CN115140685B - Forklift data driving stability control method - Google Patents
Forklift data driving stability control method Download PDFInfo
- Publication number
- CN115140685B CN115140685B CN202210528993.4A CN202210528993A CN115140685B CN 115140685 B CN115140685 B CN 115140685B CN 202210528993 A CN202210528993 A CN 202210528993A CN 115140685 B CN115140685 B CN 115140685B
- Authority
- CN
- China
- Prior art keywords
- forklift
- state
- neural network
- data
- gray
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000013528 artificial neural network Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims abstract description 14
- 230000001133 acceleration Effects 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000003062 neural network model Methods 0.000 claims description 6
- 230000002265 prevention Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 239000012535 impurity Substances 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 3
- 230000003313 weakening effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 9
- 230000001276 controlling effect Effects 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 6
- 238000000354 decomposition reaction Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 230000006872 improvement Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000005316 response function Methods 0.000 description 2
- 241000764238 Isis Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66F—HOISTING, LIFTING, HAULING OR PUSHING, NOT OTHERWISE PROVIDED FOR, e.g. DEVICES WHICH APPLY A LIFTING OR PUSHING FORCE DIRECTLY TO THE SURFACE OF A LOAD
- B66F9/00—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes
- B66F9/06—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes movable, with their loads, on wheels or the like, e.g. fork-lift trucks
- B66F9/075—Constructional features or details
- B66F9/20—Means for actuating or controlling masts, platforms, or forks
- B66F9/22—Hydraulic devices or systems
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66F—HOISTING, LIFTING, HAULING OR PUSHING, NOT OTHERWISE PROVIDED FOR, e.g. DEVICES WHICH APPLY A LIFTING OR PUSHING FORCE DIRECTLY TO THE SURFACE OF A LOAD
- B66F17/00—Safety devices, e.g. for limiting or indicating lifting force
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66F—HOISTING, LIFTING, HAULING OR PUSHING, NOT OTHERWISE PROVIDED FOR, e.g. DEVICES WHICH APPLY A LIFTING OR PUSHING FORCE DIRECTLY TO THE SURFACE OF A LOAD
- B66F9/00—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes
- B66F9/06—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes movable, with their loads, on wheels or the like, e.g. fork-lift trucks
- B66F9/075—Constructional features or details
- B66F9/07504—Accessories, e.g. for towing, charging, locking
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66F—HOISTING, LIFTING, HAULING OR PUSHING, NOT OTHERWISE PROVIDED FOR, e.g. DEVICES WHICH APPLY A LIFTING OR PUSHING FORCE DIRECTLY TO THE SURFACE OF A LOAD
- B66F9/00—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes
- B66F9/06—Devices for lifting or lowering bulky or heavy goods for loading or unloading purposes movable, with their loads, on wheels or the like, e.g. fork-lift trucks
- B66F9/075—Constructional features or details
- B66F9/0755—Position control; Position detectors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Abstract
The invention discloses a forklift data driving stability control method, which comprises the following steps: step 1, acquiring a forklift running state parameter, and establishing a forklift state prediction model, wherein the forklift state prediction model is a gray radial basis function (GM-RBF) neural network prediction model; training the prediction model until the gray radial basis function GM-RBF neural network prediction model can accurately predict the running state of the forklift; step 3, acquiring and processing real-time running parameters of the forklift, inputting the processed real-time running parameters of the forklift into the trained gray radial basis GM-RBF neural network prediction model in the step 2, and then outputting forklift state signals after model processing; and 4, controlling the state of the forklift according to the forklift state signal output in the step 3, dividing the motion state of the balanced weight type forklift, rapidly and accurately determining the motion state of the forklift, rapidly adjusting the motion state when the forklift is not in a safe state, and ensuring the stable posture of the forklift.
Description
Technical Field
The invention relates to the field of forklift stability control, in particular to a forklift data driving stability control method.
