Integrated online measurement system for thermal physical property parameters of cutting fluid

Abstract

The invention relates to the technical field of cutting fluid measurement, and discloses an integrated online measurement system for thermal physical property parameters of cutting fluid. According to the system, a multi-mode sensing array is arranged in a cutting machining area, and temperature field distribution data, flow state data and pressure gradient data of cutting fluid are collected; building a cutting fluid thermophysical property coupling model based on the acquired data, and calculating the dynamic thermal conductivity, the specific heat capacity characteristic and the viscosity coefficient of the cutting fluid through the model; and generating a reference measurement result of the thermal physical property parameters of the cutting fluid according to the dynamic thermal conductivity, the specific heat capacity characteristic and the viscosity coefficient obtained by calculation. According to the system, comprehensive collection of multi-dimensional state information of the cutting fluid can be achieved, the limitation that traditional off-line measurement is large in deviation and existing on-line measurement is single and dispersed is broken through, the authenticity and comprehensiveness of parameter measurement are improved, accurate parameter reference is provided for real-time regulation and control in the cutting machining process, and the requirement of modern machining for cutting fluid parameter monitoring is met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cutting fluid measurement, in particular to a cutting fluid thermophysical property parameter integrated online measurement system. BACKGROUND

[0002] In the modern mechanical processing field, cutting fluid is an important auxiliary medium to ensure processing precision, prolong tool life and improve processing efficiency. The accurate control of the thermophysical property parameters of the cutting fluid has a direct impact on the stability of the processing process and the quality of the final product. The thermal conductivity, specific heat capacity and viscosity coefficient of the cutting fluid will change dynamically with the change of temperature, flow state and pressure fluctuation in the processing process. The real-time changes of these parameters are closely related to the heat balance maintenance of the cutting area, the chip discharge efficiency and the processing surface quality.

[0003] The industry mostly adopts offline measurement mode for the measurement of the thermophysical property parameters of cutting fluid. This kind of offline measurement method usually needs to extract the cutting fluid from the processing system and detect the parameters in the laboratory environment using professional instruments. However, this offline measurement method has obvious limitations. On the one hand, the temperature, pressure and other environmental conditions of the cutting fluid in the process of extraction, transportation and laboratory measurement are quite different from the actual processing scene, which makes the measured parameters unable to truly reflect the dynamic characteristics in the processing process, and further makes the measurement results deviate from the actual application requirements, making it difficult to provide effective reference for real-time regulation of the processing process. On the other hand, offline measurement cannot realize continuous monitoring of the parameters, and can only obtain parameter data at a certain time point, which cannot capture the dynamic change rule of the thermophysical property parameters of the cutting fluid in the processing process. When the processing conditions change and the cutting fluid parameters appear abnormal, it is difficult to find out and take corresponding adjustment measures in time, which may affect the processing quality, and even cause tool damage, processing scrap and other problems.

[0004] The existing part of the online measurement technology often only measures a single thermophysical property parameter, and lacks the integrated measurement capability of multiple parameters. For example, some online measurement devices can only measure the temperature of the cutting fluid, and cannot simultaneously obtain the thermal conductivity, viscosity coefficient and other key parameters thereof; another device can measure part of the parameters, but the measurement of each parameter is independent of each other, and the correlation between the parameters is not established, so it is difficult to comprehensively reflect the overall thermophysical properties of the cutting fluid. Such single or scattered measurement method makes it difficult for the processing personnel to fully grasp the state of the cutting fluid in the actual processing process, and it is difficult to optimize and control the processing process as a whole, and it is difficult to meet the comprehensive demand of modern high-precision and high-efficiency mechanical processing for cutting fluid parameter monitoring. In the data acquisition link of the existing online measurement system, the multi-dimensional state information of the cutting fluid in the cutting processing area is not comprehensive enough, and most of the single state data of a local area is collected, and the temperature field distribution, flow state and pressure gradient of the cutting fluid in the whole processing area cannot be obtained, which leads to the lack of sufficient data source support for the subsequent parameter calculation model, and the precision of the calculated thermophysical property parameters is low, which further affects the precise control effect of the processing process. SUMMARY

[0005] The purpose of the present application is to provide a cutting fluid thermophysical property parameter integrated online measurement system to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purpose, the present application provides a cutting fluid thermophysical property parameter integrated online measurement system, which comprises: A multi-modal sensing array is arranged in the cutting processing area to collect temperature field distribution data, flow state data and pressure gradient data of the cutting fluid; A cutting fluid thermophysical property coupling model is constructed based on the temperature field distribution data, flow state data and pressure gradient data, and the dynamic thermal conductivity, specific heat capacity and viscosity coefficient of the cutting fluid are calculated through the cutting fluid thermophysical property coupling model; The reference measurement results of the cutting fluid thermophysical property parameters are generated according to the dynamic thermal conductivity, specific heat capacity and viscosity coefficient.

[0007] Preferably, the multi-modal sensing array arranged in the cutting processing area comprises: Distributed temperature sensors, ultrasonic flow rate detectors and micro pressure difference sensors are arranged along the flow path of the cutting fluid to obtain the local temperature gradient, instantaneous flow rate distribution and regional pressure difference change of the cutting fluid; A three-dimensional state characterization matrix of the cutting fluid is established based on the local temperature gradient, instantaneous flow rate distribution and regional pressure difference change, and the temperature field distribution data, flow state data and pressure gradient data are extracted through the three-dimensional state characterization matrix.

