Multifunctional integrated cable simulation working condition test platform

Abstract

The invention relates to the field of cable simulation working condition testing, in particular to a multifunctional integrated cable simulation working condition testing platform, which comprises a cable testing module, a testing module and a testing module, and is characterized in that the cable testing module is used for acquiring temperature parameters and load parameters of a heater and a multi-degree-of-freedom loading head for testing a tested cable; the neural network module is used for determining control parameters according to the temperature parameters and the load parameters through a control parameter fitting model based on a BP neural network architecture; and the control module is used for generating the execution temperature of the heater and the execution acting force of the multi-degree-of-freedom loading head according to the control parameters through a control algorithm. The multifunctional test platform does not need to depend on an artificial preset model and can autonomously learn control parameters corresponding to complex extreme working conditions, the adaptability of the test platform to complex working condition tests is improved, and stable and accurate parameter control over the multi-degree-of-freedom loading head and the heater of the platform is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cable simulation working condition test, and in particular to a multifunctional integrated cable simulation working condition test platform. BACKGROUND

[0002] As a key component in power transmission, communication network and industrial industry, the reliability of cable is directly related to the safe and stable operation of infrastructure. Moreover, with the rapid development of new energy, rail transportation, deep sea engineering and other fields, cables need to serve in complex working conditions for a long time, such as soil pressure in buried environment, dynamic tension in overhead laying, heat radiation beside high-temperature pipeline and other complex scenes, which puts forward strict requirements on the mechanical properties, temperature resistance and durability of cables.

[0003] Currently, cable reliability test mainly relies on multiple independent devices, such as using a tension machine for tensile property test, a press machine for simulating soil extrusion, and a high-low temperature chamber for controlling environmental temperature. However, this decentralized test mode has significant limitations, and it is difficult to reproduce actual working conditions such as buried pressure, overhead tension, and combined temperature and force in high-temperature pipeline pressure. Independent devices can only simulate a single physical field and cannot reflect the influence of multi-stress coupling on cable insulation layer aging, structural deformation and failure mechanism, resulting in deviation between test results and real service performance. Moreover, multiple devices are scattered, leading to a tedious test process, multiple sample transfers, low efficiency and easy introduction of errors, high procurement and operation cost of multiple devices, and low laboratory space utilization.

[0004] Therefore, at present, there is a tendency to use an integrated test platform with multi-parameter collaborative control and intelligent dynamic adjustment capability to realize cable simulation test under complex actual working conditions.

[0005] Among them, how the multifunctional integrated cable simulation working condition test platform realizes the simulation of complex extreme working conditions through the collaborative control of the multi-degree-of-freedom loading head and the heater is a technical problem to be solved at present. SUMMARY

[0006] Therefore, the present application provides a multifunctional integrated cable simulation working condition test platform, which realizes the collaborative application of temperature and multi-degree-of-freedom mechanical load through a cable test module, can simulate and test the complex extreme working conditions of temperature mechanics that cables face in actual operation, and can accurately capture the nonlinear relationship between temperature and load parameters through a control parameter fitting model based on a BP neural network architecture, so that the multifunctional test platform does not need to rely on artificial preset models, can independently learn the control parameters corresponding to complex extreme working conditions, improves the adaptability of the test platform to complex working condition test, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the platform.

[0007] To achieve the above object, the application provides a multifunctional integrated cable simulation working condition test platform, which comprises: a cable test module, which is used for collecting temperature parameters and load parameters of a test cable tested by a heater and a multi-degree-of-freedom loading head; a neural network module, which is used for determining control parameters by fitting a control parameter fitting model based on a BP neural network architecture with the temperature parameters and the load parameters and adjusting model parameters of the control parameter fitting model through back propagation of a test stationary loss function; a control module, which is used for generating an execution temperature of the heater and an execution force of the multi-degree-of-freedom loading head through a control algorithm with the control parameters.

[0008] Further, the neural network module comprises: a neural network forward propagation submodule, which is used for generating the control parameters by fitting the control parameter fitting model based on the BP neural network architecture with the temperature parameters and the load parameters; a neural network back propagation submodule, which is used for adjusting model parameters of the control parameter fitting model through back propagation of the test stationary loss function with the load parameters.

