Bearing load real-time monitoring and hydraulic pressure self-adaptive adjusting system and method

Abstract

The invention provides a bearing load real-time monitoring and hydraulic pressure self-adaptive adjusting system and method, and the system comprises a signal sensing unit which is used for collecting a bearing vibration signal, a hydraulic system pressure signal and a main shaft rotating speed signal; and the signal processing and calculating unit is connected with the signal sensing unit and is used for processing the collected signals, estimating the bearing load based on a pre-trained soft measurement model and generating a hydraulic control instruction. The dependence of traditional bearing load monitoring on an expensive force sensor is broken through, and by integrating operation signals such as vibration, hydraulic pressure and rotating speed which are easy to obtain, the bearing load monitoring precision is improved. According to the method, the incidence relation between signals and loads is established through off-line modeling, non-intrusive load real-time estimation is achieved, meanwhile, feed-forward and feedback composite control is adopted, hydraulic pressure is adjusted in a self-adaptive mode according to the estimated loads, traditional empirical fixed setting is replaced, monitoring, adjusting, early warning and maintenance prompting functions are integrated, and real-time monitoring is achieved. The technical problems that a traditional scheme is high in cost, blind in adjustment and dispersed in function are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of mechanical equipment condition monitoring technology, specifically to a bearing load real-time monitoring and hydraulic pressure adaptive adjustment system and method. Background Technology

[0002] In large, heavy rotating machinery, bearings, as core components that transmit loads and support rotating parts, directly determine the operational safety, processing / operational accuracy, and service life of the equipment under dynamic loads. For example, if the load on the work roll bearings of a large rolling mill exceeds its limit, it may lead to bearing seizure, roll system breakage, and production shutdown; excessive load fluctuations in the main shaft bearings of wind turbines will exacerbate the wear of blades and gearboxes, reducing the unit's power generation efficiency.

[0003] Currently, real-time monitoring of bearing loads in the industry mainly relies on direct measurement methods, which involve permanently installing strain gauge force sensors, piezoelectric force sensors, etc., on the bearing housing or supporting structure. This method has three major drawbacks: 1. High cost – High-precision force sensors typically cost tens of thousands of yuan each and require customized installation structures, making large-scale applications uneconomical; 2. Complex installation – Machining of the equipment body (such as drilling and grooving) is required, which may affect structural strength and is difficult to arrange in confined spaces; 3. Low reliability – Force sensors directly bear the vibration and impact of the equipment, making them prone to failure under complex conditions such as high temperatures, oil contamination, and dust, resulting in high maintenance costs.

[0004] Meanwhile, the hydraulic support systems of rotating machinery (such as the hydraulic pressing system of rolling mills and the yaw hydraulic support system of wind turbines), as the core actuators for load regulation, generally adopt an empirical static mode for pressure setting: a fixed pressure value is preset according to the rated operating conditions of the equipment, or segmented switching is performed according to a few types of operating conditions. This mode cannot respond to dynamic changes in bearing load, leading to two problems: first, "overload risk"—when the actual load suddenly increases, the fixed hydraulic pressure cannot provide sufficient support stiffness, which can easily cause bearing impact damage; second, "energy waste"—when the actual load is low, excessively high hydraulic pressure will increase the energy consumption of the pump station, while also aggravating the wear of hydraulic components and equipment vibration.

[0005] In existing technologies, some solutions attempt to infer the load from a single vibration signal, but fail to combine key operating parameters such as hydraulic pressure and rotational speed, resulting in low estimation accuracy (average relative error is usually >15%). Other solutions only perform closed-loop control on hydraulic pressure, but lack a direct correlation with the actual load on the bearing, and the adjustment target is detached from the core stress state of the equipment. In summary, the industry urgently needs an economical, reliable, non-invasive, multi-parameter correlated bearing load sensing method, as well as a hydraulic intelligent adjustment strategy linked to it, to solve the dual pain points of "difficult measurement" and "blind adjustment". Summary of the Invention

[0006] Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a real-time bearing load monitoring and hydraulic pressure adaptive adjustment system and method, which solves the dual problems of measurement difficulty and adjustment blindness in existing technologies.