[0008] Preferably, the step of constructing the cutting fluid thermophysical property coupling model based on the temperature field distribution data, flow state data, and pressure gradient data includes: Establish the multi-physics coupling relationship between the thermal conduction and flow characteristics of the cutting fluid, and initialize the core parameter set of the cutting fluid thermal property coupling model; The temperature field distribution data, flow state data, and pressure gradient data are input into the cutting fluid thermo-physical property coupling model, and the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient are solved by the thermo-fluid coupling equation.

[0009] Preferably, the calculation of the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of the cutting fluid using the cutting fluid thermophysical property coupling model includes: The thermal conductivity characteristics of the cutting fluid are calculated based on the temperature field distribution data, and the dynamic response characteristics of the thermal conductivity are corrected by combining the flow state data. Based on the synergistic relationship between pressure gradient data and flow state data, the viscosity coefficient of the cutting fluid is calculated, and the calculation results of specific heat capacity are optimized by the correlation between thermal conductivity and viscosity coefficient.

[0010] Preferably, the benchmark measurement results for generating the thermophysical property parameters of the cutting fluid based on the dynamic thermal conductivity, specific heat capacity characteristics, and viscosity coefficient include: The dynamic thermal conductivity, specific heat capacity, and viscosity coefficient are input into an adaptive signal processing algorithm to eliminate measurement interference components and generate an optimized set of thermal property parameters. A benchmark measurement database for the thermophysical properties of cutting fluids is constructed based on the optimized set of thermophysical parameters.

[0011] Preferably, the system further includes: Obtain historical operating condition data of the cutting fluid, and establish a cutting fluid performance evolution prediction model based on the historical operating condition data; The benchmark measurement database is input into the cutting fluid performance evolution prediction model to deduce the evolution trend of the cutting fluid thermophysical property parameters.

[0012] Preferably, the step of establishing a cutting fluid performance evolution prediction model based on the historical operating condition data includes: The characteristics of thermal conductivity, specific heat capacity and viscosity coefficient of cutting fluid in different processing cycles are extracted to construct the initial parameter space of the cutting fluid performance evolution prediction model; The initial parameter space is trained using a time-series prediction algorithm to establish a computational framework for the cutting fluid performance evolution prediction model.

[0013] Preferably, inputting the benchmark measurement database into the cutting fluid performance evolution prediction model includes: Adjust the prediction parameters of the cutting fluid performance evolution prediction model based on the deviation between the benchmark measurement data and historical data; The evolution trend of the thermophysical properties of the cutting fluid is output based on the adjusted prediction parameters.

[0014] Preferably, the system further includes: Based on the evolution trend map, a cutting fluid maintenance strategy is generated, which includes a component adjustment scheme and a replacement cycle recommendation. The parameter calculation mechanism for updating the cutting fluid thermophysical property coupling model based on the aforementioned cutting fluid maintenance strategy.

[0015] Preferably, updating the cutting fluid thermophysical property coupling model based on the cutting fluid maintenance strategy includes: Optimize the feature parameter extraction method of the cutting fluid thermophysical property coupling model based on the composition adjustment scheme; Based on the replacement cycle, it is recommended to adjust the data update frequency of the cutting fluid thermophysical coupling model.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This integrated online measurement system for cutting fluid thermophysical properties comprehensively collects temperature field distribution data, flow state data, and pressure gradient data of the cutting fluid by setting up a multimodal sensor array in the cutting area. Compared with traditional offline measurement methods, it effectively avoids measurement deviations caused by the cutting fluid being out of sync with the actual machining environment, ensuring that the collected data accurately reflects the actual state of the cutting fluid during machining. The application of the multimodal sensor array breaks through the limitations of existing online measurement technologies that can only collect single or partial data, achieving comprehensive capture of multi-dimensional state information of the cutting fluid throughout the machining area. This provides rich and reliable data source support for the subsequent accurate calculation of thermophysical properties, giving the parameter calculation a more solid foundation.

[0017] A coupled thermophysical property model of the cutting fluid was constructed based on the collected multi-dimensional data. This model was then used to calculate the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of the cutting fluid, achieving integrated calculation of multiple thermophysical parameters. This method of constructing a coupled model based on multi-dimensional data establishes the correlation between various parameters, comprehensively reflecting the overall thermophysical characteristics of the cutting fluid. Compared to existing single-parameter or discrete parameter measurement techniques, this method better reflects the objective laws governing the mutual influence and correlation of various cutting fluid parameters during actual machining. The calculated dynamic thermal conductivity, specific heat capacity, and viscosity coefficient more comprehensively and accurately reflect the dynamic changes of the cutting fluid during machining, providing a strong basis for machining personnel to fully understand the state of the cutting fluid.

[0018] This system generates benchmark measurement results based on calculated dynamic thermal conductivity, specific heat capacity, and viscosity coefficient, enabling online and continuous monitoring of the thermophysical properties of the cutting fluid. Machining personnel can use the real-time benchmark measurement results to understand the changing trends of the cutting fluid's thermophysical properties during machining. When changes in machining conditions lead to abnormalities in cutting fluid parameters, the system can quickly detect these changes and adjust the machining process accordingly, avoiding problems such as undetected parameter anomalies affecting machining quality or causing equipment malfunctions. Furthermore, the integrated online measurement method eliminates the need for cutting fluid extraction and offline processing, saving measurement time and improving parameter measurement efficiency. This provides timely parameter support for real-time control of the machining process, helping machining personnel optimize machining parameters based on the actual state of the cutting fluid, thereby improving machining efficiency and quality.