Background
The method improves the industrial production efficiency and quality, and has great significance for guaranteeing the national basic material transportation requirement and improving the material living standard. The balanced-weight forklift is one of core equipment for industrial transportation, plays an irreplaceable role in the fields of ports, construction, engineering construction and the like, and effectively improves the efficiency and quality of industrial operation, thereby improving the industrial productivity. The balanced type forklift is used as a carrying machine, the operation environment is relatively complex, the forklift body and the front axle are fixedly connected so as to be convenient for carrying and carrying, and the rear axle and the forklift body are connected in a hinged mode, so that the forklift body can swing up and down around a hinged point to meet the profiling function of a tire, but when the forklift turns at a high speed, due to the fact that the forklift is large in mass, small in track and high in gravity center, rollover accidents of the forklift are frequent, the life safety of drivers is endangered, property loss is caused, the industrial transportation progress is influenced, and the balanced type forklift has practical significance for preventing the study and engineering application of rollover of the forklift.
Because the fork truck whole vehicle dynamics system is a responsible nonlinear system, accurate judgment of the stability state of the fork truck is difficult to realize, and the judgment of the stability state of the fork truck at present is mainly achieved through expert experience and real vehicle test calibration, and a practical method and a stability control method for accurate judgment of the stability are lacked, so that the stability judgment and control method of the balanced-weight fork truck needs to be studied deeply to improve the stability of the balanced-weight fork truck.
Disclosure of Invention
The invention aims to provide a data driving stability control method for a balanced-weight forklift, which aims to solve the problems in the background art, so that rollover prevention control of the balanced-weight forklift can be optimized according to the motion state of the balanced-weight forklift, the best forklift posture is kept, and the working efficiency and stability of the balanced-weight forklift are improved.
In order to achieve the above purpose, the present invention provides the following technical solutions:
a forklift data driving stable control method comprises the following steps:
step 1, acquiring a forklift running state parameter, and establishing a forklift state prediction model according to the acquired parameter data, wherein the forklift state prediction model is a gray radial basis function (GM-RBF) neural network prediction model;
training a gray radial basis GM-RBF neural network prediction model until the gray radial basis GM-RBF neural network prediction model can accurately predict the running state of the forklift;
step 3, acquiring and processing real-time running parameters of the forklift, inputting the processed real-time running parameters of the forklift into the trained gray radial basis GM-RBF neural network prediction model in the step 2, and then outputting forklift state signals after model processing;
and 4, controlling the state of the forklift according to the forklift state signal output in the step 3.
As a further scheme of the invention: the forklift running state parameters in the step 1 comprise lateral acceleration, load and roll angle of the forklift.
As a further scheme of the invention: the establishment of the gray radial basis function GM-RBF neural network prediction model in the step 1 comprises the following steps:
step 1.1, acquiring lateral acceleration, load and roll angle data through a sensor arranged on a forklift;
step 1.2, processing the data by using a GM (1, 3) gray model, weakening random factors of the data, and finding out internal time dynamic relation between the data;
step 1.3, processing the data by using an RBF neural network;
and 1.4, adopting a serial gray neural network model, obtaining different prediction results by taking different data from the same array, combining a plurality of gray prediction results by using an RBF neural network, obtaining a series of prediction values after predicting a plurality of sequences of forklift movement by using a gray GM (1, 3) model, taking the prediction values as input samples of the combined model, taking actual values as output samples, combining the two by using the RBF neural network model by adopting a difference combination method, and simulating the deviation relation between the prediction values and the actual values and the correlation between the sequences by training the RBF neural network.
As a further scheme of the invention: the gray radial basis GM-RBF neural network prediction model training process in the step 2 comprises the following steps:
step 2.1, carrying out three-order discrete wavelet transformation on the lateral acceleration, load and roll angle data of the forklift, which are acquired by the sensor, by adopting a Malet algorithm, and repeatedly decomposing for three times, thereby finally obtaining useful characteristics of removing impurity signals;
step 2.2, carrying out data normalization, and replacing the original data by dividing the difference between the original data and the mean value thereof by the standard deviation;
step 2.3, based on the processed data, the data are randomly divided into two types, wherein one type accounts for 80% of the total number and is used for training a GM-RBF model to evaluate the rollover risk of the forklift, and the other type accounts for 20% of the total number and is used for verifying the accuracy.