[0009] Further, the load parameters comprise a force displacement sequence and a force parameter sequence, the control parameter fitting model comprises a thermodynamic compensation mechanism, a loading head resonance suppression submodule, a gating fusion layer, a convolution pooling layer, a full connection layer and a constraint layer, and the neural network forward propagation submodule comprises: an input integration unit, which is used for integrating the temperature parameters, the force displacement sequence, the force parameter sequence, a temperature set value, a displacement set value and a force set value into channels of a basic input vector respectively and generating a thermodynamic input vector based on the basic input vector and a thermal conductivity coefficient of a cable material and a loading head motion input vector based on the basic input vector and a stiffness of the cable material; a thermodynamic processing unit, which is used for generating thermodynamic characteristics by the thermodynamic compensation mechanism with the thermodynamic input vector; a resonance suppression unit, which is used for generating loading head characteristics by the loading head resonance suppression submodule with the loading head motion input vector; a gating fusion unit, which is used for generating comprehensive physical characteristics by the gating fusion layer with the thermodynamic characteristics and the loading head characteristics; a convolution unit, which is used for generating comprehensive time sequence characteristics by the convolution pooling layer with the basic input vector; a mapping unit, which is used for generating original control parameters by feature mapping of the comprehensive time sequence characteristics and the comprehensive physical characteristics through the full connection layer; A constraint unit configured to perform output constraint on the original control parameter by a constraint layer based on the cable material thermal conductivity and the cable material stiffness to generate the control parameter.

[0010] Further, the thermodynamic compensation mechanism includes an expansion convolution layer and a thermal delay compensation layer, and the thermodynamic processing unit includes: An expansion convolution sub-unit configured to perform time sequence extraction on the thermodynamic flow input vector by the expansion convolution layer to generate thermodynamic time sequence features; A thermal delay compensation sub-unit configured to perform thermal delay compensation on the current thermodynamic time sequence features and the historical thermodynamic time sequence features by the thermal delay compensation layer to generate the thermodynamic features.

[0011] Further, the loading head resonance suppression sub-model includes a differential feature extraction layer and a resonance suppression operation layer, and the resonance suppression unit includes: A differential enhancement extraction sub-unit configured to reduce fixed fluctuations on the loading head motion input vector by the differential operation to generate loading head time sequence features; A resonance suppression sub-unit configured to perform resonance suppression operation on the mechanical time sequence features by the resonance suppression operation layer to generate the loading head features.

[0012] Further, the convolution pooling layer includes a first convolution pooling layer, a second convolution pooling layer, and a flattening layer, and a convolution kernel size of the first convolution pooling layer is smaller than a convolution kernel size of the second convolution pooling layer, and the convolution unit includes: A first convolution pooling sub-unit configured to perform the first convolution pooling layer on the basic input vector to generate local time sequence features; A second convolution pooling sub-unit configured to perform the second convolution pooling layer on the local time sequence features to generate global time sequence features; A flattening layer sub-unit configured to perform the flattening layer on the global features to generate the comprehensive time sequence features.

[0013] Further, the control parameter includes a temperature control parameter and a force control parameter, and the constraint unit includes: A thermodynamic constraint sub-unit configured to perform thermodynamic constraint on the temperature control parameter based on the temperature parameter and a maximum thermal energy of the heater; A loading head constraint sub-unit configured to perform force constraint on the force control parameter based on the cable material stiffness and a maximum stress of the loading head.

[0014] Especially, the thermodynamic compensation mechanism and the loading head resonance suppression sub-model respectively learn the temperature field hysteresis effect and the mechanical system resonance interference, and the gating fusion layer realizes adaptive coupling of physical field characteristics through dynamic weight distribution. This hierarchical learning architecture enables the BP neural network to accurately fit the strong non-linear relationship in extreme scenarios, reduces the adjustment error of the control parameters, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the test platform corresponding to complex extreme working conditions.

