A method for analyzing the load of engineering vehicles

Abstract

The present application provides a method for analyzing the load of an engineering operation vehicle, comprising: predicting the change trend of the hydraulic cylinder pressure according to the fork descent speed and the load, establishing a pressure prediction model through a support vector machine algorithm, and when the predicted pressure exceeds a safety threshold, controlling the buffer valve opening in advance to achieve predictive control of the pressure; establishing a fuzzy rule base in combination with the forklift's load, fork descent speed, and hydraulic cylinder pressure to achieve fuzzy adaptive control of the buffer valve opening, automatically adjusting the buffer valve opening according to different working conditions, and ensuring that the hydraulic system pressure is stable within a safe range; establishing a nonlinear mapping relationship between the load, fork descent speed, and buffer valve opening, training and optimizing a neural network model, achieving intelligent control of the buffer valve opening, and automatically adjusting the buffer valve opening according to the load and fork descent speed collected in real time.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for analyzing the load of an engineering operation vehicle. Background Art

[0002] When the forklift is fully loaded and the fork is rapidly lowered, the pressure in the hydraulic cylinder will rise sharply due to the inertial force of the load, resulting in a large fluctuation in the system pressure. This pressure pulsation will have an adverse effect on the stability and reliability of the hydraulic system and may even cause safety accidents. In order to alleviate this problem, a buffer valve is usually installed in the hydraulic circuit, and the opening of the throttle valve core is used to adjust the return oil flow, thereby achieving the purpose of buffering the pressure pulsation. However, the opening area of ​​the buffer valve will directly affect the dynamic response characteristics of the system. If the opening area is too small, the buffering effect will not be obvious and the pressure pulsation will still be serious; if the opening area is too large, the fork will be lowered too slowly, affecting the operating efficiency. How to optimize the opening area of ​​the buffer valve according to different load conditions is a technical problem that needs to be solved urgently. In addition, there is a complex nonlinear relationship between the load and the buffer pressure, which is affected by many factors, such as hydraulic oil temperature, pipeline resistance, seal friction, etc. Summary of the invention

[0003] The present invention provides a method for analyzing the load of an engineering operation vehicle, which mainly includes:

[0004] Obtain the load weight of the forklift under full load, identify the descending speed of the fork when it descends quickly, calculate the pressure change curve of the hydraulic cylinder under full load condition of the forklift through the established hydraulic system mathematical model, and judge whether the pressure exceeds the preset safety threshold. If it exceeds the threshold, the pressure regulation function of the hydraulic buffer valve is triggered;

[0005] A pressure sensor is used to detect the inlet and outlet pressures of the hydraulic buffer valve in real time to obtain the pressure pulsation signal. The pressure pulsation signal is analyzed in the time-frequency domain to obtain the pressure pulsation amplitude at different frequencies, and it is judged whether the pressure pulsation exceeds the preset safety threshold. If it exceeds the threshold, the opening of the buffer valve is controlled to reduce the pressure pulsation.

[0006] Obtain the opening area of ​​the buffer valve. For different buffer valve opening areas, analyze the dynamic response characteristics of the hydraulic system through simulation, obtain the change curves of system pressure, flow, and valve opening, thereby optimizing the buffer valve opening control strategy and determining the target buffer valve opening area to obtain the target system dynamic response performance;

[0007] Under different load conditions, the hydraulic cylinder pressure and buffer valve opening data are collected, and a polynomial regression model of load and buffer valve pressure is established through least squares fitting to obtain the correlation between load and target buffer pressure. According to the correlation between load and target buffer pressure, adaptive adjustment of buffer valve pressure is achieved.

[0008] According to the fork lowering speed and load, the change trend of the hydraulic cylinder pressure is predicted, and the pressure prediction model is established through the support vector machine algorithm. When the predicted pressure exceeds the safety threshold, the buffer valve opening is controlled in advance to achieve predictive control of the pressure;

[0009] Combined with the forklift's load, fork lowering speed, and hydraulic cylinder pressure, a fuzzy rule base is established to achieve fuzzy adaptive control of the buffer valve opening. The buffer valve opening is automatically adjusted according to different working conditions to ensure that the hydraulic system pressure is stable within a safe range.

[0010] A nonlinear mapping relationship between load weight, fork descent speed and buffer valve opening is established, and the neural network model is trained and optimized to realize intelligent control of buffer valve opening. The buffer valve opening is automatically adjusted according to the load weight and fork descent speed collected in real time.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses a method for analyzing the load of an engineering operation vehicle. The present invention obtains the load weight of a forklift under a fully loaded state and the descending speed of a forklift when the fork is rapidly descended, establishes a hydraulic system mathematical model, calculates the pressure change curve of the hydraulic cylinder under a fully loaded condition, and determines whether the pressure exceeds a preset safety threshold. If the pressure exceeds the threshold, the pressure adjustment function of the hydraulic buffer valve is triggered. At the same time, the present invention uses a pressure sensor to detect the inlet and outlet pressures of the hydraulic buffer valve in real time, obtains a pressure pulsation signal, and determines whether the pressure pulsation exceeds the safety threshold through time-frequency domain analysis. If the pressure pulsation exceeds the threshold, the buffer valve opening is controlled to reduce the pressure pulsation. In addition, the present invention also optimizes the buffer valve opening control strategy through simulation analysis, and establishes a polynomial regression model of the load weight and the buffer valve pressure to achieve adaptive adjustment of the buffer valve pressure. Finally, the present invention combines the load weight, the fork descent speed, and the hydraulic cylinder pressure of the forklift to establish a fuzzy rule base and a neural network model to achieve fuzzy adaptive control and intelligent control of the buffer valve opening. The present invention effectively solves the pressure shock and pressure pulsation problems of the forklift hydraulic system under a fully loaded condition, ensures that the pressure of the hydraulic system is stable within a safe range, and improves the safety and reliability of the forklift. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of a method for analyzing the load of an engineering operation vehicle.

[0014] Figure 2 A schematic diagram of a method for analyzing the load of an engineering operation vehicle according to the present invention.