As a further scheme of the invention: and when the accuracy of the training set and the verification set of the gray radial basis GM-RBF neural network prediction model is not less than 80%, judging that the gray radial basis GM-RBF neural network prediction model can accurately predict the running state of the forklift.
As a further scheme of the invention: the forklift state is divided into a safe state, a dangerous state and a unsteady state, and the forklift state division comprises the following steps:
step 3.1, acquiring real-time data of lateral acceleration, load and roll angle of the forklift, and inputting the real-time data into an established GM-RBF neural network combined prediction model;
step 3.2, outputting the combined prediction model through the GM-RBF neural network as；
And 3.3, dividing the forklift state according to the output value of the model, defining a safe state when the output value is 0, defining a dangerous state when the output value is 0-1, and defining a unstable state when the output value is 1.
As a further scheme of the invention: the step 4 of controlling the forklift state comprises the following steps:
step 4.1, acquiring a forklift state, and executing the step 4.2 if the forklift is in a safe state; step 4.3 is executed if the forklift is in a dangerous state, and step 4.4 is executed if the forklift is in an unstable state;
step 4.2, when the forklift is in a safe state, the electromagnetic valve is fully opened, the oil cylinder can freely move, and the vehicle body and the rear axle can freely rotate;
step 4.3, when the forklift is in a dangerous state, the hydraulic cylinder adjusts the lateral force based on self-optimizing fuzzy control, so that the forklift is prevented from being deteriorated to a unsteady state as much as possible;
and 4.4, when the forklift is in an unstable state, the electromagnetic valve is fully closed, the oil cylinder is directly locked, the vehicle body is fixedly connected with the rear axle, the stable area of the forklift is enlarged, the rollover prevention moment of the forklift is increased, and the stability of the forklift is improved.
Compared with the prior art, the invention has the beneficial effects that:
1. the invention divides the motion state of the balanced type forklift, can rapidly and accurately determine the motion state of the forklift, can rapidly adjust the motion state when the forklift is not in a safe state, and ensures the stable posture of the forklift.
2. The invention uses the grey system radial basis function neural network combination model, adopts the grey system to weaken the random factor of the data, and the radial basis function is the nearest neighbor cluster learning algorithm, so that the change of the motion state of the forklift can be tracked and detected in real time, an accurate nonlinear dynamics model of the forklift is not required to be established, and the invention is suitable for judging the stability of the forklift with different parameters.
3. According to the invention, the anti-rollover supporting force is regulated by changing the valve of the hydraulic cylinder of the balanced-weight forklift, the operation is convenient and easy to realize, the structure of the balanced-weight forklift does not need to be changed on a large scale, the improvement difficulty is low, the improvement cost is low, and the working efficiency of the balanced-weight forklift can be obviously improved after the improvement is finished.
Drawings
FIG. 1 is a schematic diagram of an RBF neural network topology of the present invention;
FIG. 2 is a block diagram of a GM-RBF neural network of the present invention;
FIG. 3 is a schematic diagram of a signal decomposition process according to the present invention;
FIG. 4 is a schematic diagram of the model training results of the present invention;
FIG. 5 is a schematic diagram of the identification training error of the GM-RBF neural network of the present invention;
FIG. 6 is a schematic diagram of a lateral stability control strategy for a forklift in accordance with the present invention;
FIG. 7 is a schematic diagram of the application of the GM-RBF neural network prediction model system of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1-7, in an embodiment of the present invention, a method for controlling data driving stability of a forklift includes the following steps:
step 1, acquiring a forklift running state parameter, and establishing a forklift state prediction model according to the acquired parameter data, wherein the forklift state prediction model is a gray radial basis function (GM-RBF) neural network prediction model;
the method specifically comprises the following steps:
step 1, establishing a grey radial basis function GM-RBF neural network prediction model;
the method for establishing the grey radial basis function GM-RBF neural network prediction model comprises the following steps:
step 1.1, acquiring lateral acceleration, load and roll angle data through a sensor arranged on a balanced forklift;
step 1.2, processing the data by using a GM (1, 3) gray model, weakening random factors of the data, and finding out internal time dynamic relation between the data;
the gray GM (1, 3) model prediction model is built as follows:
the original forklift data are accumulated once to generate corresponding sequences: order theFor the original sequence
(1)
Record the generation number as
(2)
Recording deviceAnd->The following relation is satisfied:
(3)
then isIs generated by one accumulation of (a) and (b). So the r-th summation is generated as
(4)
Establishing an ash differential equation, and generating a close-proximity mean value sequence:
(5)
is->Is a sequence of generation of the immediate mean of (i) that is
(6)
Wherein,for the development factor->、/>For driving coefficient +.>、/>Is the driving term.