[0015] Further, the neural network back propagation sub-module comprises: A loss function construction unit is configured to construct the test platform loss function based on the control parameter error term and the control parameter change rate.

[0016] Further, the neural network back propagation sub-module further comprises: A back propagation unit is configured to sequentially perform back propagation adjustment of the control parameter fitting model in the control parameter direction, the convolution direction, the thermodynamic direction and the resonance suppression direction based on the test platform loss function.

[0017] Further, the heater and the multi-degree-of-freedom loading head are integrated in the environmental simulation cabin. The multi-degree-of-freedom loading head comprises a hydraulic cylinder array and a universal joint for controlling the hydraulic cylinder array. The heater comprises at least a semiconductor refrigeration sheet, an infrared heating plate and a convection fan.

[0018] Especially, through real-time feedback of the load parameter sequence, the network weight is dynamically corrected along the gradient path in the control parameter direction, the convolution direction, the thermodynamic direction and the resonance suppression direction. Compared with the traditional single direction gradient descent, the parameter convergence speed in extreme working conditions is improved.

[0019] Compared with the prior art, the present application has the beneficial effects that the present application realizes the cooperative application of temperature and multi-degree-of-freedom mechanical load through the cable test module, can simulate the complex extreme working conditions of temperature and mechanics that the cable faces in actual operation, and through the control parameter fitting model based on the BP neural network architecture, can accurately capture the non-linear relationship between temperature and load parameters, so that the multi-functional test platform does not need to rely on artificial preset model, can independently learn the corresponding control parameters of complex extreme working conditions, improves the adaptability of the test platform to complex working condition test, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the platform.

[0020] Especially, the application learns temperature field hysteresis effect and mechanical system resonance interference through a thermodynamic compensation mechanism and a loading head resonance suppression model respectively, and a dynamic weight distribution is realized by a gating fusion layer to achieve adaptive coupling of physical field characteristics. This hierarchical learning architecture enables the BP neural network to accurately fit the strong non-linear relationship in extreme scenarios, reduces the adjustment error of the control parameters, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the test platform under complex extreme working conditions.

[0021] Especially, the application dynamically corrects the network weight along the gradient path in the control parameter direction, the convolution direction, the thermodynamic direction and the resonance suppression direction through real-time feedback of the load parameter sequence. Compared with the traditional single direction gradient descent, the parameter convergence speed under extreme conditions is improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 The structural schematic diagram of the multifunctional integrated cable simulation working condition test platform of the embodiment of the application is shown in the figure. Figure 2 The flowchart of the multifunctional integrated cable simulation working condition test platform of the embodiment of the application is shown in the figure. Figure 3 The forward propagation process schematic diagram of the control parameter fitting model of the multifunctional integrated cable simulation working condition test platform of the embodiment of the application is shown in the figure. Figure 4 The backward propagation process schematic diagram of the control parameter fitting model of the multifunctional integrated cable simulation working condition test platform of the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0023] In order to make the purpose and advantages of the application more clear and apparent, the application will be further described below in conjunction with the embodiments; it should be understood that the specific embodiments described herein are only used to explain the application, and do not limit the application.

[0024] The preferred embodiments of the application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the application, and are not intended to limit the protection scope of the application.

[0025] It should be noted that in the description of the application, the terms "up", "down", "left", "right", "in", "out" and the like indicate the direction or positional relationship terms based on the direction or positional relationship shown in the drawings, which are only for the convenience of description, and do not indicate or imply that the device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the application.

[0026] Moreover, it needs to be explained that in the description of the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connecting" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0027] As shown in Figures 1 to 4 The present application provides a multifunctional integrated cable simulation working condition test platform, which realizes the cooperative application of temperature and multi-degree-of-freedom mechanical load through a cable test module, can simulate and test the complex extreme working conditions of temperature mechanics that the cable faces in actual operation, can accurately capture the nonlinear relationship between temperature and load parameters through a control parameter fitting model based on a BP neural network architecture, so that the multifunctional test platform does not need to rely on artificial preset model, can autonomously learn the control parameters corresponding to complex extreme working conditions, improves the adaptability of the test platform to complex working condition test, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the platform.