[0019] The cutting fluid thermophysical property coupling model constructed by this system calculates parameters through comprehensive analysis of multi-dimensional data, achieving higher calculation accuracy compared to existing parameter calculation models based on single data. Higher accuracy in thermophysical property parameters provides more precise references for refined control of the machining process. For example, in scenarios with high requirements for cutting fluid performance, such as high-speed cutting and precision machining, accurate parameter data helps machining personnel to more rationally select the type of cutting fluid, adjust the supply method and quantity, further improving the surface finish and machining efficiency, and reducing machining costs. Simultaneously, the integrated design of this system combines data acquisition, model calculation, and result generation, simplifying the measurement process, reducing operational complexity, and facilitating its application on different types of cutting equipment. It can provide cutting fluid parameter monitoring support for more machining scenarios, promoting the development of cutting fluid parameter monitoring and machining process optimization throughout the machining industry. Attached Figure Description

[0020] Figure 1 This is a timing diagram of the integrated online measurement system for the thermophysical properties of cutting fluid described in this invention. Figure 2 A schematic diagram illustrating the working principle of multimodal sensor array setup and data extraction; Figure 3 This diagram illustrates the working principle for calculating the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of cutting fluid. Detailed Implementation

[0021] 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.

[0022] Please see Figure 1 This invention provides an integrated online measurement system for the thermophysical properties of cutting fluid. The system includes a multimodal sensor array positioned in the machining zone to acquire real-time thermophysical state information of the cutting fluid. The multimodal sensor array is configured to acquire temperature field distribution data, flow state data, and pressure gradient data of the cutting fluid, reflecting its real-time thermodynamic behavior during machining. The acquired data is transmitted to a data processing unit, which constructs a coupled thermophysical property model of the cutting fluid based on the temperature field distribution data, flow state data, and pressure gradient data. This model solves for the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of the cutting fluid using multiphysics coupling equations. The solution process involves integrated calculations of heat conduction and fluid dynamics to ensure the accuracy of parameter calculations and dynamic response characteristics. Finally, the system generates benchmark measurement results of the thermophysical properties of the cutting fluid based on the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient. These results characterize the real-time performance state of the cutting fluid and provide a data basis for machining optimization. The system's implementation relies on a high-precision sensor array and real-time data processing algorithms, ensuring that the measurement process is performed online without interrupting machining operations.

[0023] Example 1: See Figure 2In the specific deployment of the multimodal sensor array, the system arranges sensors along the complete flow path of the cutting fluid from the supply pump outlet through the delivery pipeline, the machining area, to the return tank. The distributed temperature sensors are corrosion-resistant thin-film platinum resistance temperature sensors, which are embedded in the fixture wall or specially designed detection channels near the cutting area in a grid pattern with a spacing of no more than ten millimeters. These sensors are powered by a high-precision constant current source and use a four-wire connection method to eliminate lead resistance error. Their measurement data are continuously collected at millisecond intervals to capture the local temperature gradient changes of the cutting fluid at the tool-workpiece interface. This dense arrangement can effectively sense micro-temperature fluctuations caused by uneven distribution of machining heat sources or flow stagnation. The ultrasonic flow velocity detector employs one or more pairs of piezoelectric transducers mounted at a specific angle on both sides of a transparent or acoustically perforated flow channel. The transducers operate in pulse mode, with their emission frequency adjusted to the 1-5 MHz range based on the acoustic characteristics of the cutting fluid. The instantaneous flow velocity is calculated by measuring the time difference between the upstream and downstream propagation of the ultrasonic waves. The detector array covers the flow channel cross-section in a staggered arrangement, thus obtaining the instantaneous flow velocity distribution profile from the pipe wall to the central region. The micro-differential pressure sensor uses a differential pressure transmitter with a ceramic diaphragm. Its pressure tapping point is precisely set at a hydraulically characteristic location in the flow path, such as the constriction section of a nozzle, downstream of a pipe bend, or before or after a filter. The sensor transmits pressure signals through a capillary tube and uses a temperature compensation algorithm to eliminate environmental thermal drift, monitoring pressure difference changes between different areas. All sensor signals are connected to a multi-channel synchronous data acquisition instrument via shielded cables. This instrument has a built-in anti-aliasing filter and records data from each channel in parallel at a sampling rate of at least 10 kHz, ensuring the synchronization of spatiotemporal data.