Establishing a whitening differential equation:
(7)
approximating the time response function: parameter arrayLeast squares estimation of (c) satisfies
(8)
Wherein the method comprises the steps of
(9)
When (when)With smaller variation amplitude, the approximate time response function is
(10)
And performing subtraction reduction on the original number sequence to obtain a reduction value:
(11)
the final result is checked for accuracy. The final model fitting accuracy results are as follows, the development coefficientDrive factor->Mean fitting variance->And mean fit relative variance->The model is small and has good accuracy. The GM (1, 3) predictive model requires little sample information, but if the sample data is oscillating more, the model predictive effect is not very ideal. RBF neural networks having a strong fit to data are therefore considered as a combined model.
Step 1.3, processing the data by using an RBF neural network, and having higher learning efficiency and stronger fitting capacity;
the RBF neural network comprises the following components: the input layer is composed of signal source nodes and plays a role in transmitting data information; the hidden layer is a Gaussian kernel function, and the neural network is more accurate by adjusting parameters of the Gaussian function; the output layer is used for carrying out linear weighting on information output by neurons of the hidden layer and outputting the information, and adopts a linear optimization strategy. The topology is shown in fig. 1.
Wherein,for the 3-dimensional input layer of RBF, +.>The structure of the high-voltage power amplifier is 3-p-1, and the high-voltage power amplifier is p-dimensional hidden layer, and Y is 1-dimensional output layer. />For the output weights.
The RBF structure can be expressed as:
(12)
wherein the Gaussian basis function of RBF is
(13)
In the method, in the process of the invention,indicate->A center vector of the individual perceptual variables; />Indicate->A width of each hidden layer; />Representing the number of hidden layer neurons; />Representation vector->Is a norm of (c).
RBF neural network determines its center by self-organizing learningThe position enables the center of the hidden layer neuron to be positioned in an important area, so that information contained in the data can be expressed better. Learning for RBF neural networks requires determining three parameters: center vector->Width->Sum weight->。
(14)
Wherein,is the minimum value of the kth characteristic data, < ->Is the maximum value of the kth feature data.
Width vectorThe smaller the response of the neuron center is, the smaller; />The larger the influence range of the neurons is, the more accurate the neural network expression is, and the smoothness among the neurons is also better. Width->The initialization calculation formula of (c) is as follows:
(15)
wherein the method comprises the steps of<1 is a width adjustment coefficient that can improve the local response capability. Weight->A gradient descent method is generally adopted, and the formula is as follows:
(16)
wherein the method comprises the steps ofIs the scaling factor between neurons. />For learning factors->Is an evaluation function.
(17)
Wherein,is the desired output value; />And outputting the value for the network. The formula for calculating the root mean square error RMS of the output is as follows:
(18)
if RMS is less than or equal to epsilon, the model accuracy is considered to be satisfied, and training is finished.
And 1.4, adopting a serial gray neural network model, obtaining different prediction results by taking different data from the same array, and combining a plurality of gray prediction results by using an RBF neural network, wherein a structure diagram of the GM-RBF neural network is shown in figure 2. A series of predictions is obtained after predicting multiple sequences of forklift motion using a gray GM (1, 3) model. The predicted values are used as input samples of the combined model, the actual values are used as output samples, the two are combined by utilizing an RBF neural network model through a difference combining method, and the RBF neural network is trained to simulate the deviation relation between the predicted values and the actual values and the correlation between the sequences.