[0028] As shown in Figure 1 and 2 The present embodiment provides a multifunctional integrated cable simulation working condition test platform, which comprises: A cable test module is used to collect temperature parameters and load parameters of the test cable tested by the heater and the multi-degree-of-freedom loading head; A neural network module is used to determine control parameters through a control parameter fitting model based on a BP neural network architecture, and to perform backward propagation adjustment of model parameters of the control parameter fitting model through a test flat loss function; A control module is used to generate an execution temperature of the heater and an execution force of the multi-degree-of-freedom loading head through a control algorithm.

[0029] Especially, the multifunctional integrated cable simulation working condition test platform realizes stable and accurate test parameter adjustment under extreme temperature and force working conditions of cable test, and improves the adaptability of the test platform to complex working condition test.

[0030] Further, the heater and the multi-degree-of-freedom loading head are integrally arranged in an environment simulation cabin; The multi-degree-of-freedom loading head comprises a hydraulic cylinder array and a universal joint for controlling the hydraulic cylinder array; The heater comprises at least a semiconductor refrigeration sheet, an infrared heating plate and a convection fan.

[0031] Especially, through the combination of the hydraulic cylinder array of the multi-degree-of-freedom loading head and the universal joint, the multi-directional mechanical loads, such as tensile, bending, torsion and other composite stresses, that the cable bears in actual use can be simulated, and the temperature control function in the environmental simulation cabin is combined to realize the cooperative test of mechanics and environmental factors, and the actual complex working conditions are approached more closely.

[0032] Specifically, the multi-degree-of-freedom loading head of the multifunctional integrated cable simulation working condition test platform adopts a mechanism arrangement of coupling of a hydraulic cylinder array and a universal joint, and includes: three groups of symmetrically arranged servo hydraulic cylinders (axial ±50kN, lateral ±10kN); the output end of the hydraulic cylinder array is connected with an adaptive universal joint, and the vector synthesis of tensile, compressive and lateral thrust is realized through a spherical hinge; a six-dimensional force sensor is arranged at the end of the loading head to realize real-time feedback of the composite load value.

[0033] Specifically, the multifunctional integrated cable simulation working condition test platform adopts a compact environmental simulation cabin, the inside of which is integrated with the multi-degree-of-freedom loading head and the heater, the environmental simulation cabin adopts a double-layer stainless steel cabin body, the interlayer of which is filled with ceramic fiber thermal insulation material. The heater is preferably a semiconductor refrigeration fin array arranged at the top of the environmental simulation cabin and an infrared heating plate installed at the bottom, and the temperature uniformity in the cabin is realized through the convection fan inside the environmental simulation cabin.

[0034] As shown in Figure 3 Further, the neural network module includes: a neural network forward propagation sub-module, configured to fit the temperature parameter and the load parameter through a control parameter fitting model based on a BP neural network architecture to generate the control parameter; a neural network backward propagation sub-module, configured to perform backward propagation adjustment of model parameters of the control parameter fitting model through a test stationary loss function based on the load parameter.

[0035] Especially, through the forward propagation of the BP neural network architecture, the temperature and load parameters can be nonlinearly fitted, the control parameter that is adaptive to complex working conditions is dynamically generated, through the backward propagation sub-module combined with the test stationary loss function, the model parameters can be dynamically adjusted based on the real-time load parameter feedback, the online iterative optimization of the control parameter fitting model is realized, and the adaptability of the system to working condition fluctuations and the long-term operation stability are improved. Therefore, compared with the traditional linear model, the coupling relationship between the parameters can be captured, the precision of the control strategy is improved, the accurate performance parameters of the test cable under complex working conditions are realized, and then the foundation for more accurate performance improvement of the cable under complex working conditions is laid.