[0007] Technical solution

[0008] To achieve the above objectives, the present invention provides the following technical solution: a bearing load real-time monitoring and hydraulic pressure adaptive adjustment system, comprising: The signal sensing unit is used to collect bearing vibration signals, hydraulic system pressure signals, and spindle speed signals. The signal processing and computing unit, connected to the signal sensing unit, is used to process the acquired signals, estimate the bearing load based on the pre-trained soft measurement model, and generate hydraulic control commands. A hydraulic actuator, connected to the signal processing and computing unit, is used to receive control commands and adjust the hydraulic system pressure. The data storage and interaction unit is connected to the signal processing and computing unit and is used for data storage, human-computer interaction, and communication with external systems.

[0009] Preferably, the signal sensing unit includes: A vibration acceleration sensor is installed on the vibration-sensitive part of the bearing housing, with a frequency response range of 5Hz-10kHz. The pressure sensor is installed on the hydraulic system's actuator oil circuit, with a range of 0-40MPa. The speed sensor is mounted coaxially with the spindle and has a resolution of no less than 1000 pulses / revolution.

[0010] Preferably, the signal processing and computing unit includes: The data acquisition card supports multi-channel synchronous acquisition, with a single-channel sampling rate of no less than 10kHz. The core controller uses an industrial computer or programmable logic controller and is equipped with a load soft measurement model and hydraulic regulation control algorithm.

[0011] Preferably, the hydraulic actuator includes: The electro-hydraulic control element is an electro-hydraulic proportional valve or servo valve, with a control accuracy of no more than 1% FS and a response time of no more than 50ms; The drive module is used to convert control signals into valve-controlled current signals; Hydraulic auxiliary components, including pump stations, oil tanks, and filters.

[0012] A method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure includes the following steps: Offline modeling stage: S1. Temporarily install a reference force sensor to collect the actual load value, vibration signal, hydraulic pressure value and speed signal under typical working conditions; S2. After denoising the vibration signal, perform spectrum analysis, screen out target frequency bands with a correlation ≥ 0.85 with the real load, and calculate the root mean square value of the vibration signal in the frequency band as the vibration characteristic value. S3, vibration characteristic value, hydraulic pressure value, and rotational speed are the inputs, and the actual load value is the output. The load soft measurement model is trained using machine learning algorithm, and the model parameters are stored after verification. Online application stage: S4. Real-time acquisition of vibration, pressure, and rotation speed signals; repeat the feature extraction process in S2 to obtain real-time vibration feature values. S5. Substitute the real-time characteristic values ​​and signals into the load soft measurement model to estimate the real-time bearing load; S6. Based on the deviation between the expected load and the estimated load, a hydraulic control command is generated through a feedforward + feedback PID control algorithm to drive the hydraulic actuator to adjust the pressure. S7. Monitor and estimate loads to provide over-limit early warning and trend maintenance prompts.

[0013] Preferably, the offline modeling stage includes: Plan the operating conditions of the speed and pressure gradients covering the operating range of the equipment; Collect vibration, pressure, rotational speed, and actual load signals; The vibration signal is denoised and its features are extracted to select frequency bands that are highly correlated with the load. A soft measurement model is established using machine learning algorithms, with vibration characteristics, pressure, and rotational speed as inputs and actual load as output; After verifying the model's accuracy and ensuring the average relative error does not exceed 5%, store the model parameters.

[0014] Preferably, the online application phase includes: The root mean square value of the vibration signal in the target frequency band is calculated in real time and used as the feature input. Substitute the values ​​into the soft measurement model to obtain the real-time load estimate; Based on the deviation between the expected load and the estimated load, a hydraulic control signal is generated by combining the feedforward preload command and PID feedback correction.

[0015] Beneficial effects

[0016] This invention provides a system and method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure. It has the following beneficial effects: This invention breaks through the reliance on expensive force sensors in traditional bearing load monitoring. By integrating easily obtainable operating signals such as vibration, hydraulic pressure, and rotational speed, and establishing the correlation between signals and loads through offline modeling, it achieves non-intrusive real-time load estimation. At the same time, it adopts a composite control of feedforward and feedback, which adaptively adjusts the hydraulic pressure according to the estimated load, replacing the traditional experience-based fixed settings. It also integrates monitoring, adjustment, early warning, and maintenance prompt functions, effectively solving the technical pain points of high cost, blind adjustment, and fragmented functions of traditional solutions.

