Open-pit mine transport equipment multi-source information fusion monitoring method based on automatic driving

Abstract

The application discloses an open-pit mine transport equipment multi-source information fusion monitoring method based on automatic driving, which comprises the following steps: step one, collecting the position signal, the speed signal and the steering wheel rotation angle of the transport equipment according to the sampling period; step two, performing data preprocessing on the to-be-fused data of the multi-source information; step three, constructing a support vector machine regression model; step four, establishing a particle swarm optimization algorithm model, solving the optimal solution of the penalty coefficient and the kernel function parameter of the support vector machine regression model, and obtaining the optimal parameter combination; and step five, bringing the optimal parameter combination into the support vector machine regression model, so that the monitoring result is more accurate, and the transport equipment progress route monitoring result is obtained. The application has the characteristics of greatly shortening the operation time, improving the monitoring precision and improving the accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of transportation equipment data fusion, more particularly, the present application relates to a kind of open-pit mine transportation equipment multi-source information fusion monitoring method based on automatic driving. BACKGROUND

[0002] Iron ore is one of the most widely used metals in daily life. Iron ore has many uses, from making cars to making a knife, which is an important raw material. Due to its importance, how to collect iron ore resources is also attracting people's attention. Iron ore mining is to carefully study, summarize and identify to obtain iron ore minerals.

[0003] The diversification of iron ore production environment conditions makes the iron ore production and harvesting face great challenges, therefore, with the development of science and technology, higher requirements are put forward for the production mode of iron ore, and intelligent mechanized iron ore has gradually replaced the traditional mode.

[0004] Transportation equipment will inevitably be affected by external environment, road conditions, changes in load, vibration of itself, reception of signals and a series of influences on the path of automatic driving. For automated transportation equipment, necessary influencing factors are generally monitored by multiple sensors to effectively obtain data information at different time nodes and spatial dimensions. In transmission and decision-making, multi-source information fusion problems will inevitably arise.

[0005] In the existing multi-source information fusion research, some use hierarchical structure and genetic neural network algorithm for processing, and others use gray least square support vector machine algorithm for processing. However, both of these two fusion algorithms cannot achieve the expected accuracy, resulting in insufficient monitoring accuracy, and due to the large amount of data, the operation time is too long and the real-time performance is not good enough. SUMMARY

[0006] The purpose of the present application is to design and develop a kind of open-pit mine transportation equipment multi-source information fusion monitoring method based on automatic driving. By combining the self-adaptive particle swarm optimization algorithm with the support vector machine regression model, the progress route of the transportation equipment can be accurately monitored, thereby avoiding the deviation of the transportation equipment route and improving the monitoring accuracy.

[0007] The technical scheme provided by the present application is as follows:

[0008] A kind of open-pit mine transportation equipment multi-source information fusion monitoring method based on automatic driving, comprising the following steps:

[0009] Step one, according to the sampling period, collect the position signal, vehicle speed signal and steering wheel rotation angle of the transportation equipment;

[0010] Step two, data preprocessing is performed on the to-be-fused data of multi-source information;

[0011] Step three, constructing a support vector machine regression model;

[0012] Step four, establishing a particle swarm optimization algorithm model to solve the optimal solution of the penalty coefficient and kernel function parameter of the support vector machine regression model, and obtaining an optimal parameter combination;

[0013] Step five, bringing the optimal parameter combination into the support vector machine regression model to obtain a travel route prediction result of the transportation equipment.

[0014] Preferably, the to-be-fused data of the multi-source information comprises:

[0015] initial position signal, previous time prediction position signal, previous time real-time position signal, average vehicle speed signal, previous time prediction vehicle speed signal, previous time real-time vehicle speed signal, real-time maximum steering wheel rotation angle, previous time prediction steering wheel rotation angle and previous time real-time steering wheel rotation angle.

[0016] Preferably, the position signal, the vehicle speed signal and the rotation angle of the steering wheel of the transportation equipment are monitored by a sensor.

[0017] Preferably, the data preprocessing comprises outlier rejection and default value processing.

[0018] Preferably, the outlier rejection comprises the following steps:

[0019] Step 1, arranging the data collected by the sensor in ascending order;

[0020] Step 2, obtaining high-end statistics and low-end statistics according to different sample sizes;

[0021] Step 3, when the confidence value is different, the number of sample sizes corresponds to different critical values :

[0022] When , , is an outlier and is rejected ; when , , is an outlier and is rejected , otherwise there is no outlier;

[0023] In the formula, is the high-end statistics, is the low-end statistics, and .

[0024] Preferably, the default value processing comprises:

[0025] coordinates between known coordinates and coordinates inserted satisfy:

[0026] .