[0015] Figure 3 It is another schematic diagram of a method for analyzing the load of an engineering operation vehicle according to the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1-3 In this embodiment, a method for analyzing the load of an engineering operation vehicle may specifically include:

[0018] S101, obtaining the load weight of the forklift under the fully loaded state, identifying the descending speed of the fork when it descends rapidly, calculating the pressure change curve of the hydraulic cylinder under the fully loaded condition of the forklift through the established hydraulic system mathematical model, and judging whether the pressure exceeds the preset safety threshold, if it exceeds the threshold, triggering the pressure adjustment function of the hydraulic buffer valve.

[0019] The measurement data of the gravity sensor and the lifting angle sensor are obtained, and the fork load moment value is obtained according to the measurement data and the fork arm length according to the load moment calculation formula M=F×L×cosθ, wherein F is the gravity value of the fork, L is the fork arm length, and θ is the lifting angle of the fork; according to the fork load moment value, a Kalman filter is used to perform noise filtering processing, and the fork position change data is collected within a fixed sampling time interval through a displacement sensor, and the hydraulic oil flow value is obtained according to the flow calculation formula Q=v×A, wherein v is the fork descent speed, and A is the piston area; according to the hydraulic oil flow value, the pipeline pressure loss value is calculated according to the Bernoulli equation ΔP=ρv2 / 2, wherein Where ρ is the density of hydraulic oil. The hydraulic oil pressure data is collected through the oil pump pressure sensor, and the hydraulic cylinder pressure value is obtained according to the pressure calculation formula P=F / A+ΔP; the hydraulic cylinder pressure value is curve-fitted by cubic spline interpolation to obtain the hydraulic cylinder pressure change curve, and it is judged whether the hydraulic cylinder pressure value exceeds the preset pressure threshold; if the hydraulic cylinder pressure value exceeds the preset pressure threshold, the hydraulic buffer valve is triggered, and a proportional-integral controller is used to output a control signal according to the control formula u=Kp×e+Ki×∫edt, wherein e is the deviation value between the actual pressure and the threshold, Kp is the proportional coefficient, and Ki is the integral coefficient, and the buffer valve opening is adjusted according to the control signal.

[0020] Specifically, the fork load data and lifting angle data are obtained through the gravity sensor and the lifting height sensor, and the load moment M = F × L × cosθ is calculated, where F is the gravity value of the fork, L is the fork arm length, and θ is the lifting angle of the fork. The sensor data is subjected to noise filtering by the Kalman filter to obtain the corrected load moment value. The displacement sensor is used to collect the fork position change data within a fixed 10ms sampling time interval, and the fork descent speed v is calculated. The hydraulic oil flow Q = v × A is calculated according to the piston area A, and the pipeline pressure loss is calculated according to the Bernoulli equation ΔP = ρv2 / 2, where ρ is the hydraulic oil density. The hydraulic oil pressure data is collected from the oil pump pressure sensor, and the hydraulic cylinder pressure is calculated according to P = F / A + ΔP, where P is the hydraulic cylinder pressure, F is the load force, A is the piston area, and ΔP is the pipeline pressure loss. The interpolation points are selected at intervals of 50ms and the cubic spline interpolation is used to calculate the hydraulic cylinder pressure change curve. If the hydraulic cylinder pressure value exceeds the preset pressure threshold, the hydraulic buffer valve is triggered, and the proportional integral controller output u=Kp×e+Ki×∫edt is used to control the buffer valve opening, where e is the deviation between the actual pressure and the threshold, Kp is the proportional coefficient, and Ki is the integral coefficient. The buffer chamber pressure value is obtained from the hydraulic buffer valve feedback data, and the buffer valve opening is dynamically adjusted according to the pressure difference between the buffer pressure and the main pressure. The forklift load sensor installs a strain gauge at the root of the fork to collect strain signals, and the angle sensor collects the forklift lifting angle signal. The fork length is fixed to 1200mm. The fork is calibrated to zero position signal in the empty state. When the load is 1000kg, the strain signal is 2000με, and the lifting angle is in the range of 0-60 degrees. The Kalman filter sets the process noise covariance to 0.01 and the measurement noise covariance to 0.1. The steady-state load torque value is obtained by iterative calculation. The displacement sensor collects the fork position data within a 10ms sampling period. The fork lowering speed is in the range of 0-300mm / s. The piston diameter is 80mm. According to the piston area of ​​5026mm 2 The calculated hydraulic oil flow rate is 0-90L / min and the hydraulic oil density is 0.9g / cm 3 , the pipeline pressure loss is obtained through the Bernoulli equation. The oil pump pressure sensor has a range of 0-35MPa and a sampling frequency of 100Hz. When the fork is fully loaded with 1000kg, the corresponding hydraulic cylinder pressure is 16MPa. After considering the pipeline pressure drop, the actual working pressure is 18MPa. The pressure data points are selected at intervals of 50ms to perform cubic spline interpolation to fit the pressure curve. The preset pressure threshold of the hydraulic buffer valve is 20MPa, the proportional coefficient Kp of the proportional-integral controller is set to 0.8, the integral coefficient Ki is set to 0.05, the buffer valve opening range is 0-100%, and when the pressure difference between the buffer chamber pressure and the main pressure obtained by feedback is 2MPa, the buffer valve opening is adjusted to 50% to achieve dynamic pressure balance.

[0021] S102. Use a pressure sensor to detect the inlet and outlet pressures of the hydraulic buffer valve in real time, obtain a pressure pulsation signal, perform time-frequency domain analysis on the pressure pulsation signal, obtain the pressure pulsation amplitude at different frequencies, and determine whether the pressure pulsation exceeds a preset safety threshold. If it exceeds the threshold, control the opening of the buffer valve to reduce the pressure pulsation.

[0022] A pressure sensor is used to collect the inlet and outlet pressure signals of the hydraulic buffer valve, and the pressure signal is sampled and quantized through a data acquisition card to obtain a pressure pulsation sampling data sequence; based on the pressure pulsation sampling data sequence, a bandpass filter is used to remove power frequency interference and high-frequency noise, and a filtered pressure signal is obtained through wavelet decomposition processing; a fast Fourier transform is performed on the filtered pressure signal to obtain a pressure pulsation amplitude spectrum within a frequency range; a frequency component with an amplitude greater than twice the mean is extracted from the pressure pulsation amplitude spectrum as a main frequency component, and a linear mapping relationship of the opening adjustment amount is calculated based on the main frequency component; if the amplitude of the main frequency component exceeds a preset safety threshold, the recursive least squares method is used to calculate the optimal opening value, and the buffer valve opening is compensated and adjusted according to the opening adjustment rate.