[0024] The raw sensor data collected needs to be preprocessed to extract effective information. The local temperature gradient is obtained by combining the difference between adjacent temperature sensor measurement points with spatial coordinate interpolation. The instantaneous flow velocity distribution is reconstructed by performing correlation analysis and statistical processing on the ultrasonic flight time series. The regional pressure difference change is highlighted by differential calculation and moving average filtering. These processed data are organized into a multi-dimensional data structure, namely the cutting fluid three-dimensional state characterization matrix. The three spatial dimensions of this matrix correspond to the X, Y, and Z axis positions in the flow channel coordinate system, respectively. The time dimension is indexed by the acquisition timestamp. The physical quantity dimension includes three data types: temperature, flow velocity, and pressure. The matrix construction uses the Kriging interpolation algorithm to fuse discrete measurement point data into continuous field information. At the same time, the local coordinate system of the sensor is transformed into the global machining coordinate system through coordinate transformation. The process of extracting temperature field distribution data from the three-dimensional state characterization matrix involves isothermal surface reconstruction and heat flux density vector calculation of temperature dimension data. The extraction of flow state data is achieved by identifying vorticity and tracing streamlines in the velocity field. Pressure gradient data is obtained by directional differentiation of the pressure field. These extraction operations are completed using a specialized numerical calculation library, ensuring the accuracy and efficiency of data conversion.

[0025] Constructing a coupled thermo-physical property model for cutting fluids requires establishing a mathematical description of the interaction between heat and flow. This model establishes a multi-physics coupling relationship between the thermal conduction and flow characteristics of the cutting fluid based on the laws of conservation of energy and momentum, considering the constitutive behavior of the cutting fluid as a non-Newtonian fluid and the potential influence of latent heat of phase change. The model initialization phase requires loading a set of core parameters, including the cutting fluid's reference density, reference thermal conductivity, standard specific heat capacity, and viscosity-temperature coefficient. These initial values ​​are derived from the cutting fluid supplier's technical data sheet or calibrated through offline experiments. Simultaneously, solver control parameters such as convergence tolerance and maximum number of iterations need to be set. Preprocessed temperature field distribution data, flow state data, and pressure gradient data are input as boundary conditions and field variables into the initialized model. Temperature field data is used to define the thermal boundary of the computational domain, flow state data is used to set the inlet velocity conditions, and pressure gradient data is coupled to the source terms of the momentum equation.

[0026] The model solution employs a discretization scheme based on the finite volume method, dividing the spatial region occupied by the cutting fluid into hundreds of thousands to millions of control volume meshes. The governing equations are discretized on each control volume. The thermal-fluid coupling equations are solved using a separate algorithm to sequentially process the velocity and temperature fields. The coupling of pressure and velocity is achieved through a SIMPLE-like algorithm. The dynamic thermal conductivity is calculated by inverting the energy equation containing the flow term. The specific heat capacity is indirectly evaluated by analyzing the change in heat required per unit temperature rise. The viscosity coefficient is iteratively corrected by matching the simulated shear stress field with the measured velocity gradient field. The solution process is performed on a high-speed workstation, utilizing parallel computing technology to accelerate the solution of large linear equation systems. Each calculation iteration updates the physical property parameter field until the residuals reach below a set threshold. The final outputs spatially distributed dynamic thermal conductivity, time-varying specific heat capacity, and viscosity coefficient related to shear rate. These parameters reflect the instantaneous thermophysical state of the cutting fluid under real-world operating conditions. The optimized arrangement of the sensor array ensures the spatial resolution and timeliness of the source data. The construction of the three-dimensional state characterization matrix realizes the structured integration of multi-source heterogeneous data. The establishment and solution of the thermophysical coupling model transforms macroscopic measurable field information into microscopic physical property parameters. This implementation method makes it possible to continuously monitor the changes in the thermophysical properties of the cutting fluid online without stopping the machine or taking samples.

[0027] Example 2: See Figure 3 The core of calculating the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of cutting fluid lies in the collaborative analysis and inversion calculation of multi-physics field data. The system first analyzes the thermal conductivity characteristics of the cutting fluid based on the temperature field distribution data collected by the multi-modal sensor array arranged in the machining area. This analysis does not simply apply the static thermal conductivity formula, but treats the temperature field data as a dynamically changing scalar field. By analyzing the rate of change of temperature at each point in the field with time and the spatial gradient distribution, the heat transfer path and intensity are inferred. The calculation process is carried out on a discretized three-dimensional grid model. Each grid cell is assigned an instantaneous temperature value obtained by interpolation from the sensor data. By calculating the temperature difference and spatial distance between adjacent cells and combining it with the known time step, the magnitude and direction of the heat flux density vector can be estimated point by point, thereby obtaining a spatial distribution map reflecting the effective thermal conductivity of the cutting fluid under the current flow and thermal boundary conditions.

[0028] The dynamic response characteristics correction of thermal conductivity based on flow state data is an indispensable step, because the flow state of the cutting fluid in the machining zone (such as laminar flow, turbulent flow, or the presence of vortices) significantly affects its heat transport efficiency. The system spatially registers and temporally aligns the instantaneous velocity distribution field and temperature field acquired by the ultrasonic flow velocity detector, so that each calculation unit simultaneously has temperature value and velocity vector information. The correction process introduces the physical concept of convective heat transfer from fluid mechanics, that is, the macroscopic motion of the fluid enhances heat transport. The magnitude of the correction is related to the local flow velocity, fluid density, and an enhancement factor related to the flow state. For areas with high flow velocity or strong turbulence, the system will correspondingly increase the effective thermal conductivity value of the area to reflect the contribution of convective heat transfer. Conversely, in areas with stagnant flow or very low flow velocity, the thermal conductivity mainly depends on molecular thermal diffusion. This correction makes the calculated dynamic thermal conductivity no longer an intrinsic property of the material, but an equivalent parameter closely coupled with the current flow field, reflecting the overall heat transport capacity.