[0027] Preferably, the data preprocessing further comprises normalizing the to-be-fused data of multi-source information:

[0028] ;

[0029] wherein, is the output data after normalization, is the original data, is the minimum value of the variable , and is the maximum value of the variable .

[0030] Preferably, the support vector machine regression model satisfies:

[0031] ;

[0032] ;

[0033] ;

[0034] wherein, is a kernel function, and corresponds to and inner products of the mapping to the feature space, used to replace the complex operation of the feature space, is a kernel function parameter, and σ is a standard deviation of the kernel function, also known as a scale parameter, and b is a bias term, also known as an intercept term.

[0035] Preferably, the support particle swarm optimization algorithm model specifically comprises:

[0036] The updating of the speed and position of the particle swarm optimization algorithm model satisfies:

[0037] ;

[0038] ;

[0039] wherein, is the speed of the i th particle in an n-dimensional space when the iteration number is , is an inertia weight, is the iteration number, and is the position of the i th particle when the iteration number is . the first particle in the d-dimensional space, the first particle in the d-dimensional space, the velocity of the first particle in the d-dimensional space, a first acceleration factor, a random number distributed in the interval [0, 1], a random number distributed in the interval [0, 1], the iteration number is n, the first particle in the d-dimensional space, the first particle in the d-dimensional space, the first particle in the d-dimensional space, the iteration number is n, the first particle in the d-dimensional space, the first particle in the d-dimensional space, the first particle in the d-dimensional space, a second acceleration factor, a random number distributed in the interval [0, 1], a random number distributed in the interval [0, 1], the iteration number is n, the first particle in the d-dimensional space, the first particle in the d-dimensional space, the iteration number is n, the first particle in the d-dimensional space, the first particle in the d-dimensional space, the first particle in the d-dimensional space;

[0040] wherein the inertia weight satisfies:

[0041] ;

[0042] wherein, a minimum weight value, a maximum weight value, a fitness value of the particle at any time, a minimum fitness value in the particle swarm, an average fitness value of the particle swarm;

[0043] the first acceleration factor and the second acceleration factor satisfy:

[0044] ;

[0045] wherein, , are initial values of the and, respectively, , are termination values of the and, respectively. Preferably, the parameters of the particle swarm optimization algorithm model are initialized as:

[0046] Preferably, the parameters of the particle swarm optimization algorithm model are initialized as:

[0047] ​​​The population size is 50, the iteration number is 1000, the initial value of the inertia weight is 0.6, and the initial values of the first acceleration factor and the second acceleration factor are both 2.

[0048] The beneficial effects of the present application are:

[0049] The monitoring method for the route of the open-pit mine transportation equipment based on automatic driving developed by the present application is capable of simplifying the operation process, greatly shortening the operation time, and improving the monitoring accuracy and the accuracy, and is capable of determining the travel route of the transportation equipment in advance, judging whether the travel route of the transportation equipment meets the expectation, achieving the purpose of revising the route in advance, and optimizing the automatic driving of the transportation equipment. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The present application provides a monitoring method for the route of the open-pit mine transportation equipment based on automatic driving. DETAILED DESCRIPTION

[0051] The present application will be further described in detail below, so that those skilled in the art can implement the present application according to the description.

[0052] As shown in the accompanying drawings, Figure 1 The monitoring method for the route of the open-pit mine transportation equipment based on automatic driving provided by the present application comprises the following steps:

[0053] Step one, according to the sampling period, the position signal, the speed signal and the rotation angle of the steering wheel of the transportation equipment are collected, and the position signal, the speed signal and the rotation angle of the steering wheel are the to-be-fused data of the multi-source information;

[0054] The to-be-fused data of the multi-source signal specifically includes: initial position signal, previous time prediction position signal, previous time real-time position signal, average speed signal, previous time prediction speed signal, previous time real-time speed signal, real-time maximum rotation angle of the steering wheel, previous time prediction steering wheel rotation angle and previous time real-time steering wheel rotation angle.

[0055] Step two, data preprocessing is performed on the to-be-fused data of the multi-source information;

[0056] The data preprocessing includes: elimination of outliers and default value processing.

[0057] The elimination of outliers includes:

[0058] The data collected by the sensor is arranged in ascending order as follows: The different sample sizes result in different statistics, which are as follows:

[0059] In case, , ;

[0060] In case, , ;

[0061] In case, , ;

[0062] In case, , ;

[0063] In formula, is the number of sample size, is the first high-end statistics, is the first low-end statistics, is the second high-end statistics, is the second low-end statistics, is the third high-end statistics, is the third low-end statistics, is the fourth high-end statistics, is the fourth low-end statistics, for convenience, the , , , , , , and are unified as , and correspond to ;

[0064] When the confidence value is different, the sample size corresponds to different critical value , when , , is the abnormal value, and is removed ; when , , is the abnormal value, and is removed , otherwise there is no abnormal value; the specific value of the critical value is shown in Table 1.