[0023] Specifically, the pressure signals of the hydraulic buffer valve inlet and outlet are collected through a pressure sensor with a range of 0-40MPa and an accuracy of 0.1%, and a 16-bit high-precision data acquisition card is used to sample and quantify the pressure pulsation signal. The sampling frequency is set to 1000Hz, and the minimum resolution is 0.6kPa to obtain the pressure pulsation sampling data sequence. The Butterworth bandpass filter with a passband of 10-200Hz and a stopband attenuation of 40dB is used to remove the power frequency interference and high-frequency noise from the collected pressure pulsation data sequence to obtain the filtered pressure signal. The signal is decomposed into 4 layers using the db4 wavelet, and the root mean square value and peak-to-peak value of the pressure pulsation signal are calculated to judge the signal processing quality. The pressure pulsation signal after filtering is subjected to a 2048-point fast Fourier transform to obtain the pressure pulsation amplitude spectrum of different frequencies in the range of 0-500Hz. The frequency component with an amplitude greater than 2 times the mean is extracted from the amplitude spectrum as the main frequency component, and the linear mapping relationship between the amplitude corresponding to the main frequency component and the opening adjustment amount is calculated. If the amplitude of the main frequency component exceeds the preset safety threshold, the recursive least square method with a forgetting factor of 0.95 is used to calculate the optimal opening value, and the pressure change before and after the opening adjustment is obtained from the feedback data. The buffer valve opening is proportionally compensated according to the limit of the opening adjustment rate of 10% / s, and the compensation coefficient is proportional to the pressure amplitude. The pressure sensor collects hydraulic signals at the inlet and outlet installation positions of the buffer valve. When the steady-state value of the inlet pressure is 20MPa, the outlet pressure is 18MPa, and the peak-to-peak value of the pressure pulsation is 2MPa. The sampling resolution of the data acquisition card is 0.6kPa. 5000 data points are obtained by continuously collecting for 5 seconds at a sampling frequency of 1000Hz. The Butterworth bandpass filter with a passband of 10-200Hz is used to filter the original signal. After filtering, the peak-to-peak value of the signal is reduced to 1.5MPa, and the signal-to-noise ratio is increased to 40dB. The db4 wavelet is used to decompose the filtered signal in 4 layers. The correlation coefficient between the reconstructed signal and the original signal after decomposition is 0.95, the root mean square value is 0.4MPa, and the peak-to-peak value is 1.2MPa. The 2048-point data was fast Fourier transformed to obtain the amplitude spectrum in the frequency range of 0-500Hz. The main pulsation frequencies appeared at 80Hz, 160Hz and 240Hz, with corresponding amplitudes of 0.3MPa, 0.2MPa and 0.1MPa, respectively, and the average was 0.12MPa. The frequency component with an amplitude greater than 0.24MPa was extracted as the main frequency component. When the amplitude of the 80Hz frequency component of 0.3MPa exceeded the preset threshold of 0.25MPa, the recursive least squares method was iterated 5 times to obtain an opening compensation value of 15%, and the pressure change was 0.1MPa / 1% opening. Under the condition that the opening adjustment rate was limited to 10% / s, the buffer valve opening was gradually adjusted from 35% to 50%, and the pressure pulsation amplitude was reduced to 0.2MPa.

[0024] S103, obtaining the buffer valve opening area, and analyzing the dynamic response characteristics of the hydraulic system through simulation for different buffer valve opening areas, obtaining the change curves of system pressure, flow, and valve opening, thereby optimizing the buffer valve opening control strategy and determining the target buffer valve opening area to obtain the target system dynamic response performance.

[0025] A displacement sensor with a measuring range is used to collect the valve core displacement signal, and the valve port contour line is identified by the Canny edge detection algorithm according to the displacement signal to obtain the valve port opening area value; for the valve port opening area value, under the pressure inlet boundary and the pressure outlet boundary conditions, the flow channel between the hydraulic cylinder and the buffer valve is calculated by the grid division method to obtain the pressure distribution field and the flow field data; according to the pressure distribution field and the flow field data, the time domain finite difference algorithm is used to calculate the response characteristics to obtain the pressure rise time, the flow response delay, the pressure overshoot and the flow fluctuation value; for the pressure rise time, the flow response delay, the pressure overshoot and the flow fluctuation value, the cubic polynomial least squares fitting algorithm is used to calculate to obtain the opening and dynamic characteristic correlation curve; if the opening and dynamic characteristic correlation curve meets the pressure response time weight, overshoot weight and flow delay weight requirements, the genetic algorithm is used to optimize the opening response curve to obtain the optimal opening area value.