Training a gray radial basis GM-RBF neural network prediction model until the gray radial basis GM-RBF neural network prediction model can accurately predict the running state of the forklift;
the GM-RBF neural network prediction model learning training comprises the following steps:
step 2.1, carrying out three-order discrete wavelet transformation on the lateral acceleration, load and roll angle data of the forklift, which are acquired by the sensor, by adopting a Malet algorithm, and repeatedly decomposing for three times, thereby finally obtaining useful characteristics of removing impurity signals;
the formula of the Mallet algorithm is as follows:
(19)
wherein the method comprises the steps ofFor input signals j is the decomposition scale, +.>Coefficients for low-pass and high-pass filters, < >>Is->Low-frequency smoothing component at scale, +.>Is->High frequency detailed components at scale. Original signal +.>The low-frequency smoothed component and the high-frequency detailed component are decomposed, and then the low-frequency smoothed component is taken as a signal for the next decomposition. The decomposition was repeated three times, the final useful feature being a model with the impurity signal removed, the signal decomposition process being as shown in fig. 3. After wavelet transformation, the clutter interference signals are removed, and the method is smoother and higher in accuracy than the original signals.
And 2.2, carrying out data normalization, and replacing the original data by dividing the difference between the original data and the mean value thereof by the standard deviation.
Z-score is a common normalization method that reflects the relative standard distance between data and mean, eliminates the influence of different dimensions, and ensures the comparability between data. The calculation formula is as follows:
(20)
in the middle ofIs the original data; />For the mean value of the raw data, +.>Is the standard deviation of the original data.
Step 2.3, based on the processed data, the data are randomly divided into two types, wherein one type accounts for 80% of the total number and is used for training a GM-RBF model to evaluate the rollover risk of the forklift, and the other type accounts for 20% of the total number and is used for verifying the accuracy. The model training set and the verification set both have accuracy exceeding 80%, which indicates that the model has better prediction effect, and the model training result is shown in figure 4
Step 3, acquiring and processing real-time running parameters of the forklift, inputting the processed real-time running parameters of the forklift into the trained gray radial basis GM-RBF neural network prediction model in the step 2, and then outputting forklift state signals after model processing, wherein the forklift states are divided into a safe state, a dangerous state and a unsteady state, and the forklift state division comprises the following steps:
step 3.1, acquiring real-time data of lateral acceleration, load and roll angle of the forklift, and inputting the real-time data into an established GM-RBF neural network combined prediction model;
step 3.2, outputting the combined prediction model through the GM-RBF neural network as；
Step 3.3, dividing the forklift state according to the output value of the model, defining a safe state when the output value is 0, defining a dangerous state when the output value is 0-1, and defining an unstable state when the output value is 1; the GM-RBF neural network identification training error is schematically shown in FIG. 5.
And 4, controlling the state of the forklift according to the forklift state signal output in the step 3.
The step 4 of controlling the forklift state comprises the following steps:
step 4.1, acquiring a forklift state, and executing the step 4.2 if the forklift is in a safe state; step 4.3 is executed if the forklift is in a dangerous state, and step 4.4 is executed if the forklift is in an unstable state;
step 4.2, when the forklift is in a safe state, the electromagnetic valve is fully opened, the oil cylinder can freely move, so that the automobile body and the rear axle can freely rotate, and the forklift does not have strong bumpy feeling when encountering uneven road surfaces;
and 4.3, when the forklift is in a dangerous state, the hydraulic cylinder adjusts the supporting force based on self-optimizing fuzzy control, so that the forklift is prevented from being deteriorated to a unsteady state as much as possible, and the method is specifically carried out according to the following steps:
in order to control the efficient operation of the solenoid valve to regulate the cylinder lateral force, the solenoid valve must operate at an optimal point throughout the process. However, because the optimal point can be changed along with the change of working conditions, and the mathematical model of the operation of the electromagnetic valve cannot be accurately established, the opening set value of the electromagnetic valve is obtainedIt is difficult to determine in time. In order to ensure that the electromagnetic valve is always in an optimal state, a self-optimizing algorithm for adjusting the lateral force of the oil cylinder by the electromagnetic valve is provided.