[0017] This invention eliminates the purchase and maintenance costs of high-precision force sensors, significantly reducing equipment modification costs; it does not require large-scale modifications to the equipment itself, is compatible with various rotating machinery, and has strong compatibility. Its dynamic adjustment capability can avoid overload damage and underload energy waste. Combined with load trend analysis, it enables predictive maintenance, reduces unplanned downtime, and comprehensively improves the safety, stability, and economy of equipment operation. Attached Figure Description

[0018] Figure 1 This is a flowchart of a bearing load real-time monitoring and hydraulic pressure adaptive adjustment system and method proposed in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example

[0021] like Figure 1 As shown, this embodiment of the invention provides a bearing load real-time monitoring and hydraulic pressure adaptive adjustment system, comprising: The signal sensing unit is used to collect bearing vibration signals, hydraulic system pressure signals, and spindle speed signals. (1) Vibration acceleration sensor: a piezoelectric sensor (model: PCB 352C33) is selected, with a frequency response of 5Hz-10kHz and a measurement range of ±50g. It is installed on the radial end cover of the work roller bearing seat (close to the rolling contact point) and is installed by magnetic adsorption (for easy disassembly and calibration). (2) Pressure sensor: A strain gauge pressure sensor (model: HBM P3MB) with a range of 0-40MPa and an accuracy of 0.25%FS is selected and installed in the oil cylinder inlet pipeline of the hydraulic pressing system of the rolling mill. (3) Speed ​​sensor: A photoelectric encoder (model: Pepperl+Fuchs RVI58N) with a resolution of 2048 pulses / revolution is selected and installed coaxially with the main shaft of the work roller and connected by a coupling; The signal processing and computing unit, connected to the signal sensing unit, is used to process the acquired signals, estimate the bearing load based on the pre-trained soft measurement model, and generate hydraulic control commands. (1) Data acquisition card: NI9234 (4 analog inputs, sampling rate 50kHz / channel) + NI9401 (8 digital inputs) are selected and integrated in NIcDAQ-9178 chassis; (2) Core controller: An industrial computer (IPC, configured with Core i7-10700, 16GB memory, 512GB SSD) was selected and LabVIEW software was installed for data processing and algorithm deployment; The hydraulic actuator, connected to the signal processing and computing unit, is used to receive control commands and regulate the hydraulic system pressure; (1) Electro-hydraulic control components: an electro-hydraulic proportional relief valve (model: Atos AGMZO-A-10 / 210) was selected, with a control accuracy of 0.5%FS and a response time of 30ms; (2) Drive module: Atos E-ME-K-30-10 proportional amplifier is selected; (3) Auxiliary components: hydraulic pump station (22kW motor + axial piston pump, flow rate 100L / min), 1000L oil tank, 10μm return oil filter; The data storage and interaction unit is connected to the signal processing and computing unit and is used for data storage, human-computer interaction, and communication with external systems.

[0022] Storage module: Built-in 512GB SSD + external 2TB hard drive; Human-computer interaction module: 15-inch touch screen (model: Weintek MT8150iE); Communication module: Supports Profinet protocol for communication with the rolling mill central control system (Siemens S7-1500 PLC). The signal sensing unit includes: A vibration acceleration sensor is installed on the vibration-sensitive part of the bearing housing, with a frequency response range of 5Hz-10kHz. The pressure sensor is installed on the hydraulic system's actuator oil circuit, with a range of 0-40MPa. The speed sensor is mounted coaxially with the spindle and has a resolution of no less than 1000 pulses / revolution.

[0023] The signal processing and computing unit includes: The data acquisition card supports multi-channel synchronous acquisition, with a single-channel sampling rate of no less than 10kHz. The core controller uses an industrial computer or programmable logic controller and is equipped with a load soft measurement model and hydraulic regulation control algorithm.

[0024] The hydraulic actuator includes: The electro-hydraulic control element is an electro-hydraulic proportional valve or servo valve, with a control accuracy of no more than 1% FS and a response time of no more than 50ms; The drive module is used to convert control signals into valve-controlled current signals; Hydraulic auxiliary components, including pump stations, oil tanks, and filters.