[0026] Specifically, the displacement signal of the buffer valve core is collected by a displacement sensor with a range of 0-10mm and an accuracy of 0.01mm. The Canny edge detection algorithm with a detection threshold of 50 is used to identify the valve port contour line. According to the corresponding relationship between the opening and the area in the valve core structure size, the valve port opening area value within the opening range of 0%-100% is calculated, and discretization is performed according to the 0.1mm grid size. Under the conditions of 20MPa at the pressure inlet boundary and 18MPa at the outlet boundary, 20,000 grid units are divided into the flow channel between the hydraulic cylinder and the buffer valve, and the height of the first layer of the wall grid is set to 0.02mm. The Reynolds stress turbulent flow meter is used to calculate the pressure distribution field and flow field data within the opening range of 0%-100%, and the pressure flow characteristic curve under each opening is obtained. According to the 0.1ms time step within the 100ms sampling time, the time domain finite difference algorithm is used to calculate the response characteristics of the pressure distribution field and flow field data, and the pressure rise time, flow response delay, pressure overshoot and flow fluctuation value are obtained. The cubic polynomial least squares fitting is used to obtain the correlation curve between the opening and the dynamic characteristics. The genetic algorithm was set with a population size of 50, number of iterations of 100, crossover probability of 0.8, and mutation probability of 0.1. The pressure response time weight was 0.4, the overshoot weight was 0.3, and the flow delay weight was 0.3 as the optimization targets. The opening response curve was optimized and calculated within the opening range of 0-100%, and the optimal opening area value that met the requirements of pressure rise time less than 50ms, overshoot less than 5%, and flow response delay less than 30ms was obtained. The displacement sensor collects the valve core displacement data in real time at the installation point of the buffer valve. When the valve core displacement increases from 0mm to 5mm, the Canny edge detection algorithm detects that the length of the valve port contour line increases from 150 pixels to 300 pixels. When the edge detection threshold is set to 50, the detection accuracy reaches 0.02mm, and the valve port opening area increases from 0mm 2 Increased to 80mm 2, 800 grid units are obtained after discretization according to the 0.1mm grid size. Under the conditions of inlet pressure 20MPa and outlet pressure 18MPa, the flow rate is calculated to be 0L / min when the opening is 0%, 40L / min when the opening is 25%, 85L / min when the opening is 50%, 120L / min when the opening is 75%, and 150L / min when the opening is 100%. 100ms data is collected continuously in the 0.1ms sampling interval. When the opening is 25%, the pressure rise time is 80ms, the overshoot is 8%, and the flow response delay is 45ms; when the opening is 50%, the pressure rise time is 65ms, the overshoot is 6%, and the flow response delay is 38ms; when the opening is 75%, the pressure rise time is 45ms, the overshoot is 4%, and the flow response delay is 25ms. After 30 generations of optimization calculations using the genetic algorithm, the optimal opening was 72%, at which the pressure rise time was 48ms, the overshoot was 4.8%, and the flow response delay was 28ms, meeting the dynamic response index requirements.

[0027] S104. Under different load conditions, collect the hydraulic cylinder pressure and buffer valve opening data, establish a polynomial regression model of the load and buffer valve pressure through least squares fitting, obtain the correlation between the load and the target buffer pressure, and realize adaptive adjustment of the buffer valve pressure based on the correlation between the load and the target buffer pressure.

[0028] The load signal of the hydraulic forklift is collected by a load sensor, and the load signal is sampled by a data acquisition card for sixteen-bit quantization to obtain load data, and the load data is processed by a median filter; according to the load data, a pressure sensor is used to collect the hydraulic cylinder circuit pressure and the inlet and outlet pressures of the buffer valve, and the pressure sensor obtains a corresponding data group of the buffer valve opening value and the pressure data value; for the corresponding data group, a third-order polynomial least squares method is used for curve fitting, and the polynomial coefficients are obtained by the curve fitting; according to the polynomial coefficients, the Lagrange interpolation method is used to smooth the curve, and the Lagrange interpolation method obtains a smooth curve with a pressure smoothness error less than a preset threshold; for the smooth curve, a mean filter is used to calculate the target buffer pressure value, and the target buffer pressure value is adjusted by a proportional integral controller for closed-loop opening, and the proportional integral controller performs adaptive control on the buffer valve opening.

[0029] Specifically, the load data of hydraulic forklifts under different loads are collected through a load sensor with a range of 0-3000kg and an accuracy of 0.1%. The load signal is sampled and quantized with 16 bits using a high-precision data acquisition card. The sampling frequency is set to 100Hz, and the data group corresponding to the load and hydraulic cylinder pressure in the range of 0-2000kg is obtained. The sampled data is processed using a median filter with a window length of 10. A pressure sensor is used to collect the hydraulic cylinder circuit pressure and the buffer valve inlet and outlet pressures in real time within a 5-second time window. The pressure data is collected every 10ms. The steady-state data with a pressure fluctuation of less than 1% is selected to record the buffer valve opening value and pressure data value in the opening range of 0-100%, and the corresponding data group of load and pressure is obtained. The interpolation points of the collected data group are selected at intervals of 100kg, and the third-order polynomial least squares method with coefficients a0, a1, a2, and a3 is used to fit the regression curve. The fitting accuracy is set to 0.98, and the fitting coefficient is updated every 500 milliseconds. The Lagrange interpolation method with a pressure smoothness error of less than 0.5% is used to smooth the curve. According to the fitted polynomial coefficients, a mean filter with a sliding window length of 10 is used to calculate the real-time target buffer pressure value. Through a proportional integral controller with a proportional coefficient of 0.8 and an integral time of 0.5s, the buffer valve opening is closed-loop regulated within a 20ms control cycle to achieve pressure adaptive control. The load sensor outputs a signal of 4mA when the forklift loads 0kg, and an output signal of 20mA when the load is 3000kg. A 16-bit sampling accuracy data acquisition card is used to collect the load signal, with a minimum resolution of 0.05kg. The measurement accuracy reaches 0.1% under a median filter window length of 10 data points. The pressure sensor collects data within a 5-second time window. When the load is 500kg, the hydraulic cylinder pressure is 8MPa, and the buffer pressure is 6MPa when the opening is 25%; when the load is 1000kg, the hydraulic cylinder pressure is 12MPa, and the buffer pressure is 9MPa when the opening is 50%; when the load is 1500kg, the hydraulic cylinder pressure is 16MPa, and the buffer pressure is 12MPa when the opening is 75%; when the load is 2000kg, the hydraulic cylinder pressure is 20MPa, and the buffer pressure is 15MPa when the opening is 100%. According to the 100kg load interval, 20 interpolation points are selected, and the coefficients a0=5.2, a1=0.008, a2=-2.1×10-6, a3=4.5×10-10 are obtained by fitting with a third-order polynomial. The correlation coefficient between the fitting curve and the measured data is 0.992. Under the action of the proportional-integral controller, when the load increases from 1000kg to 1500kg, the buffer valve opening is adjusted from 50% to 75% within 0.5 seconds, the buffer pressure rises from 9MPa to 12MPa, the pressure overshoot is 0.3MPa, and the adjustment time is 0.8 seconds.

[0030] S105. Predict the change trend of the hydraulic cylinder pressure according to the fork lowering speed and the load, establish a pressure prediction model through the support vector machine algorithm, and when the predicted pressure exceeds the safety threshold, control the buffer valve opening in advance to achieve predictive control of the pressure.

[0031] The hydraulic cylinder pressure signal is collected by the pressure sensor, and the fork displacement signal is collected by the displacement sensor. The data is quantized according to the pressure signal and the displacement signal to obtain a quantized data sequence; the pressure change process samples are selected for the quantized data sequence, and the pressure change process samples are normalized by the maximum and minimum method to obtain a normalized data set, and the normalized data set includes a load parameter, a descent speed parameter, and a pressure change parameter; the support vector machine prediction model is trained according to the normalized data set, and the support vector machine prediction model adopts a radial basis kernel function, and the support vector parameters are obtained by cross-validation; the support vector machine prediction model is used to predict the pressure change trend, and the prediction result is smoothed by a sliding average filter to obtain a smoothed predicted pressure value; if the smoothed predicted pressure value exceeds the safety threshold, a fuzzy controller is used to perform opening compensation, and the opening compensation value is obtained by defuzzification calculation using the center of gravity method according to the pressure deviation and the deviation change rate, and the buffer valve opening is adjusted.

[0032] Specifically, the hydraulic cylinder pressure signal is collected through a pressure sensor with a range of 0-40MPa and an accuracy of 0.1%, and the fork displacement signal is collected through a displacement sensor with a range of 0-2m and an accuracy of 0.1mm. A high-precision data acquisition card is used to quantize the pressure and displacement signals in 16 bits, and data is collected every 10ms to record the corresponding relationship between the pressure change and the descent speed in the range where the pressure fluctuation is less than 2% and the descent speed is constant. 1000 sets of complete pressure change processes are selected from the pressure data sequence collected by the pressure sensor, including three variables: load, descent speed, and pressure change. The maximum and minimum value method is used to normalize the original data, and the data is divided into a training set and a validation set according to a ratio of 8:2. The training data is cross-validated at 5 folds, and the validation error is less than 3%. The support vector machine algorithm is used to establish a pressure prediction model, and the radial basis kernel function is selected. The penalty factor is set to 100, the slack variable is 0.01, the number of support vectors obtained by training is 200, the cross-validation accuracy reaches 95%, the prediction time lead is set to 500ms, and the prediction results are smoothed by a sliding average filter. According to the pressure change trend predicted by the support vector machine, if the predicted pressure value exceeds the safety threshold of 20MPa, the fuzzy controller based on 7 fuzzy subsets is used for opening compensation. The pressure deviation and deviation change rate are used as input. The opening compensation value is obtained by the center of gravity method to adjust the buffer valve opening. When the fork is unloaded and the lowering speed is 300mm / s, the displacement sensor sampling signal changes from 0mm to 1800mm, and the hydraulic cylinder pressure is maintained at 5MPa; when the load is 500kg, the lowering speed is 250mm / s, and the hydraulic cylinder pressure is 8MPa; when the load is 1000kg, the lowering speed is 200mm / s, and the hydraulic cylinder pressure is 12MPa; when the load is 1500kg, the lowering speed is 150mm / s, and the hydraulic cylinder pressure is 16MPa; when the load is 2000kg, the lowering speed is 100mm / s, and the hydraulic cylinder pressure is 20MPa. After normalization by the maximum and minimum value method, the normalized value of load weight ranges from 0.25 to 1, the normalized value of descent speed ranges from 0.33 to 1, and the normalized value of pressure ranges from 0.25 to 1. The support vector machine trains 800 sets of data and verifies 200 sets of data. The average error of 5-fold cross validation is 2.8%. The pressure change trend is predicted 500ms in advance. When the load increases from 1000kg to 1500kg, the predicted pressure value rises from 12MPa to 15.8MPa within 300ms, and the actual pressure reaches 16MPa after 500ms, with a prediction error of 1.2%. The input of the fuzzy controller is divided into 7 fuzzy subsets, the pressure deviation is -3MPa to +3MPa, the deviation change rate is -0.1MPa / ms to +0.1MPa / ms, the output opening compensation value is -20% to +20%, and the opening compensation value is 15% obtained by using the center of gravity method to solve the fuzziness, and the actual pressure is reduced to 14.5MPa.

[0033] S106. Based on the load capacity of the forklift, the fork lowering speed, and the hydraulic cylinder pressure, a fuzzy rule base is established to realize fuzzy adaptive control of the buffer valve opening, and the buffer valve opening is automatically adjusted according to different working conditions to ensure that the hydraulic system pressure is stable within a safe range.

[0034] The forklift load, fork descent speed and hydraulic cylinder pressure signals collected by the sensor are obtained, the sampling frequency of the sensor is a preset frequency value, and the forklift load, fork descent speed and hydraulic cylinder pressure signals are quantized to form the first input data; according to the first input data, the load, descent speed and hydraulic cylinder pressure are fuzzy quantized by using a pre-established Gaussian membership function to obtain the membership values ​​of the load, descent speed and hydraulic cylinder pressure; for the membership values ​​of the load, descent speed and hydraulic cylinder pressure, a fuzzy rule base is constructed by using a preset weight coefficient, and the rule activation strength is calculated by using a minimum operator to obtain the activation degree of the fuzzy rule base; according to the activation degree of the fuzzy rule base, a maximum operator is used to perform rule synthesis operation to obtain the fuzzy control amount of the buffer valve opening, and the buffer valve opening is divided into a preset number of fuzzy subsets; for the fuzzy control amount of the buffer valve opening, a centroid method is used to perform defuzzification calculation, and the opening output value is smoothed by a sliding average filter to obtain the precise control amount of the buffer valve opening.

[0035] Specifically, the three input variables of forklift load, fork lowering speed and hydraulic cylinder pressure are collected by sensors with sampling frequency of 100Hz and accuracy of 0.1%, and the variable ranges are 0-2500kg, 0-400mm / s and 0-25MPa respectively. The load is divided into four fuzzy subsets: zero load, light load, medium load and heavy load, the lowering speed is divided into three fuzzy subsets: slow speed, medium speed and fast speed, and the hydraulic cylinder pressure is divided into three fuzzy subsets: low pressure, medium pressure and high pressure. The fuzzy variables are quantified using the Gaussian membership function with a standard deviation of 20% of the center value, the overlap of adjacent membership functions is set to 25%, the center values ​​of the load membership are set to 0kg, 500kg, 1000kg and 2000kg, the center values ​​of the lowering speed membership are set to 100mm / s, 200mm / s and 300mm / s, and the center values ​​of the pressure membership are set to 5MPa, 10MPa and 20MPa. A fuzzy rule base containing 36 rules is constructed based on the input variable weights of 0.5, 0.3, and 0.2. The rule antecedent is a combination of load, descent speed, and hydraulic cylinder pressure. The minimum operator is used to calculate the rule activation strength, and the maximum operator is used for rule synthesis. The rule consequence is the buffer valve opening value. The opening is divided into three fuzzy subsets: small opening, medium opening, and large opening, with center values ​​of 25%, 50%, and 75%, respectively. The centroid method is used to defuzzify the rule reasoning results. The three input variable values ​​under the current working conditions are collected in real time within a 20ms control cycle. The opening output is smoothed by a sliding average filter with a window length of 5, and the opening change rate is limited to 10% / s to obtain the precise control value of the buffer valve opening. The sensor data collected when the forklift is lifted without load shows: the load is 0kg, the descent speed is 300mm / s, and the pressure is 5MPa, corresponding to the zero-load subset membership of 0.95, the fast subset membership of 0.85, the low-pressure subset membership of 0.9, the rule activation strength of 0.85, and the output opening of 25%. When the forklift is loaded with 500kg, the descent speed drops to 250mm / s, the pressure rises to 8MPa, corresponding to the light-load subset membership of 0.8, the medium-speed subset membership of 0.75, the low-pressure subset membership of 0.6, the rule activation strength of 0.6, and the output opening of 35%. When the load increases to 1000kg, the descent speed drops to 200mm / s, the pressure rises to 12MPa, corresponding to the medium-load subset membership of 0.85, the medium-speed subset membership of 0.9, the medium-pressure subset membership of 0.8, the rule activation strength of 0.8, and the output opening of 55%. When the load reaches 2000kg, the descent speed drops to 100mm / s and the pressure rises to 20MPa, corresponding to the heavy load subset membership of 0.9, the slow speed subset membership of 0.85, the high pressure subset membership of 0.9, the rule activation intensity of 0.85, and the output opening of 75%.After sliding average filtering, the opening output curve is smooth. When the load increases from 1000kg to 2000kg, the opening gradually increases from 55% to 75%, and the change rate remains at 8% / s. The pressure rises steadily to 20MPa, meeting the safety control requirements.

[0036] S107, establishing a nonlinear mapping relationship between the load weight, the fork lowering speed and the buffer valve opening, training and optimizing the neural network model, realizing intelligent control of the buffer valve opening, and automatically adjusting the buffer valve opening according to the load weight and fork lowering speed collected in real time.

[0037] The load signal and the descent speed signal collected by the sensor are obtained, and 16-bit quantization processing is performed through a data acquisition card according to the load signal and the descent speed signal, and the quantization signal is filtered by a median filter to obtain training sample data; according to the training sample data, the load, the descent speed and the buffer valve opening value are normalized to obtain normalized data, and the normalized data are standardized by a z-score method to obtain standardized training data; the double hidden layer neural network is trained by the standardized training data, and the neural network weight is initialized by a xaver method. If the training error is less than a preset threshold and the test set accuracy exceeds a specified threshold, a trained neural network model is obtained; the load signal and the descent speed signal collected in real time are processed according to the trained neural network model, and the buffer valve opening value is output through the neural network model; the buffer valve opening value is smoothed by a sliding average filter, and if the smoothed opening value exceeds the standard deviation threshold, the abnormal value is eliminated, and the opening value is limited by an opening limiter to obtain a final opening output value.

[0038] Specifically, the sensor collects data under the conditions of load 0-2000kg interval 200kg, descent speed 0-300mm / s interval 30mm / s, and uses a high-precision data acquisition card to quantize the signal to 16 bits. The sampling frequency is set to 100Hz, and the median filter is used to remove noise from the sampled data to obtain 1000 sets of complete training sample data, and the abnormal data exceeding 3 times the standard deviation is eliminated. The collected training sample data is normalized, the load is normalized to the range of 0-1, the descent speed is normalized to the range of 0-1, and the buffer valve opening value is normalized to the range of 0-1. After randomly shuffling the data set 3 times, it is divided into training set and test set in a ratio of 8:2, and the mean and standard deviation of the training set are used to perform z-score normalization on the data. A feedforward neural network with two hidden layers was constructed, and the weights were initialized using the xaver method. The number of nodes in the first hidden layer was set to 8, and the number of nodes in the second hidden layer was set to 4. The hyperbolic tangent function was used as the activation function, and the learning rate was set to 0.01. The iteration was stopped when the training error was less than 0.001 within 1000 iterations. The training was verified on the test set every 100 times, and the accuracy reached more than 95%. The real-time collected load and descent speed data were processed by the trained neural network, and the output results were smoothed using a sliding average filter with a window length of 10. The outliers exceeding 3 times the standard deviation were eliminated, the opening output was limited to 0-100%, and the opening change rate was controlled within 10% / s. During the forklift data collection process, when the load is 0kg, the descent speed is 300mm / s, and the opening value is 25%; when the load is 400kg, the descent speed is 270mm / s, and the opening value is 35%; when the load is 800kg, the descent speed is 240mm / s, and the opening value is 45%; when the load is 1200kg, the descent speed is 210mm / s, and the opening value is 55%; when the load is 1600kg, the descent speed is 180mm / s, and the opening value is 65%; when the load is 2000kg, the descent speed is 150mm / s, and the opening value is 75%. After data normalization, the mean value of the load data is 0.5, the standard deviation is 0.2, the mean value of the descent speed data is 0.6, the standard deviation is 0.15, and the mean value of the opening value data is 0.4, and the standard deviation is 0.18. After 200 neural network trainings, the training error dropped to 0.0009, the test set verification accuracy reached 96.5%, the weights of the 8 nodes in the first hidden layer ranged from -0.8 to 0.8, and the weights of the 4 nodes in the second hidden layer ranged from -0.5 to 0.5. During real-time control, when the load increased from 800kg to 1200kg, the collected descent speed decreased from 240mm / s to 210mm / s. The opening value calculated by the neural network increased from 45% to 55%. After the sliding average filter, the opening change rate remained at 8% / s, and the pressure transitioned smoothly.