The self-optimizing algorithm increases at the beginning of each self-optimizing periodThe fuzzy logic controller controls the electromagnetic valve to stably work at a set value. When the system is stable, the algorithm is according to +.>And (3) to determine whether the current set point is optimal. If it is not optimal, < >>Will increase (or decrease) in the next self-optimization cycle. Finally, the best point can be found. For example, the last self-optimizing value is set?>. At the beginning of the next cycle, +.>。/>For the set value of the current period, +.>Is a fixed self-optimizing step. When the control device is stable, if +.>Increased, increased over the previous cycle>Will increase->In the next self-optimization cycle. If->Decrease in the next self-optimization cycle, +.>Is the optimal point; otherwise, the set value will be increased again +.>. However, the self-optimization period must be set longer in consideration of the stability of the system and large delay;
and 4.4, when the forklift is in an unstable state, the electromagnetic valve is fully closed, and the oil cylinder is directly locked, so that the body is fixedly connected with the rear axle, the stable area of the forklift is enlarged, the rollover prevention moment of the forklift is greatly increased, and the stability of the forklift is improved.
In summary, the invention provides a method for controlling rollover prevention of a balanced forklift based on a GM-RBF neural network, which can divide the motion state of the forklift according to the GM-RBF neural network and adjust the valve of a hydraulic cylinder in different motion states, so as to optimize the rollover prevention control of the balanced forklift, and the structure of the balanced forklift is not greatly changed, so that the normal operation of the balanced forklift is not affected simply and conveniently, the working efficiency of the balanced forklift can be effectively improved, and the application schematic diagram of the GM-RBF neural network prediction model system is shown in fig. 7.
According to the embodiment, the gray radial basis GM-RBF neural network prediction model is built, lateral acceleration, load and roll angle of forklift motion data are obtained through the sensor, and the model is input into the GM-RBF neural network combination model for training, so that the model can accurately predict the running state of the forklift, the forklift state is divided into a safe state, a dangerous state and an unstable state, and hydraulic cylinder hierarchical control is performed according to different forklift stability states, so that the stability of the forklift is improved. The invention can accurately judge the stability state of the forklift, improve the running posture of the forklift and improve the stability of the forklift.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.
Claims (7)
1. The data driving stability control method for the forklift is characterized by comprising the following steps of:
step 1, acquiring a forklift running state parameter, and establishing a forklift state prediction model according to the acquired parameter data, wherein the forklift state prediction model is a gray radial basis function (GM-RBF) neural network prediction model;
training a gray radial basis GM-RBF neural network prediction model until the gray radial basis GM-RBF neural network prediction model can accurately predict the running state of the forklift;
step 3, acquiring and processing real-time running parameters of the forklift, inputting the processed real-time running parameters of the forklift into the trained gray radial basis GM-RBF neural network prediction model in the step 2, and then outputting forklift state signals after model processing;
and 4, controlling the state of the forklift according to the forklift state signal output in the step 3.
2. The method according to claim 1, wherein the forklift driving state parameters in step 1 include lateral acceleration, load and roll angle of the forklift.
3. The forklift data driving stability control method according to claim 1, wherein the establishment of the gray radial basis function GM-RBF neural network prediction model in step 1 comprises the steps of:
step 1.1, acquiring lateral acceleration, load and roll angle data through a sensor arranged on a forklift;
step 1.2, processing the data by using a GM (1, 3) gray model, weakening random factors of the data, and finding out internal time dynamic relation between the data;
step 1.3, processing the data by using an RBF neural network;
and 1.4, adopting a serial gray neural network model, obtaining different prediction results by taking different data from the same array, combining a plurality of gray prediction results by using an RBF neural network, obtaining a series of prediction values after predicting a plurality of sequences of forklift movement by using a gray GM (1, 3) model, taking the prediction values as input samples of the combined model, taking actual values as output samples, combining the two by using the RBF neural network model by adopting a difference combination method, and simulating the deviation relation between the prediction values and the actual values and the correlation between the sequences by training the RBF neural network.