[0029] The calculation of the viscosity coefficient mainly relies on the synergistic relationship between pressure gradient data and flow state data. The system transforms the regional pressure difference change information measured by the micro-differential pressure sensor into a main pressure gradient field along the flow direction. According to the basic principles of fluid mechanics, the pressure drop generated by a viscous fluid to overcome internal friction during flow is directly related to the fluid's viscosity coefficient and velocity gradient. The system uses the obtained instantaneous velocity distribution data to calculate the shear rate tensor at each point in the flow field through numerical differentiation. Subsequently, the calculated pressure gradient vector and shear rate tensor are substituted into a preset constitutive model (such as a generalized Newtonian fluid model or a more complex non-Newtonian fluid model) for fitting and solving, thereby retrieving the viscosity coefficient of the cutting fluid. The selection of this constitutive model can be preset according to the known type of cutting fluid (such as emulsion, semi-synthetic fluid, or fully synthetic fluid). The solution process is an iterative optimization process, aiming to find a viscosity coefficient distribution field that minimizes the error between the theoretical pressure gradient calculated from it and the velocity field and the measured pressure gradient.

[0030] The correlation between thermal conductivity and viscosity coefficient was further used to optimize the calculation results of specific heat capacity. Specific heat capacity is a parameter that measures the amount of heat required for a unit mass of fluid to increase its temperature by one unit. In static measurements, it is usually considered a constant. However, in dynamically flowing cutting fluids, its effective value is affected by viscous dissipation. Viscous dissipation refers to the phenomenon of heat generation in a fluid due to internal shear friction. This heat contributes to the temperature increase of the fluid, thus becoming a non-negligible internal heat source in the energy balance equation. The system substitutes the previously calculated dynamic thermal conductivity field and viscosity field, along with the flow velocity field and temperature field data, into the energy conservation equation for global solution. In this equation, the viscous dissipation term participates in the calculation as a source term. By adjusting the value of specific heat capacity, the temperature change rate calculated by the equation and the temperature change rate measured by the sensor are optimally matched in the global range. This process is actually a calibration of specific heat capacity based on the full-field energy balance, so that the final specific heat capacity characteristic can reflect the comprehensive influence of viscous heat generation on the thermal inertia of the fluid. The three parameters—dynamic thermal conductivity, viscosity coefficient, and specific heat capacity—are not solved independently but are coupled with each other. Therefore, the calculation module adopts an alternating iterative solution strategy: first, two parameters are fixed to solve for the third, and then the new solution is used to update the other parameters. This process is repeated until the changes in all parameter fields are less than the set convergence threshold. The final output of dynamic thermal conductivity, specific heat capacity, and viscosity coefficient are three sets of data fields that vary with space and time. Together, they constitute a complete and dynamic description of the thermophysical properties of the cutting fluid under real working conditions, providing accurate input for subsequent performance evaluation and prediction.

[0031] Example 3: The generation of benchmark measurement results for the thermophysical properties of cutting fluid begins with the preprocessing of the original data sequences of dynamic thermal conductivity, specific heat capacity, and viscosity coefficient. These data, directly output from the thermophysical property coupling model, typically contain measurement noise and high-frequency interference components introduced by operating condition fluctuations. The system employs an adaptive signal processing algorithm based on variational mode decomposition to separate useful signals from noise. This algorithm adaptively decomposes the input signal into a series of mode functions with specific center frequencies. Modes whose frequency components do not conform to the actual variation law of cutting fluid properties are identified as interference and eliminated. For singularities in the viscosity coefficient sequence caused by the passage of instantaneous bubbles, the algorithm can decompose them into independent modes and remove them. For slow background drift, it is eliminated by subtracting low-frequency modes that are unrelated to the trend of property changes. This process can be represented as the signal... Decomposition:

[0032] in: The first one obtained by decomposition One modal function, For residual components, A time-series signal representing the original dynamic thermal conductivity, specific heat capacity, or viscosity coefficient; This represents the total number of modes adaptively determined by the algorithm; Indicates the first A mode function is an amplitude-frequency modulated signal with a specific center frequency and a finite bandwidth; The residual components after decomposition typically contain signal trend terms or noise. By evaluating the frequency characteristics of each mode function and its correlation with the processing cycle, the system automatically selects a subset of modes containing effective physical property change information for reconstruction, thereby obtaining a purified optimized set of thermal property parameters. The reconstructed signal is smooth and retains the true physical change trend.

[0033] The benchmark measurement database, built based on the optimized set of thermal property parameters, adopts a hybrid architecture combining time-series and relational databases. Each record contains a timestamp accurate to milliseconds, a set of spatial coordinates (corresponding to the position of the measurement point in the flow field), and the optimized thermal conductivity value at that moment. Specific heat capacity and viscosity coefficient value The database also stores context information acquired during data collection, such as spindle speed, feed rate, and workpiece material code. This information is obtained from the CNC system in real time via the OPCUA protocol. Data integrity constraints are implemented during the database creation process. For example, foreign key constraints ensure that each property record corresponds to a valid machining task number, and physical rationality checks are set for the numerical range to prevent abnormal values ​​caused by sensor failures from being entered into the database.