[0039] The above only lists some preferred embodiments of the present invention, but the present invention is not limited thereto, and many improvements and changes can be made. As long as the improvements and changes are made on the basis of the basic principles of the present invention, they should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for analyzing the load of an engineering vehicle, characterized in that: The method comprises: Obtain the load weight of the forklift under full load, identify the descending speed of the fork when it descends quickly, calculate the pressure change curve of the hydraulic cylinder under full load condition of the forklift through the established hydraulic system mathematical model, and judge whether the pressure exceeds the preset safety threshold. If it exceeds the threshold, the pressure regulation function of the hydraulic buffer valve is triggered; A pressure sensor is used to detect the inlet and outlet pressures of the hydraulic buffer valve in real time to obtain the pressure pulsation signal. The pressure pulsation signal is analyzed in the time-frequency domain to obtain the pressure pulsation amplitude at different frequencies, and it is judged whether the pressure pulsation exceeds the preset safety threshold. If it exceeds the threshold, the opening of the buffer valve is controlled to reduce the pressure pulsation. Obtain the opening area of ​​the buffer valve. For different buffer valve opening areas, analyze the dynamic response characteristics of the hydraulic system through simulation, obtain the change curves of system pressure, flow, and valve opening, thereby optimizing the buffer valve opening control strategy and determining the target buffer valve opening area to obtain the target system dynamic response performance; Under different load conditions, the hydraulic cylinder pressure and buffer valve opening data are collected, and a polynomial regression model of load and buffer valve pressure is established through least squares fitting to obtain the correlation between load and target buffer pressure. According to the correlation between load and target buffer pressure, adaptive adjustment of buffer valve pressure is achieved. According to the fork lowering speed and load, the change trend of the hydraulic cylinder pressure is predicted, and the pressure prediction model is established through the support vector machine algorithm. When the predicted pressure exceeds the safety threshold, the buffer valve opening is controlled in advance to achieve predictive control of the pressure; Combined with the forklift's load, fork lowering speed, and hydraulic cylinder pressure, a fuzzy rule base is established to achieve fuzzy adaptive control of the buffer valve opening. The buffer valve opening is automatically adjusted according to different working conditions to ensure that the hydraulic system pressure is stable within a safe range, including: Acquire the forklift load, fork lowering speed and hydraulic cylinder pressure signals collected by the sensor, wherein the sampling frequency of the sensor is a preset frequency value, and the forklift load, fork lowering speed and hydraulic cylinder pressure signals are quantized to form first input data; According to the first input data, a pre-established Gaussian membership function is used to perform fuzzy quantization processing on the load, the descent speed and the hydraulic cylinder pressure to obtain membership values ​​of the load, the descent speed and the hydraulic cylinder pressure; According to the membership values ​​of the load, the lowering speed and the hydraulic cylinder pressure, a fuzzy rule base is constructed by using preset weight coefficients, and the activation strength of the rule is calculated by a minimum operator to obtain the activation degree of the fuzzy rule base; According to the activation degree of the fuzzy rule base, a maximum operator is used to perform rule synthesis operation to obtain a fuzzy control amount of the buffer valve opening, and the buffer valve opening is divided into a preset number of fuzzy subsets; For the fuzzy control amount of the buffer valve opening, the centroid method is used to perform defuzzification calculation, and the opening output value is smoothed by a sliding average filter to obtain the precise control amount of the buffer valve opening; A nonlinear mapping relationship between load weight, fork descent speed and buffer valve opening is established, and the neural network model is trained and optimized to realize intelligent control of buffer valve opening. The buffer valve opening is automatically adjusted according to the load weight and fork descent speed collected in real time.

2. The method according to claim 1, characterized in that The method of obtaining the load of the forklift under the fully loaded state, identifying the descending speed of the fork when the fork is rapidly descended, calculating the pressure change curve of the hydraulic cylinder under the fully loaded working condition of the forklift through the established hydraulic system mathematical model, and judging whether the pressure exceeds the preset safety threshold, and triggering the pressure adjustment function of the hydraulic buffer valve if the pressure exceeds the threshold, includes: Obtain measurement data of the gravity sensor and the lifting angle sensor, and obtain the fork load moment value according to the measurement data and the fork arm length according to the load moment calculation formula M=F×L×cosθ, where F is the gravity value of the fork, L is the fork arm length, and θ is the fork lifting angle; According to the fork load moment value, a Kalman filter is used to perform noise filtering processing, and the fork position change data is collected by a displacement sensor within a fixed sampling time interval, and the hydraulic oil flow value is obtained according to the flow calculation formula Q=v×A, where v is the fork descending speed and A is the piston area; According to the hydraulic oil flow value according to the Bernoulli equation ΔP=ρv 2 / 2 Calculate the pipeline pressure loss value, where ρ is the hydraulic oil density. The hydraulic oil pressure data is collected through the oil pump pressure sensor, and the hydraulic cylinder pressure value is obtained according to the pressure calculation formula P=F / A+ΔP; Using cubic spline interpolation to perform curve fitting processing on the hydraulic cylinder pressure value to obtain a hydraulic cylinder pressure change curve, and determine whether the hydraulic cylinder pressure value exceeds a preset pressure threshold; If the hydraulic cylinder pressure value exceeds the preset pressure threshold, the hydraulic buffer valve is triggered, and a proportional-integral controller is used to output a control signal according to the control formula u=Kp×e+Ki×∫edt, where e is the deviation between the actual pressure and the threshold, Kp is the proportional coefficient, and Ki is the integral coefficient. The buffer valve opening is adjusted according to the control signal.

3. The method according to claim 1, characterized in that The method uses a pressure sensor to detect the inlet and outlet pressures of the hydraulic buffer valve in real time, obtains a pressure pulsation signal, performs time-frequency domain analysis on the pressure pulsation signal, obtains pressure pulsation amplitudes at different frequencies, and determines whether the pressure pulsation exceeds a preset safety threshold. If the pressure pulsation exceeds the threshold, the opening of the buffer valve is controlled to reduce the pressure pulsation, including: A pressure sensor is used to collect the inlet and outlet pressure signals of the hydraulic buffer valve, and the pressure signal is sampled and quantified by a data acquisition card to obtain a pressure pulsation sampling data sequence; According to the pressure pulsation sampling data sequence, a bandpass filter is used to remove power frequency interference and high-frequency noise, and a filtered pressure signal is obtained through wavelet decomposition processing; Performing a fast Fourier transform on the filtered pressure signal to obtain a pressure pulsation amplitude spectrum within a frequency range; Extracting frequency components with amplitudes greater than twice the mean value from the pressure pulsation amplitude spectrum as main frequency components, and calculating a linear mapping relationship of the opening adjustment amount according to the main frequency components; If the amplitude of the main frequency component exceeds the preset safety threshold, the optimal opening value is calculated using the recursive least squares method, and the buffer valve opening is compensated and adjusted according to the opening adjustment rate.

4. The method according to claim 1, characterized in that: The method of obtaining the buffer valve opening area, analyzing the dynamic response characteristics of the hydraulic system through simulation for different buffer valve opening areas, obtaining the change curves of system pressure, flow, and valve opening, thereby optimizing the buffer valve opening control strategy, determining the target buffer valve opening area, and obtaining the target system dynamic response performance, includes: A displacement sensor with a measuring range is used to collect a valve core displacement signal, and a Canny edge detection algorithm is used to identify the valve port contour line according to the displacement signal to obtain a valve port opening area value; According to the valve opening area value, under the pressure inlet boundary and pressure outlet boundary conditions, the flow channel between the hydraulic cylinder and the buffer valve is calculated by a grid division method to obtain the pressure distribution field and flow field data; According to the pressure distribution field and the flow field data, a time-domain finite difference algorithm is used to calculate the response characteristics to obtain the pressure rise time, flow response delay, pressure overshoot and flow fluctuation value; The pressure rise time, the flow response delay, the pressure overshoot and the flow fluctuation value are calculated by using a cubic polynomial least squares fitting algorithm to obtain a correlation curve between the opening and the dynamic characteristics; If the opening degree and dynamic characteristic correlation curve meets the pressure response time weight, overshoot weight and flow delay weight requirements, a genetic algorithm is used to optimize the opening degree response curve to obtain the optimal opening area value.

5. The method according to claim 1, characterized in that: Under different load conditions, the hydraulic cylinder pressure and buffer valve opening data are collected, and a polynomial regression model of the load and buffer valve pressure is established by least squares fitting to obtain the correlation between the load and the target buffer pressure. According to the correlation between the load and the target buffer pressure, adaptive adjustment of the buffer valve pressure is achieved, including: The load signal of the hydraulic forklift is collected by the load sensor, the load signal is sampled by a data acquisition card for 16-bit quantization to obtain load data, and the load data is processed by a median filter; According to the load data, a pressure sensor is used to collect the hydraulic cylinder circuit pressure and the buffer valve inlet and outlet pressures, and the pressure sensor obtains a corresponding data set of the buffer valve opening value and the pressure data value; For the corresponding data group, a third-order polynomial least square method is used to perform curve fitting, and the curve fitting obtains polynomial coefficients; According to the polynomial coefficients, the curve is smoothed by using a Lagrange interpolation method, wherein the Lagrange interpolation method obtains a smooth curve whose pressure smoothness error is less than a preset threshold value; For the smooth curve, a mean filter is used to calculate a target buffer pressure value, and the target buffer pressure value is adjusted by a proportional-integral controller for closed-loop opening, and the proportional-integral controller performs adaptive control on the buffer valve opening.

6. The method according to claim 1, characterized in that The method predicts the change trend of the hydraulic cylinder pressure according to the fork lowering speed and the load, establishes a pressure prediction model through a support vector machine algorithm, and controls the opening of the buffer valve in advance when the predicted pressure exceeds the safety threshold to achieve predictive control of the pressure, including: Collecting a hydraulic cylinder pressure signal through a pressure sensor and a fork displacement signal through a displacement sensor, and performing data quantization according to the pressure signal and the displacement signal to obtain a quantized data sequence; A pressure change process sample is selected for the quantitative data sequence, and a maximum and minimum value method is used to normalize the pressure change process sample to obtain a normalized data set, wherein the normalized data set includes a load parameter, a descent speed parameter, and a pressure change parameter; Training a support vector machine prediction model according to the normalized data set, wherein the support vector machine prediction model adopts a radial basis kernel function and obtains support vector parameters through cross validation; The support vector machine prediction model is used to predict the pressure change trend, and the prediction result is smoothed by a sliding average filter to obtain a smoothed predicted pressure value; If the smoothed predicted pressure value exceeds the safety threshold, a fuzzy controller is used to perform opening compensation, and the opening compensation value is obtained by defuzzification calculation using the center of gravity method according to the pressure deviation and the deviation change rate, and the buffer valve opening is adjusted.

7. The method according to claim 1, characterized in that The method of establishing a nonlinear mapping relationship between the load weight, the fork lowering speed and the buffer valve opening, training and optimizing the neural network model, realizing intelligent control of the buffer valve opening, and automatically adjusting the buffer valve opening according to the load weight and fork lowering speed collected in real time, includes: Obtain the load signal and the descent speed signal collected by the sensor, perform 16-bit quantization processing on the load signal and the descent speed signal through the data acquisition card, and use the median filter to filter the quantized signal to obtain training sample data; According to the training sample data, the load weight, the descent speed and the buffer valve opening value are normalized to obtain normalized data, and the normalized data are normalized by using a z-score method to obtain standardized training data; The double hidden layer neural network is trained using the standardized training data, and the weights of the neural network are initialized using the Xavier method. If the training error is less than a preset threshold and the accuracy of the test set exceeds a specified threshold, a trained neural network model is obtained; Processing the real-time collected load signal and descent speed signal according to the trained neural network model, and outputting the buffer valve opening value through the neural network model; The buffer valve opening value is smoothed by a sliding average filter. If the smoothed opening value exceeds the standard deviation threshold, abnormal values ​​are eliminated, and the opening value is limited by an opening limiter to obtain a final opening output value.

Images