4. The forklift data driving stability control method according to claim 1, wherein the gray radial basis GM-RBF neural network prediction model training process in step 2 comprises the following steps:
step 2.1, carrying out three-order discrete wavelet transformation on the lateral acceleration, load and roll angle data of the forklift, which are acquired by the sensor, by adopting a Malet algorithm, and repeatedly decomposing for three times, thereby finally obtaining useful characteristics of removing impurity signals;
step 2.2, carrying out data normalization, and replacing the original data by dividing the difference between the original data and the mean value thereof by the standard deviation;
step 2.3, based on the processed data, the data are randomly divided into two types, wherein one type accounts for 80% of the total number and is used for training a GM-RBF model to evaluate the rollover risk of the forklift, and the other type accounts for 20% of the total number and is used for verifying the accuracy.
5. The method for controlling data driving stability of a forklift according to claim 4, wherein when the accuracy of the training set and the verification set of the gray radial basis GM-RBF neural network prediction model is not less than 80%, the gray radial basis GM-RBF neural network prediction model is determined to be capable of accurately predicting the running state of the forklift.
6. The method for controlling data driving stability of a forklift according to claim 1, wherein the forklift states are classified into a safe state, a dangerous state, and a unstable state, and the forklift state classification includes the steps of:
step 3.1, acquiring real-time data of lateral acceleration, load and roll angle of the forklift, and inputting the real-time data into an established GM-RBF neural network combined prediction model;
step 3.2, outputting the combined prediction model through the GM-RBF neural network as；
And 3.3, dividing the forklift state according to the output value of the model, defining a safe state when the output value is 0, defining a dangerous state when the output value is 0-1, and defining a unstable state when the output value is 1.
7. The method for controlling the stability of a forklift data drive according to claim 6, wherein the controlling of the forklift state in step 4 comprises the steps of:
step 4.1, acquiring a forklift state, and executing the step 4.2 if the forklift is in a safe state; step 4.3 is executed if the forklift is in a dangerous state, and step 4.4 is executed if the forklift is in an unstable state;
step 4.2, when the forklift is in a safe state, the electromagnetic valve is fully opened, the oil cylinder can freely move, and the vehicle body and the rear axle can freely rotate;
step 4.3, when the forklift is in a dangerous state, the hydraulic cylinder adjusts the lateral force based on self-optimizing fuzzy control, so that the forklift is prevented from being deteriorated to a unsteady state as much as possible;
and 4.4, when the forklift is in an unstable state, the electromagnetic valve is fully closed, the oil cylinder is directly locked, the vehicle body is fixedly connected with the rear axle, the stable area of the forklift is enlarged, the rollover prevention moment of the forklift is increased, and the stability of the forklift is improved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210528993.4A CN115140685B (en) | 2022-05-16 | 2022-05-16 | Forklift data driving stability control method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210528993.4A CN115140685B (en) | 2022-05-16 | 2022-05-16 | Forklift data driving stability control method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115140685A CN115140685A (en) | 2022-10-04 |
| CN115140685B true CN115140685B (en) | 2024-02-13 |
Family
ID=83406726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210528993.4A Active CN115140685B (en) | 2022-05-16 | 2022-05-16 | Forklift data driving stability control method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115140685B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117842909A (en) * | 2024-03-08 | 2024-04-09 | 美通重工有限公司 | Automatic leveling method and system for high-altitude vehicle platform |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104270281A (en) * | 2014-09-03 | 2015-01-07 | 河海大学 | Web Service QoS (Quality of Service) combined forecasting method |