[0065] Table 1 Critical value

[0066]

[0067] The default value processing includes:

[0068] In this embodiment, the default value is processed by linear interpolation method, and the current interpolation can be obtained according to the data before and after the interpolation point. Specifically, given the coordinates and , the interpolated value and its corresponding in the interval can be obtained according to the following formula:

[0069] ;

[0070] In the formula, is known, and the value of can be directly obtained according to the formula, that is, the default value of the current position.

[0071] The data preprocessing also includes normalizing the data:

[0072] Since the types of the nine data are different, in order to eliminate the influence of the dimensions between the data and improve the monitoring accuracy, all the data are normalized to as follows, and the calculation formula is:

[0073] ;

[0074] In the formula, is the output data after normalization, is the original data, is the minimum value of the variable , and is the maximum value of the variable .

[0075] The data after data preprocessing can not only eliminate abnormal values, but also reduce the data dialing, and the transmission of effective values can reduce energy consumption and improve network transmission rate and transmission quality.

[0076] Step three, construct a support vector machine regression model, and initialize the parameters in the model:

[0077] The training sample set is , wherein, is the output vector corresponding to the input vector , and therefore the regression function is:

[0078] ;

[0079] In the formula, is the normal vector, The input vector is mapped to a feature vector, The offset term is determined;

[0080] The training of the regression model is to find the optimal normal vector and offset term so that The approximation is as close as possible This process is ultimately converted into a convex optimization problem, which is in the form of:

[0081] ;

[0082] ;

[0083] In the formula, is the penalty coefficient, is the relaxation variable, is the insensitive loss function;

[0084] The original problem is converted into a dual problem by introducing the Lagrange multiplier, and the calculation is as follows:

[0085] ;

[0086] In the formula, , is the Lagrange coefficient.

[0087] In the nonlinear case, a suitable kernel function is introduced to map the sample to a high-dimensional space to convert it to linear, and the support vector machine regression model satisfies:

[0088] ;

[0089] ;

[0090] ;

[0091] In the formula, is the kernel function, and corresponds to and The inner product of the mapping to the feature space is used to replace the complex operation of the feature space, simplify the calculation process, and shorten the operation time, is the kernel function parameter;

[0092] Parameter initialization specifically includes: the penalty factor range is , and the kernel function parameter value range is .

[0093] Step four, establish a particle swarm optimization algorithm model to solve the optimal solution of the penalty coefficient and the kernel function parameter of the support vector machine regression model:

[0094] 1. The update of the speed and position of the particle swarm optimization algorithm model satisfies:

[0095] ;

[0096] ;

[0097] wherein, is the velocity of the i-th particle in the d-dimensional space at the iteration number k, is the velocity of the i-th particle in the d-dimensional space at the iteration number k, is the inertia weight, is the velocity of the i-th particle in the d-dimensional space at the iteration number k, is the position vector of the i-th particle in the d-dimensional space at the iteration number k, is the first acceleration factor, is a random number distributed in the interval [0, 1], is the optimal solution obtained by the i-th particle in the d-dimensional space at the iteration number k, is the position vector of the i-th particle in the d-dimensional space at the iteration number k, is the second acceleration factor, is a random number distributed in the interval [0, 1], is the optimal solution obtained by the population in the d-dimensional space at the iteration number k, is the position vector of the i-th particle in the d-dimensional space at the iteration number k; wherein, the inertia weight satisfies: ; is the position vector of the i-th particle in the d-dimensional space at the iteration number k;

[0098] wherein, the inertia weight satisfies:

[0099] ;

[0100] wherein, is the minimum weight value, is the maximum weight value, is the fitness value of the particle at any time, is the minimum fitness value in the particle swarm, is the average fitness value of the particle swarm;

[0101] The first acceleration factor is large to increase the iteration time, and the second acceleration factor is large to cause the particle to converge to a local optimal value too early, resulting in premature algorithm, therefore, adjusting the two acceleration factors to adaptively change in the optimization process is beneficial to converge to the global optimal, and the two acceleration factors satisfy:​​​​​​​​​​​​​​

[0102] ;

[0103] In the formula, 、 respectively and initial value of 、 respectively and terminal value of

[0104] 2, initialize the particle swarm:

[0105] In the particle swarm optimization algorithm, the penalty coefficient and the kernel function parameter are mapped as particles, the population size is 50, the iteration number is 1000, the initial value of the inertia weight is 0.6, and the initial values of the first acceleration factor and the second acceleration factor are both 2.

[0106] 3, the mean square error function of the support vector machine regression model is taken as the fitness function, the fitness of each particle is calculated, the individual extreme value and the global optimal value are updated, until the iteration termination condition is met, and the optimal parameter combination is obtained.

[0107] Step five, the optimal parameter combination is brought into the support vector machine regression model, so that the monitoring result is more accurate, and the optimal result of the travel route monitoring of the transportation equipment is obtained.

[0108] The open-pit mine transportation equipment multi-source information fusion monitoring method based on automatic driving designed and developed by the application, through the adaptive particle swarm optimization algorithm combined with the support vector machine regression model, simplifies the operation process, greatly shortens the operation time, and at the same time can improve the monitoring precision and improve the accuracy.

[0109] Although the embodiments of the application have been disclosed as above, it is not limited to the use listed in the specification and embodiments, and can be fully applied to various fields suitable for the application. For those skilled in the art, other modifications can be easily realized, therefore the application is not limited to specific details and the examples shown and described herein.

Claims

1. A method for monitoring a multi-source information fusion of an automatic driving based open-pit mine transport device, characterized in that, It comprises the following steps: Step one, collecting the multi-source information to be fused data according to the sampling period, wherein the multi-source information to be fused data comprises the position signal, the speed signal and the steering wheel rotation angle of the transportation equipment; The position signal, the speed signal and the steering wheel rotation angle of the transportation equipment in step one comprise the initial position signal, the previous time prediction position signal, the previous time real-time position signal, the average speed signal, the previous time prediction speed signal, the previous time real-time speed signal, the real-time maximum steering wheel rotation angle, the previous time prediction steering wheel rotation angle and the previous time real-time steering wheel rotation angle; Step two, data preprocessing of the multi-source information to be fused data; Step three, constructing a support vector machine regression model; The support vector machine regression model satisfies: ； ； ； wherein is the kernel function, and corresponds to is the inner product mapped to the feature space to replace the complex operation of the feature space, is the inner product mapped to the feature space to replace the complex operation of the feature space, is the kernel function parameter, and σ is the standard deviation of the kernel function, and b is the bias term; Step four, establishing a particle swarm optimization algorithm model to solve the optimal solution of the penalty coefficient and the kernel function parameter of the support vector machine regression model, and obtaining the optimal parameter combination; The particle swarm optimization algorithm model specifically comprises: The update of the velocity and position of the particle swarm optimization algorithm model satisfies: ； ； In the formula, For the number of iterations Time Individual particles Velocity in 2D space For inertial weights, For the number of iterations Time Individual particles Velocity in 2D space As the first acceleration factor, For distribution in Random numbers in an interval For the number of iterations Time Individual particles The optimal solution obtained in 3D space For the number of iterations Time Individual particles Position vector in 3D space The second acceleration factor, For distribution in Random numbers in an interval For the number of iterations The group at that time The optimal solution in 3D space. For the number of iterations Time Individual particles Position vector in 3D space; Wherein, the inertia weight satisfies: ； wherein is the minimum value of the weight, is the maximum value of the weight, is the fitness value of the particle at any time, is the minimum fitness value in the particle swarm, is the average fitness value of the particle swarm; The first acceleration factor and the second acceleration factor satisfy: ； wherein , are the initial values of and , , are the final values of and . Step five, bringing the optimal parameter combination into the support vector machine regression model to obtain the transportation equipment progress route monitoring result.

2. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method according to claim 1, characterized in that, The position signal, the speed signal and the steering wheel rotation angle of the transportation equipment are monitored by sensors.

3. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method of claim 2, wherein, The step two, data preprocessing of the multi-source information to be fused data, comprises outlier rejection and default value processing.

4. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method of claim 3, characterized in that, The outlier rejection specifically comprises the following steps: Step 1, arranging the data collected by the sensor in ascending order; Step 2, obtaining high-end statistics and low-end statistics according to different sample sizes; Step 3, when the confidence interval At different times, the number of sample sizes correspond to different critical values : When , , then is an outlier and is rejected ; when , , then is an outlier and is rejected , otherwise there is no outlier; wherein is a high-end statistic, is a low-end statistic, and .

5. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method of claim 3, wherein, The default value processing specifically comprises: coordinates between the known coordinates and coordinates satisfies: 。 6. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method according to claim 1, characterized in that, The preprocessing of the multi-source information to be fused data also comprises normalizing the multi-source information to be fused data: ； wherein is the result data after normalization processing, is different original data, is a variable is the minimum value of the variable is a variable is the maximum value of the variable 7. The automatic driving based open-pit mine transport equipment multi-source information fusion monitoring method of claim 1, wherein, The parameter initialization of the particle swarm optimization algorithm model is: The population size is 50, the iteration number is 1000, the initial value of the inertia weight is 0.6, and the initial values of the first acceleration factor and the second acceleration factor are both 2.

Images