[0025] A method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure includes the following steps: Offline modeling stage: S1. Temporarily install a reference force sensor to collect the actual load value, vibration signal, hydraulic pressure value and speed signal under typical working conditions; S2. After denoising the vibration signal, perform spectrum analysis, screen out target frequency bands with a correlation ≥ 0.85 with the real load, and calculate the root mean square value of the vibration signal in the frequency band as the vibration characteristic value. S3, vibration characteristic value, hydraulic pressure value, and rotational speed are the inputs, and the actual load value is the output. The load soft measurement model is trained using machine learning algorithm, and the model parameters are stored after verification. Online application stage: S4. Real-time acquisition of vibration, pressure, and rotation speed signals; repeat the feature extraction process in S2 to obtain real-time vibration feature values. S5. Substitute the real-time characteristic values ​​and signals into the load soft measurement model to estimate the real-time bearing load; S6. Based on the deviation between the expected load and the estimated load, a hydraulic control command is generated through a feedforward + feedback PID control algorithm to drive the hydraulic actuator to adjust the pressure. S7. Monitor and estimate loads to provide over-limit early warning and trend maintenance prompts.

[0026] The offline modeling phase includes: Plan the operating conditions of the speed and pressure gradients covering the operating range of the equipment; Collect vibration, pressure, rotational speed, and actual load signals; The vibration signal is denoised and its features are extracted to select frequency bands that are highly correlated with the load. A soft measurement model is established using machine learning algorithms, with vibration characteristics, pressure, and rotational speed as inputs and actual load as output; After verifying the model's accuracy and ensuring the average relative error does not exceed 5%, store the model parameters.

[0027] The online application phase includes: The root mean square value of the vibration signal in the target frequency band is calculated in real time and used as the feature input. Substitute the values ​​into the soft measurement model to obtain the real-time load estimate; Based on the deviation between the expected load and the estimated load, a hydraulic control signal is generated by combining the feedforward preload command and PID feedback correction.

[0028] (1) Offline modeling implementation Reference sensor deployment: A high-precision tensile and compressive force sensor (model: HBMU9B, range 0-500kN, accuracy 0.1% FS) is temporarily installed at the contact point between the work roll bearing housing and the frame, and the output signal is connected to the NI9234 data acquisition card.

[0029] Operating condition planning and data acquisition: Speed ​​gradient: Set the working roller speed to 100rpm, 200rpm, 300rpm, 400rpm, and 500rpm; Load gradient: By adjusting the hydraulic pressure to 8MPa, 12MPa, 16MPa, 20MPa, and 24MPa, the corresponding bearing loads are 100kN, 150kN, 200kN, 250kN, and 300kN. Each operating point was run stably for 60 seconds, and data was collected at a frequency of 10kHz, resulting in a total of 25 sets of calibration data.

[0030] Feature extraction and model building: Wavelet denoising: db4 wavelet is used, with 5 decomposition layers, and soft threshold denoising is applied; Frequency band selection: FFT analysis revealed that the vibration RMS in the 150Hz-1800Hz frequency band was correlated with the actual load by 0.91, thus identifying this frequency band as the target frequency band. Model training: The soft measurement model was trained using the SVR algorithm (kernel function RBF, penalty coefficient C=10, gamma=0.1).

[0031] Model validation: Seven sets of validation data were selected, and the MRE was calculated to be 3.2% < 5%, indicating that the model was qualified. After storing the model parameters, the HBM U9B sensor was removed.

[0032] (2) Implementation of online applications Real-time signal acquisition: The IPC synchronously acquires vibration, pressure, and rotational speed signals at a frequency of 10kHz via the NI acquisition card.

[0033] Load estimation: Real-time calculation of RMS vibration in the 150Hz-800Hz frequency band, substituted into the SVR model, yields F. est_real The estimated cycle time is 50ms.

[0034] Hydraulic adjustment Feedforward instruction: When rolling Q235 steel (20mm thickness), match F from the mapping table. tar9et =220kN, pre-calculated P pre =18MPa; PID control: Kp=0.8, Ki=0.05, Kd=0.02, when ΔF=10kN, ΔP=0.5MPa, P out =18.5MPa; Execution: The AGMZO valve is controlled via a proportional amplifier to adjust the hydraulic pressure to 18.5 MPa, with a response time of 40 ms.

[0035] Security Alert Over-limit warning: F max =350kN, when F est_real =320kN (0.91×F) max When this occurs, the touchscreen will trigger an audible and visual alarm. Maintenance prompt: The average load over 72 consecutive hours increased from 220kN to 235kN (an increase of 6.8%), prompting the message "Check work roll bearing wear".

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A bearing load real-time monitoring and hydraulic pressure adaptive adjustment system, characterized in that, include: The signal sensing unit is used to collect bearing vibration signals, hydraulic system pressure signals, and spindle speed signals. The signal processing and computing unit, connected to the signal sensing unit, is used to process the acquired signals, estimate the bearing load based on the pre-trained soft measurement model, and generate hydraulic control commands. A hydraulic actuator, connected to the signal processing and computing unit, is used to receive control commands and adjust the hydraulic system pressure. The data storage and interaction unit is connected to the signal processing and computing unit and is used for data storage, human-computer interaction, and communication with external systems.

2. The bearing load real-time monitoring and hydraulic pressure adaptive adjustment system according to claim 1, characterized in that: The signal sensing unit includes: A vibration acceleration sensor is installed on the vibration-sensitive part of the bearing housing, with a frequency response range of 5Hz-10kHz. The pressure sensor is installed on the hydraulic system's actuator oil circuit, with a range of 0-40MPa. The speed sensor is mounted coaxially with the spindle and has a resolution of no less than 1000 pulses / revolution.

3. The bearing load real-time monitoring and hydraulic pressure adaptive adjustment system according to claim 1, characterized in that: The signal processing and computing unit includes: The data acquisition card supports multi-channel synchronous acquisition, with a single-channel sampling rate of no less than 10kHz. The core controller uses an industrial computer or programmable logic controller and is equipped with a load soft measurement model and hydraulic regulation control algorithm.

4. The bearing load real-time monitoring and hydraulic pressure adaptive adjustment system according to claim 1, characterized in that: The hydraulic actuator includes: The electro-hydraulic control element is an electro-hydraulic proportional valve or servo valve, with a control accuracy of no more than 1% FS and a response time of no more than 50ms; The drive module is used to convert control signals into valve-controlled current signals; Hydraulic auxiliary components, including pump stations, oil tanks, and filters.

5. A method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure, applied to the system described in any one of claims 1-4, characterized in that, Includes the following steps: Offline modeling stage: S1. Temporarily install a reference force sensor to collect the actual load value, vibration signal, hydraulic pressure value and speed signal under typical working conditions; S2. After denoising the vibration signal, perform spectrum analysis, screen out target frequency bands with a correlation ≥ 0.85 with the real load, and calculate the root mean square value of the vibration signal in the frequency band as the vibration characteristic value. S3, vibration characteristic value, hydraulic pressure value, and rotational speed are the inputs, and the actual load value is the output. The load soft measurement model is trained using machine learning algorithm, and the model parameters are stored after verification. Online application stage: S4. Real-time acquisition of vibration, pressure, and rotation speed signals; repeat the feature extraction process in S2 to obtain real-time vibration feature values. S5. Substitute the real-time characteristic values ​​and signals into the load soft measurement model to estimate the real-time bearing load; S6. Based on the deviation between the expected load and the estimated load, a hydraulic control command is generated through a feedforward + feedback PID control algorithm to drive the hydraulic actuator to adjust the pressure. S7. Monitor and estimate loads to provide over-limit early warning and trend maintenance prompts.

6. The method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure according to claim 5, characterized in that: The offline modeling phase includes: Plan the operating conditions of the speed and pressure gradients covering the operating range of the equipment; Collect vibration, pressure, rotational speed, and actual load signals; The vibration signal is denoised and its features are extracted to select frequency bands that are highly correlated with the load. A soft measurement model is established using machine learning algorithms, with vibration characteristics, pressure, and rotational speed as inputs and actual load as output; After verifying the model's accuracy and ensuring the average relative error does not exceed 5%, store the model parameters.

7. The method for real-time monitoring of bearing load and adaptive adjustment of hydraulic pressure according to claim 5, characterized in that: The online application phase includes: The root mean square value of the vibration signal in the target frequency band is calculated in real time and used as the feature input. Substitute the values ​​into the soft measurement model to obtain the real-time load estimate; Based on the deviation between the expected load and the estimated load, a hydraulic control signal is generated by combining the feedforward preload command and PID feedback correction.

Images