| CN104573170A (en) * | 2014-10-13 | 2015-04-29 | 芜湖扬宇机电技术开发有限公司 | Method for determining initial pressure of optimal operation of steam turbine |
| CN107368928A (en) * | 2017-08-03 | 2017-11-21 | 西安科技大学 | A kind of combination forecasting method and system of ancient building sedimentation |
| CN111143989A (en) * | 2019-12-25 | 2020-05-12 | 北京无线电计量测试研究所 | Frequency adjustment amount calculation method, module, system, storage medium, and device |
| DE102019113765A1 (en) * | 2019-05-23 | 2020-11-26 | Jungheinrich Ag | Method for controlling a hydraulic system of a mobile work machine and mobile work machine |
-
2022
- 2022-05-16 CN CN202210528993.4A patent/CN115140685B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104270281A (en) * | 2014-09-03 | 2015-01-07 | 河海大学 | Web Service QoS (Quality of Service) combined forecasting method |
| CN104573170A (en) * | 2014-10-13 | 2015-04-29 | 芜湖扬宇机电技术开发有限公司 | Method for determining initial pressure of optimal operation of steam turbine |
| CN107368928A (en) * | 2017-08-03 | 2017-11-21 | 西安科技大学 | A kind of combination forecasting method and system of ancient building sedimentation |
| DE102019113765A1 (en) * | 2019-05-23 | 2020-11-26 | Jungheinrich Ag | Method for controlling a hydraulic system of a mobile work machine and mobile work machine |
| CN111143989A (en) * | 2019-12-25 | 2020-05-12 | 北京无线电计量测试研究所 | Frequency adjustment amount calculation method, module, system, storage medium, and device |
Non-Patent Citations (1)
| Title |
|---|
| "平衡重式叉车防侧翻分层控制研究";夏光 等;中国机械工;第30卷(第17期);第2066-2075页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115140685A (en) | 2022-10-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Papadimitrakis et al. | Active vehicle suspension control using road preview model predictive control and radial basis function networks | |
| Soleymani et al. | Adaptive fuzzy controller for vehicle active suspension system based on traffic conditions | |
| Gobbi et al. | Optimising a car chassis | |
| CN112776551B (en) | Semi-active control method and system for magneto-rheological suspension based on motion pattern | |
| CN115140685B (en) | Forklift data driving stability control method | |
| CN110386191B (en) | Electro-hydraulic intelligent steering system performance monitor and performance optimization method | |
| CN109002599B (en) | Automobile ride comfort optimization analysis method based on Taguchi experiment | |
| Lin et al. | Fuzzy logic control of vehicle suspension systems | |
| Lin et al. | Design of a grey-prediction self-organizing fuzzy controller for active suspension systems | |
| Lee et al. | Deep reinforcement learning of semi-active suspension controller for vehicle ride comfort | |
| Robert et al. | Fuzzy control of active suspension system based on quarter car model | |
| CN111967308A (en) | Online road surface unevenness identification method and system | |
| Senthil et al. | Optimal preview control of a two-dof vehicle model using stochastic optimal control theory | |
| Chen et al. | Study on PID control of vehicle semi-active suspension based on genetic algorithm | |
| Allam et al. | Vibration control of active vehicle suspension system using fuzzy logic algorithm | |
| Arslan et al. | Quarter Car Active Suspension System Control Using Fuzzy Controller | |
| Hardier | Recurrent RBF networks for suspension system modeling and wear diagnosis of a damper | |
| CN113705865B (en) | Automobile stability factor prediction method based on deep neural network | |
| Moran et al. | Optimal preview control of rear suspension using nonlinear neural networks | |
| Feng et al. | Bandwidth-limited active suspension controller for an off-road vehicle based on co-simulation technology | |
| Huang et al. | Application of a fuzzy enhance adaptive control on active suspension system | |
| Pasterkamp et al. | Optimal design of neural networks for estimation of tyre/road friction | |
| Herz et al. | Estimation of load-time curves using recurrent neural networks based on can bus signals | |
| Mulla et al. | Investigating performance of quarter-car semi-active suspension with skyhook, fuzzy logic, adaptive neuro-fuzzy inference system control strategies for ISO-classified road disturbance | |
| Seeger | Neural Network-Based Wheel Suspension Modeling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |