue is small, it can respond to small changes and improve the model's ability to identify outliers. This setting makes the weight adjustment more accurate, suitable for unstable data sources, and improves the practical application value of the model. Detailed explanation of the formula and the derivation process of the formula calculation: Assume that the volatility obtained from the data is 0.03, the mean is 0.01, the minimum fluctuation threshold Set to 0.05 and adjust these parameters according to actual conditions. These values ​​can be obtained by monitoring actual data. For example, volatility It can be calculated by the standard deviation within a certain period of time, the mean It is the average value of the data in the same time period.

[0039] ; ; ; ; ; This result shows that the adjusted weight parameter It is approximately 1.76, which means that according to the current volatility and mean, by adjusting the minimum volatility threshold, we can get a relatively high weight parameter, which can respond more sensitively to small fluctuations when processing actual data, increasing the model's ability to adapt to small changes in data. It can be further deduced that this weight can be directly used in subsequent calculations or model adjustments to optimize model performance.

[0040] S414: Call the readjusted weight parameters, re-evaluate and classify the weighted feature values ​​of each stage in combination with the phased sample group division information, calculate the comprehensive feature parameters of the operation stage through classification, classify the readjusted classification features of the operation stage in turn, and generate the phased operation mode classification results.

[0041] The characteristic distribution and corresponding dynamic weights of sample points in the acceleration, turning and stable operation stages are extracted one by one, and the weighted mean is calculated for the weighted characteristic distribution of the acceleration stage. The classification of each characteristic point is determined according to the mean interval, and the direction change and fork load fluctuation characteristic values ​​of the turning stage are extracted. The samples are classified and aggregated in combination with the weighted distribution. For the stable operation stage, the weighted normalized distribution of the vehicle speed fluctuation value of the sample point is calculated. The sample points of the three stages are gradually reclassified and aggregated to generate the classification results of the staged operation mode.

[0042] See also Figure 6 ,The specific steps for obtaining the forklift usage status classification indicator set are: S511: calling the classification result of the staged operation mode, extracting the fluctuation value range of the fork load and the body posture angle in the differentiated operation stage, and gradually classifying the fluctuation value range of the sample points by sequentially screening the dynamic change values ​​of the fork load and the body posture angle parameters for the sample points in the operation stage, and generating a sample fluctuation value set for the operation stage; Firstly, the fork load value and attitude angle parameters recorded at each sample point in the operation stage are analyzed point by point, and the fluctuation range is calculated by the time series fluctuation change of each group of sample points. The sliding window method is used to extract the local maximum and minimum values ​​according to the time series characteristics, and the difference is calculated to obtain the fluctuation range value. After the extraction of the fluctuation range value, the samples are reclassified in combination with the differentiated operation stages of the sample group classification, and the extracted fluctuation range is stored corresponding to the classified operation stage sample points to ensure that each group of fluctuation value range corresponds to the operation stage characteristics. After all the operation stage sample groups are classified and stored, the fluctuation ranges of all sample groups are gradually accumulated, and the upper and lower quartiles of the classified fluctuation range values ​​are counted and the overall fluctuation range is calculated. The fluctuation value difference characteristics of each operation stage group are re-evaluated through the calculated fluctuation range to generate an operation stage sample fluctuation value set.

[0043] S512: extracting the fluctuation value distribution of each type of sample in the sample fluctuation value set of the operation stage, calculating the fluctuation value interval characteristics of the samples in turn and combining the distribution range of each type of samples for classification statistics, and gradually classifying and summarizing the characteristic parameters of the fluctuation range of the sample points to generate the classification statistics results of the sample fluctuation value distribution; First, the distribution statistics of the fluctuation value set in each type of operation stage samples are carried out, and the overall distribution of each type of samples is summarized by calculating the basic statistics such as the mean, median, standard deviation in the sample distribution; then, the classification statistics of the distribution are carried out in combination with the upper and lower quartiles of the fluctuation value, and the classification intervals are gradually refined based on the skewed characteristics of the distribution. For example, for samples with fluctuation values ​​higher than the mean of the overall distribution by two standard deviations, they are defined as the interval with higher fluctuation values; for samples with one standard deviation lower than the mean of the overall distribution, they are defined as the interval with lower fluctuation values, and the remaining samples are divided into the interval with medium fluctuation values; after completing the classification of the distribution, the proportion of the number of samples in each interval of the sample fluctuation distribution of each type of operation stage is counted according to the divided intervals, and their contribution to the overall distribution is analyzed category by category; finally, the classified statistical results are recorded in combination with the upper and lower boundaries of the fluctuation interval, and the statistical characteristics of each category are gradually accumulated to generate the classification statistical results of the sample fluctuation value distribution.

[0044] S513: Call the sample fluctuation value distribution classification statistics result, calculate the fluctuation mean and fluctuation range of each type of sample distribution, and perform weighted operation on the fluctuation value characteristics point by point and use the formula: ; Calculate and generate sample distribution mean and fluctuation range; in, represents the weighted fluctuation value of the sample distribution, Indicates The fluctuation value of the samples, Represents the mean of the fluctuation values ​​of all sample points, Indicates The dynamic weight of the sample points, The nonlinear effect of adjusting the weights for the smoothing coefficients, represents the smoothing effect of dynamically adjusting weights on distribution characteristics, Indicates the total number of samples; formula: ; The benefit of the formula is that by introducing dynamic weights and nonlinear smoothing adjustment coefficient 2. It can effectively balance the impact of sample weights on the distribution of volatility values, further enhance the correlation between sample weights and volatility characteristics, and thus improve the level of detail in the distribution of volatility values.

[0045] Detailed explanation of the formula and the process of formula calculation and derivation: First, define the parameters: represents the weighted fluctuation value of the sample distribution, Indicates The fluctuation value of the samples, Represents the mean of the fluctuation values ​​of all sample points, Indicates The dynamic weight of the sample points, 2 is the nonlinear effect of the smoothing coefficient adjustment weight, Represents the smoothing effect of dynamically adjusting weights on distribution characteristics.

[0046] Parameter acquisition method description: :It is obtained by monitoring the fluctuation value range of the sample points, and a sliding window is used to calculate the difference between the local maximum and minimum values ​​of each group of sample points, and the difference is taken as the fluctuation value.

[0047] :The mean is calculated by the fluctuation value of the sample point. The calculation formula is: ,in is the total number of samples.

[0048] : The dynamic weight is extracted through the operation stage characteristics of the sample points, and the weight value is dynamically adjusted according to the importance of the operation stage. The comprehensive influence of the operation stage on the fork load and the vehicle body posture angle can be proportionally calculated and determined.

[0049] 2: Obtain the smoothing coefficient through experimental calibration, and dynamically set the adjustment value according to the influence range of the sample fluctuation value on the overall feature distribution.

[0050] Parameter assignment and formula derivation: Assume there are 5 sample points and the fluctuation value , , , , , weight , , , , , smoothing coefficient .

[0051] 1. Calculate the mean .

[0052] 2. Calculate the weighted volatility value point by point: right : .

[0053] right : .

[0054] right : .

[0055] right : .

[0056] right : .

[0057] 3. Calculate the denominator: .

[0058] .

[0059] .

[0060] .

[0061] .

[0062] 4. Enter the formula for calculation: .

[0063] The result shows that the weighted fluctuation value of the sample distribution is 0.383, which is related to the calculated mean, weight and fluctuation range, indicating that the fluctuation characteristics of the operation stage have completed the weighted distribution processing.

[0064] S514: Call the sample distribution mean and fluctuation range, and gradually classify and summarize the fluctuation value range and mean characteristic parameters of each type of sample in combination with the classification statistics results in the operation stage, recalculate the classification index for the dynamic change parameter characteristics of each type of operation stage sample, and generate a forklift usage status classification index set.

[0065] Firstly, the fluctuation mean and interval characteristics are respectively corresponded to each type of operation stage samples, and the dynamic change characteristics of each type of operation stage are divided by the distribution mean characteristics and the upper and lower interval boundaries; after gradually classifying the fluctuation range of each type of sample, the distribution characteristics of the operation characteristic parameters of each type of sample are calculated in combination with the change values ​​of the fork load and the vehicle body posture angle; after completing the classification of all characteristic parameters, the dynamic change characteristics of all operation stage samples are gradually summarized to generate a forklift usage status classification indicator set.

[0066] A forklift use status intelligent analysis system, the forklift use status intelligent analysis system is used to execute the forklift use status intelligent analysis method, the system comprises: The feature extraction module extracts feature values ​​from vehicle speed, fork load, and vehicle body posture angle based on forklift operation data, calculates the fluctuation mean and variance, normalizes the feature values, and obtains the initial feature weight value; The dynamic weight adjustment module calculates the weighted distance of sample points based on the initial feature weight value, obtains the weighted average distance of samples inside and outside the class, performs normalization adjustment through the difference, updates the weight of each feature, and generates a dynamic feature weight value; The cluster analysis module calculates the weighted density distribution of the forklift operation sample points based on the dynamic feature weight value, determines the density peak, performs the merge clustering of the density peak, calculates the weighted density of the cluster center, and generates the operation mode clustering result; The operation mode classification module divides the forklift operation data into acceleration, turning, and stable operation stages based on the operation mode clustering results, extracts the weighted features of the vehicle body posture angle and position change rate, calculates the fluctuation rate and adjusts the weight, and generates the staged operation mode classification results; The status classification index generation module extracts the fluctuation range of the fork load and the vehicle body posture angle in the differentiated operation stage based on the classification results of the staged operation mode, performs statistics on the distribution of fluctuation values, calculates the distribution mean and interval, and generates a forklift usage status classification index set.

[0067] Embodiment 2 This embodiment is basically the same as the first embodiment, except that, compared with the first embodiment, in this embodiment, step S1 further includes: Based on the operating data of the forklift controller and ECU computer, extract the version information including vehicle model, identification number and year; Specifically, based on the operating data of the forklift controller and ECU computer, the vehicle model, identification number and year version information are extracted. It is necessary to obtain the data stored in the ECU from the forklift control system. First, connect to the forklift control system through the communication interface of the ECU, and use the CAN bus or serial communication protocols such as UART, SPI, etc. to read the firmware version, hardware serial number and software version information in the storage unit. Then, extract the vehicle identification number. The identification number is usually stored in the EEPROM or Flash storage unit. It is necessary to obtain the information by reading the data at a specific address and verify it with the system registry content provided by the forklift controller to obtain the complete vehicle identification number. For the vehicle model information, it can be parsed The device identification field in the ECU firmware is used for confirmation. This field may contain information such as manufacturer code, device type, load capacity, etc. For example, if the device identification field returned by the ECU is "FLK-3500", it means that the forklift model is FLK-3500 and its rated load capacity is 3500kg. In order to obtain the production year of the forklift, it is necessary to parse the ECU system log or production data record, which contains the device factory time. If the log record shows that the device first ran on "2021-06-15", the vehicle may have been produced in 2021. Its accuracy can be further verified by combining other hardware information. Finally, after integrating these data, a basic information table of the vehicle is formed, including model, identification number and production year.

[0068] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for intelligent analysis of forklift usage status, characterized in that: The following steps are involved: S1: Based on the forklift operation data, feature parameters including vehicle speed, fork load, and vehicle body posture angle are extracted, the fluctuation range of the feature parameters is divided into intervals, the fluctuation mean and variance of multiple intervals are calculated, all feature parameters are normalized, and the initial feature weight value is generated; S2: Based on the initial feature weight value, calculate the weighted distance of each forklift operation sample point on the vehicle speed sensor and the position change rate, calculate the weighted average distance of samples within the class and the weighted average distance of samples between classes, normalize the difference between the two and adjust the feature weight to generate a dynamic feature weight value; S3: Based on the dynamic feature weight value, the weighted density distribution value of the forklift operation sample point is calculated, the density peak value is determined by combining the vehicle speed and the fork load feature value, the density peak value is merged and clustered, the cluster samples are re-divided, the weighted density value of the cluster center is calculated, and the operation mode clustering result is generated; S4: Based on the operation mode clustering result, the sample groups are divided according to the operation stages of acceleration, turning, and stable operation in the forklift operation data, and the weighted feature values ​​of the vehicle body posture angle and position change rate in each stage are extracted. The fluctuation rate of the weighted feature values ​​is calculated and the weights are readjusted to generate the staged operation mode classification results; S5: Based on the classification results of the staged operation modes, the fluctuation value ranges of the fork load and the vehicle body posture angle in the differentiated operation stages are extracted, the fluctuation value distribution of each type of sample is classified and counted, the distribution mean and the fluctuation range are calculated, and a forklift usage status classification indicator set is generated.

2. The intelligent analysis method for forklift usage status according to claim 1 is characterized in that: The initial feature weight value specifically includes the feature parameter fluctuation range, multi-interval fluctuation mean, variance, and normalized feature parameters. The dynamic feature weight value includes weighted distance, average distance between samples within and outside the class, and difference normalization. The operation mode clustering result specifically refers to density peak, merged clustering, and weighted density of cluster center. The staged operation mode classification result includes stage division, volatility calculation, and weight adjustment. The forklift usage status classification indicator set specifically includes the fork load fluctuation range, body posture angle fluctuation range, fluctuation value distribution statistics, distribution mean, and fluctuation interval.

3. The intelligent analysis method for forklift usage status according to claim 2 is characterized in that: The steps for obtaining the initial feature weight value are specifically as follows: S111: extracting characteristic parameters of vehicle speed, fork load and vehicle body posture angle from the forklift operation data, analyzing the fluctuation range of the characteristic parameters item by item according to the discrete interval rule, dividing the data set in each interval into separate areas, and generating fluctuation mean and fluctuation variance by performing item by item mean operation and variance calculation on sample points in the area; S112: Normalize the fluctuation mean and fluctuation variance using the formula: ； Calculate and generate a normalized feature parameter matrix; in, Represents the normalized feature matrix Line The eigenvalues ​​of the columns, Represents the original feature matrix Line The eigenvalues ​​of the columns, Representative The mean of the column features, Representative The variance of the column features, represents the smoothing coefficient used to avoid the denominator being zero; S113: Calculate the inter-column mean of the normalized feature parameter matrix, analyze its initial weight value column by column, use the matrix column mean vector calculation as the initial weight vector, and generate the initial feature weight value based on the column mean and the deviation superposition processing between the matrix column multiple eigenvalues.

4. The method for intelligent analysis of forklift usage status according to claim 3, characterized in that: The steps for obtaining the dynamic feature weight value are specifically as follows: S211: performing parameter weighting processing on the running sample points of the vehicle speed sensor and the position change rate according to the initial feature weight value, calculating the weight distribution of the vehicle speed and position change rate parameters on the sample points by calling the initial weight value, and calculating the accumulated square difference value of each sample point on the weighted distance, to obtain a weighted sample point distance set; S212: separating the weighted distance of sample points within a class and the weighted distance of sample points between classes from the weighted sample point distance set, and calculating the weighted mean values ​​of sample points within a class and between classes respectively according to the classes to which the sample points belong by calling the weighted distance information of the sample points within the set, and gradually aggregating the sample class information to obtain the weighted average distance of samples within a class and the weighted average distance of samples between classes, and generating a sample weighted average distance parameter set; S213: Based on the sample weighted average distance parameter set, the difference between the weighted average distances of samples between classes and within classes is calculated point by point, using the formula: ； Calculate and generate dynamic feature weight values; in, represents the dynamic feature weight value, Indicates The weighted average distance within the class of samples, Indicates The weighted average distance between classes of samples, Indicates The initial feature weight value of the sample point, Dynamic adjustment coefficients are used for smooth weight calculations. represents the total number of samples, Used to adjust the influence of sample point weight on difference normalization. Represents the normalized sum of weight differences.

5. The method for intelligent analysis of forklift usage status according to claim 4, characterized in that: The steps for obtaining the operation mode clustering result are specifically as follows: S311: calling the dynamic feature weight value, calculating the weighted density distribution value of the forklift operation sample point, performing weighted density calculation on each sample point through the vehicle speed and fork load feature value, accumulating the sample point density value point by point, and performing density classification processing in combination with the feature distribution parameter to generate a weighted density distribution value set; S312: Combine the weighted density distribution value set, compare the distribution values ​​of the vehicle speed and fork load characteristic values ​​with the overall density distribution point by point, and use the formula: ； Determine the local density peak of each sample point and generate a density peak set; in, represents the density peak, Indicates The weighted density value of sample points, represents the dynamic weight of vehicle speed characteristics, The dynamic weight representing the fork load characteristics, To adjust the parameters used to adjust the smoothness of the weight to density calculation, Indicates the dynamic smoothing effect of the fork load weight on the density value, The operation selects the local density maximum as the density peak; S313: performing merging clustering on the density peak set, determining the weighted distance distribution between the sample points by calculating the weighted density difference between the density peak of each sample point and other sample points, and reclassifying the clustering results by gradually aggregating the sample point attribution information to generate cluster sample division; S314: Call the cluster sample division, calculate the weighted density value of each cluster center, perform weighted summation of the vehicle speed and fork load characteristic values ​​of the sample points belonging to the cluster center in sequence, and dynamically adjust the center characteristics in combination with the mean distribution of the sample density values ​​within the class to generate the operation mode clustering result.

6. The method for intelligent analysis of forklift usage status according to claim 5, characterized in that: The steps for obtaining the classification results of the staged operation modes are specifically as follows: S411: calling the operation mode clustering result, dividing the sample groups item by item according to the acceleration, turning and stable operation stages in the forklift operation data, extracting the sample point position parameters and attitude angle feature values ​​of the operation stage, performing operation feature grouping processing on the sample groups in turn, and generating stage-by-stage sample group division; S412: extracting the characteristic values ​​of the vehicle body posture angle and position change rate in each stage of the phased sample group division, weighting the characteristic values ​​of the sample points point by point by combining the dynamic characteristic weight values, and performing weighted accumulation processing on the characteristic data of the running stage to generate a phased weighted characteristic value set; S413: Calculate the volatility of the phased weighted eigenvalue set, and readjust the weights using the formula: ； Calculate and generate the re-adjusted weight parameters; in, represents the re-adjusted weight parameter, represents the volatility of the weighted eigenvalue set, represents the mean of the weighted eigenvalue set, Represents the minimum fluctuation threshold, which is used to adjust the flexibility and sensitivity of the weights and increase the formula's ability to respond to small fluctuations; S414: Call the readjusted weight parameters, re-evaluate and classify the weighted feature values ​​of each stage in combination with the phased sample group division information, calculate the comprehensive feature parameters of the operation stage by classification, classify the readjusted classification features of the operation stage in turn, and generate the phased operation mode classification results.

7. The method for intelligent analysis of forklift usage status according to claim 6, characterized in that: The steps for obtaining the forklift usage status classification index set are specifically as follows: S511: calling the classification result of the staged operation mode, extracting the fluctuation value range of the fork load and the body posture angle in the differentiated operation stage, and gradually summarizing and classifying the fluctuation value range of the sample points by sequentially screening the dynamic change values ​​of the fork load and the body posture angle parameters for the sample points in the operation stage, and generating a sample fluctuation value set for the operation stage; S512: extracting the fluctuation value distribution of each type of samples in the sample fluctuation value set of the operation stage, calculating the fluctuation value interval characteristics of the samples in turn and combining the distribution range of each type of samples for classification statistics, and gradually classifying and summarizing the characteristic parameters of the fluctuation range of the sample points to generate the classification statistics results of the sample fluctuation value distribution; S513: Call the sample fluctuation value distribution classification statistics result, calculate the fluctuation mean and fluctuation range of each type of sample distribution, and perform weighted operation on the fluctuation value characteristics point by point and use the formula: ； Calculate and generate sample distribution mean and fluctuation range; in, represents the weighted fluctuation value of the sample distribution, Indicates The fluctuation value of the samples, Represents the mean of the fluctuation values ​​of all sample points, Indicates The dynamic weight of the sample points, The nonlinear effect of adjusting the weights for the smoothing coefficients, represents the smoothing effect of dynamically adjusting weights on distribution characteristics, Indicates the total number of samples; S514: calling the sample distribution mean and fluctuation range, and gradually classifying and summarizing the fluctuation value range and mean characteristic parameters of each type of sample in combination with the classification statistics results in the operation stage, recalculating the classification index for the dynamic change parameter characteristics of each type of operation stage sample, and generating a forklift usage status classification index set.

8. An intelligent analysis system for forklift usage status, characterized in that: According to the method for intelligent analysis of forklift usage status according to any one of claims 1 to 7, the system comprises: The feature extraction module extracts feature values ​​from vehicle speed, fork load, and vehicle body posture angle based on forklift operation data, calculates the fluctuation mean and variance, normalizes the feature values, and obtains the initial feature weight value; The dynamic weight adjustment module calculates the weighted distance of the sample points based on the initial feature weight value, obtains the weighted average distance of samples inside and outside the class, performs normalization adjustment through the difference, updates the weight of each feature, and generates a dynamic feature weight value; The cluster analysis module calculates the weighted density distribution of the forklift operation sample points based on the dynamic feature weight values, determines the density peaks, performs merging clustering of the density peaks, calculates the weighted density of the cluster centers, and generates the operation mode clustering results; The operation mode classification module divides the forklift operation data into acceleration, turning, and stable operation stages based on the operation mode clustering results, extracts weighted features of the vehicle body posture angle and position change rate, calculates the fluctuation rate and adjusts the weight, and generates a staged operation mode classification result; The state classification index generation module extracts the fluctuation range of the fork load and the body posture angle in the differentiated operation stage based on the staged operation mode classification result, performs statistics on the fluctuation value distribution, calculates the distribution mean and interval, and generates a forklift use state classification index set.

Images