[0034] Historical operating data of the cutting fluid is obtained by querying the database interface of the manufacturing execution system. The historical data covers complete records of multiple cutting fluid usage cycles, including the time and dosage of each additive replenishment, periodic laboratory sampling analysis reports, and corresponding machining load data (estimated by total metal removal). Extracting the variation characteristics of thermal conductivity, specific heat capacity, and viscosity coefficient from this historical data involves feature engineering. The system calculates the linear fitting slope of the physical property parameters within each complete life cycle to characterize its decay rate, calculates the standard deviation within the sliding window to quantify its volatility, and extracts short-term variation patterns related to maintenance events (such as replenishment and filtration). These features constitute a high-dimensional initial parameter space. The cutting fluid performance evolution prediction model based on historical operating data adopts a long short-term memory network framework. This network structure can capture long-term dependencies in the time series. The network training uses the feature sequence in the historical data as input and the physical property parameter values ​​after a certain period of time in the future as the prediction target. The training process is carried out by unfolding through time steps. The network weights are optimized using the stochastic gradient descent algorithm with dropout to prevent overfitting. The established computational framework includes an encoder-decoder structure. The encoder is responsible for learning the hidden state of the historical sequence, and the decoder gradually generates the prediction sequence of future physical property parameters based on the state.

[0035] Inputting the benchmark measurement database into the cutting fluid performance evolution prediction model is a continuous assimilation process. The deviation between the new measurement data and the model's predicted values ​​is calculated, and this deviation is used in two ways: first, by updating the model's estimate of the current hidden state of the cutting fluid (such as the effective concentration of additives and the degree of contamination) through the extended Kalman filter algorithm; second, when the deviation continuously exceeds a predetermined threshold, the model's incremental learning mechanism is triggered, using the latest data to fine-tune some weights of the network with a small learning rate, so that the model can adapt to possible operating condition drift. The evolution trend of the cutting fluid's thermophysical property parameters is inferred through iterative execution of a trained LSTM network. The model uses the current state as the initial condition and combines it with future planned machining schedules (as external condition input) to automatically infer the change path of the physical property parameters over the next few days or weeks. The inference results are output in the form of quantiles, providing not only the predicted median trajectory but also the 10% and 90% quantile trajectories to characterize the uncertainty of the prediction. This evolution trend map visually shows the time points when the parameters may exceed the operating threshold, providing a quantitative basis for predictive maintenance. The entire implementation method forms a complete technology chain from real-time data purification, database construction, historical pattern learning to future trend prediction.

[0036] Example 4: Establishing a cutting fluid performance evolution prediction model requires systematic processing of historical observation data. Taking a semi-synthetic cutting fluid used on an automotive engine block machining line as an example, its historical data records more than twelve months of continuous operation information. The data is collected from the multimodal sensor array integrated in the machining center and the work logs of the manufacturing execution system. The primary task of this stage is to extract the variation characteristics of the cutting fluid's thermal conductivity, specific heat capacity, and viscosity coefficient in each machining cycle (usually eight hours per shift) from these historical data. Feature extraction is not simply recording the mean, but rather using time series analysis methods to mine its inherent patterns. For example, the moving average of each parameter over multiple consecutive shifts is calculated to capture long-term decay trends, the moving standard deviation is calculated to assess the stability of parameter fluctuations, and the fast Fourier transform is used to identify possible periodicity in parameter changes (such as cycles related to weekly concentrated fluid replenishment). In addition, the autocorrelation coefficient of the parameter sequence is calculated to determine the degree of influence of historical values ​​on current values. These calculated feature values ​​together constitute the initial parameter space describing the performance state of the cutting fluid. This space has a high dimension and contains information on multiple aspects such as trends, fluctuations, and cycles.

[0037] After constructing the initial parameter space, dimensionality reduction is required to improve model efficiency and reduce noise interference. Principal component analysis is applied to this process. It transforms multiple potentially related feature variables into a few independent principal component variables through linear transformation. These principal components can retain most of the variance information of the original data. For example, analysis may reveal that the first three principal components contribute more than 90% of the total variance. One principal component mainly represents the long-term trend of the parameter, another principal component represents the periodic fluctuation intensity of the parameter, and the third principal component is related to sudden changes. Through dimensionality reduction, a complex parameter space containing dozens of original features is simplified into a low-dimensional space composed of a few principal components. The time-series prediction algorithm is trained on this low-dimensional parameter space to establish the computational framework for the cutting fluid performance evolution prediction model. Considering that the evolution of cutting fluid parameters has both long-term trends and short-term fluctuations, a long short-term memory network is selected as the core algorithm. This network structure can effectively learn the long-term dependencies in the time series. During the training process, historical data is divided into training and validation sets according to time order. For example, the data of the first ten months is used for training, and the data of the last two months is used to validate the model's prediction effect. The network learns the mapping relationship from the input sequence (such as the principal component values ​​of the past week) to the output sequence (such as the parameter prediction values ​​of the next week) by unfolding the time step. The training objective is to minimize the difference between the predicted value and the true value. The tens of thousands of connection weights inside the network are continuously adjusted through the backpropagation algorithm and the gradient descent method until the prediction error of the model on the validation set tends to stabilize.

[0038] Inputting the benchmark measurement database generated by the real-time online measurement system into the trained prediction model is a dynamic update process. The newly input data is first compared with the model's predictions based on historical data, generating a series of deviations. These deviations reveal the systematic differences between the model's predictions and actual observations. Based on these deviations, the system will activate the adjustment mechanism for the model's prediction parameters. For example, an online learning technique can be used to fine-tune some weights of the LSTM network with a small learning rate based on the latest deviation data. This fine-tuning allows the model to gradually adapt to possible changes in the degradation law of cutting fluid performance. For example, the thermal load may be increased due to the change of workpiece material, thereby accelerating the degradation of certain properties of the cutting fluid.

[0039] Based on the adjusted prediction parameters, the system outputs a graph showing the evolution trend of the cutting fluid's thermophysical properties. This graph is a comprehensive visualization that not only displays the predicted center trajectories of thermal conductivity, specific heat capacity, and viscosity coefficient over a future period (such as the next planned maintenance cycle), but also clearly marks the uncertainty range of the predicted values ​​(e.g., a 95% confidence interval) using shaded areas or error bars. The horizontal axis of the graph represents time, and the vertical axis represents the normalized parameter values ​​or actual physical quantity units. Important maintenance events (such as the last fluid top-up or filter cleaning) are usually marked at key time points for reference. This graphical output allows maintenance personnel to intuitively assess the health status of the cutting fluid and proactively identify potential risks where parameters may exceed permissible operating ranges. Table 1 shows partial results of feature extraction from historical viscosity coefficient data of the example cutting fluid over a given week. This data is derived from daily aggregated values ​​in a benchmark measurement database.

[0040] Table 1: Feature Extraction of Weekly Data on Cutting Fluid Viscosity Coefficient

[0041] The entire implementation process embodies a closed-loop approach that learns evolutionary patterns from historical data and uses real-time data for adaptive model correction. By extracting multidimensional features of performance degradation through feature engineering, using advanced time-series prediction algorithms to capture complex change patterns, and combining this with an online learning mechanism to keep the model sensitive to current operating conditions, the prediction results are finally output through intuitive trend graphs, providing a dynamic and reliable basis for precise management and replacement decisions of cutting fluid.

[0042] Example 5: Generating a cutting fluid maintenance strategy based on an evolution trend graph is a process of converting predicted data into specific operational instructions. The parameter change trajectory shown in the graph is compared and analyzed with the preset safe operating threshold. The preset rule engine inside the system contains multiple logical judgment statements. For example, when the graph shows that the viscosity coefficient has multiple consecutive predicted points higher than the upper limit threshold and its upward slope exceeds the set value, the rule engine will trigger the "composition adjustment" flag and further determine the imbalance type based on the change pattern of thermal conductivity and specific heat capacity. If accompanied by a significant decrease in thermal conductivity, it may be judged as consumption of lubricating components or accumulation of contaminants. At this time, the generated composition adjustment plan will specify the type of additive to be added, such as adding a functional concentrate containing extreme pressure anti-wear agents, and calculating the approximate addition range. This calculation will refer to the current concentration of the cutting fluid, the total system volume, and the predicted performance gap. For replacement cycle recommendations, the rule engine will comprehensively evaluate the time when the predicted values ​​of each parameter exceed the critical point, select the most conservative time point as the recommended replacement date, and take into account the production plan to give a recommended time window rather than a single time point. All generated strategies will be output in the form of a structured report, including maintenance type, recommended action, execution time window and expected goal.

[0043] Updating the cutting fluid thermophysical property coupling model based on the cutting fluid maintenance strategy is a key step to ensure the model's continued accuracy. When the composition adjustment scheme is implemented, the basic composition of the cutting fluid changes, which directly affects its thermophysical properties. The model update is first reflected in the optimization of the feature parameter extraction method. The system will introduce the adjusted chemical components (such as the types and quantities of newly added additives) as new input variables into the model. The empirical coefficients or weights in the model that were originally used to describe the relationship between the cutting fluid components and properties need to be recalibrated. The calibration process is carried out using newly collected high-frequency measurement data within a period of time after the adjustment. By minimizing the difference between the new data prediction values ​​and the measured values, these feature parameters are optimized in reverse. For example, the weight coefficients characterizing the degree of influence of additives on thermal conductivity will be refitted, so that the model can more accurately reflect the thermal conduction behavior of the new mixed liquid.

[0044] Adjusting the data update frequency of the cutting fluid thermophysical coupling model according to the replacement cycle recommendation involves changes to the model's adaptive learning mechanism. The performance degradation rate of the cutting fluid is not constant at different stages of its lifespan, typically accelerating towards the end of its service life. Therefore, the model needs to learn and track this change at different paces. In the initial stage after cutting fluid replacement, the system uses a higher data update frequency, for example, fine-tuning the model every half hour using the latest collected data. This helps the model quickly learn and adapt to the initial characteristics of the new fluid. When the system reaches the middle stage and performance is relatively stable, the data update frequency can be reduced to once every four hours to balance computational load and tracking accuracy. When the predictive model indicates that the cutting fluid lifespan has entered the final quarter stage, the system automatically increases the data update frequency to once every fifteen minutes. This dynamic adjustment mechanism allows the model to remain highly vigilant at critical moments, like an experienced mechanic, capturing subtle signs of accelerated performance degradation, thus providing more timely warnings for the next maintenance decision.

[0045] A closed-loop feedback loop is formed between the execution of maintenance strategies and model updates. For example, in a mass production of aluminum alloy housings, the system predicts that the cooling performance of the cutting fluid will soon be insufficient and generates a strategy to add 5% new concentrate. After the maintenance personnel execute the strategy, the system detects that the actual increase in thermal conductivity is slightly lower than the model's initial expectation. This small deviation is recorded and used to optimize the calculation logic when generating similar composition adjustment schemes in the future. The model learns that under such processing conditions, the efficiency coefficient of the additive needs to be slightly reduced. This continuous learning makes the entire system increasingly accurate, and its recommended addition ratio or replacement cycle is more in line with the actual needs of specific production environments. It transforms offline human experience-based decision-making into online, data-driven automated decision support. The generation of maintenance strategies transforms the prediction results from simple charts into clear, actionable instructions, while the online updating of the model ensures that the system's understanding can evolve in sync with the actual state of the cutting fluid, avoiding prediction bias caused by model aging.

[0046] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0047] 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. An integrated online measurement system for the thermophysical properties of cutting fluid, characterized in that, The system includes: A multimodal sensor array is set up in the cutting area to collect data on the temperature field distribution, flow state, and pressure gradient of the cutting fluid; Based on the temperature field distribution data, flow state data, and pressure gradient data, a cutting fluid thermophysical property coupling model is constructed. The dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of the cutting fluid are calculated using the cutting fluid thermophysical property coupling model. The benchmark measurement results for generating the thermophysical property parameters of the cutting fluid are based on the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient.

2. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 1, characterized in that, The step of setting up a multimodal sensing array in the cutting area includes: Distributed temperature sensors, ultrasonic flow rate detectors, and micro-differential pressure sensors are arranged along the cutting fluid flow path to obtain the local temperature gradient, instantaneous flow rate distribution, and regional pressure difference changes of the cutting fluid. A three-dimensional state characterization matrix for the cutting fluid is established based on the local temperature gradient, instantaneous flow velocity distribution, and regional pressure difference changes. Temperature field distribution data, flow state data, and pressure gradient data are then extracted from the three-dimensional state characterization matrix.

3. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 1, characterized in that, The construction of the cutting fluid thermophysical property coupling model based on the temperature field distribution data, flow state data, and pressure gradient data includes: Establish the multi-physics coupling relationship between the thermal conduction and flow characteristics of the cutting fluid, and initialize the core parameter set of the cutting fluid thermal property coupling model; The temperature field distribution data, flow state data, and pressure gradient data are input into the cutting fluid thermo-physical property coupling model, and the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient are solved by the thermo-fluid coupling equation.

4. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 3, characterized in that, The calculation of the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient of the cutting fluid using the cutting fluid thermophysical property coupling model includes: The thermal conductivity characteristics of the cutting fluid are calculated based on the temperature field distribution data, and the dynamic response characteristics of the thermal conductivity are corrected by combining the flow state data. Based on the synergistic relationship between pressure gradient data and flow state data, the viscosity coefficient of the cutting fluid is calculated, and the calculation results of specific heat capacity are optimized by the correlation between thermal conductivity and viscosity coefficient.

5. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 1, characterized in that, The benchmark measurement results for generating the thermophysical property parameters of the cutting fluid based on the dynamic thermal conductivity, specific heat capacity, and viscosity coefficient include: The dynamic thermal conductivity, specific heat capacity, and viscosity coefficient are input into an adaptive signal processing algorithm to eliminate measurement interference components and generate an optimized set of thermal property parameters. A benchmark measurement database for the thermophysical properties of cutting fluids is constructed based on the optimized set of thermophysical parameters.

6. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 5, characterized in that, The system also includes: Obtain historical operating condition data of the cutting fluid, and establish a cutting fluid performance evolution prediction model based on the historical operating condition data; The benchmark measurement database is input into the cutting fluid performance evolution prediction model to deduce the evolution trend of the cutting fluid thermophysical property parameters.

7. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 6, characterized in that, The cutting fluid performance evolution prediction model established based on the historical operating condition data includes: The characteristics of thermal conductivity, specific heat capacity and viscosity coefficient of cutting fluid in different processing cycles are extracted to construct the initial parameter space of the cutting fluid performance evolution prediction model; The initial parameter space is trained using a time-series prediction algorithm to establish a computational framework for the cutting fluid performance evolution prediction model.

8. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 7, characterized in that, The step of inputting the benchmark measurement database into the cutting fluid performance evolution prediction model includes: Adjust the prediction parameters of the cutting fluid performance evolution prediction model based on the deviation between the benchmark measurement data and historical data; The evolution trend of the thermophysical properties of the cutting fluid is output based on the adjusted prediction parameters.

9. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 8, characterized in that, The system also includes: Based on the evolution trend map, a cutting fluid maintenance strategy is generated, which includes a component adjustment scheme and a replacement cycle recommendation. The parameter calculation mechanism for updating the cutting fluid thermophysical property coupling model based on the aforementioned cutting fluid maintenance strategy.

10. The integrated online measurement system for the thermophysical properties of cutting fluid as described in claim 9, characterized in that, The updating of the cutting fluid thermophysical property coupling model based on the cutting fluid maintenance strategy includes: Optimize the feature parameter extraction method of the cutting fluid thermophysical property coupling model based on the composition adjustment scheme; Based on the replacement cycle, it is recommended to adjust the data update frequency of the cutting fluid thermophysical coupling model.

Images