[0036] Further, the load parameter includes a force displacement sequence and a force parameter sequence, the control parameter fitting model includes a thermodynamic compensation mechanism, a loading head resonance suppression submodel, a gated fusion layer, a convolution pooling layer, a fully connected layer, and a constraint layer, and the neural network forward propagation submodule includes: an input integration unit configured to integrate the temperature parameter, the force displacement sequence, the force parameter sequence, a temperature set value, a displacement set value, and a force set value into channels of a basic input vector respectively, and generate a thermodynamic input vector based on the basic input vector and a cable material thermal conductivity coefficient and a loading head motion input vector based on the basic input vector and a cable material stiffness; a thermodynamic processing unit configured to pass the thermodynamic input vector through a thermodynamic compensation mechanism to generate a thermodynamic feature; a resonance suppression unit configured to pass the loading head motion input vector through a loading head resonance suppression submodel to generate a loading head feature; a gated fusion unit configured to pass the thermodynamic feature and the loading head feature through a gated fusion layer to generate a comprehensive physical feature; a convolution unit configured to pass the basic input vector through a convolution pooling layer to generate a comprehensive time sequence feature; a mapping unit configured to perform feature mapping on the comprehensive time sequence feature and the comprehensive physical feature through a fully connected layer to generate an original control parameter; a constraint unit configured to perform output constraint on the original control parameter based on the cable material thermal conductivity coefficient and the cable material stiffness through a constraint layer to generate the control parameter.

[0037] In particular, the input integration unit constructs multi-dimensional parameters such as temperature, load, and material characteristics (cable material thermal conductivity coefficient and cable material stiffness) into a thermodynamic and kinematic dual-channel input vector, realizes deep fusion of mechanical and thermal parameters, avoids the limitations of single physical field analysis in traditional models, and improves the relevance and accuracy of control parameters under complex working conditions.

[0038] Specifically, dimension 1 of the basic input vector includes the temperature parameter, the force displacement sequence, the force parameter sequence, a temperature set value, a displacement set value, and a force set value, and dimension 2 of the basic input vector is a time step, and the last 60 sampling points (about 3 seconds of data) are taken through a sliding window.

[0039] It should be noted that the force displacement parameter sequence and the displacement set value in the embodiment are the force position of the multi-degree-of-freedom loading head on the test cable collected by a six-dimensional force sensor of the multi-degree-of-freedom loading head, and the force parameter sequence and the force set value in the embodiment are the force size of the multi-degree-of-freedom loading head on the test cable.

[0040] Specifically, the control parameters include 9 PID control parameters of a corresponding heater temperature controller, a force position controller of the multi-degree-of-freedom loading head, and a force size controller of the multi-degree-of-freedom loading head.

[0041] Further, the thermodynamic compensation mechanism includes an expansion convolution layer and a thermal delay compensation layer, and the thermodynamic processing unit includes: an expansion convolution subunit, configured to perform time sequence extraction on the thermodynamic flow input vector through the expansion convolution layer to generate thermodynamic time sequence features; a thermal delay compensation subunit, configured to perform thermal delay compensation on the current thermodynamic time sequence features and the historical thermodynamic time sequence features through the thermal delay compensation layer to generate the thermodynamic features.

[0042] In particular, the thermal delay compensation layer constructs a time correlation model of temperature field evolution by fusing the current and historical thermodynamic time sequence features, can accurately quantify and offset the thermal conduction lag effect, such as temperature response delay caused by thermal inertia of cable materials, and reduces test errors caused by temperature control lag, such as deviation of thermal stress loading time, and is particularly suitable for cycle test scenarios under complex working conditions.

[0043] Specifically, the thermodynamic flow input vector further includes at least where T represents a current temperature, i.e., the temperature parameter, represents a temperature set value, represents a heater power, represents a thermal conductivity of the cable material, represents a cooling rate of the cable material, represents a temperature change rate.

[0044] Specifically, the expansion convolution layer can be represented as: ; In the formula, represents the thermodynamic time sequence features, represents a ReLU activation function, respectively represent a convolution kernel matrix and a bias vector of the expansion convolution layer, represents the thermodynamic flow input vector.

[0045] It can be understood that the expansion convolution of the expansion convolution layer can better capture long-range thermal conduction.

[0046] Specifically, the thermal delay compensation layer can be represented as: ; In the formula, respectively represent the thermodynamic time sequence features of the current t and the historical t-1. represents a thermodynamic characteristic, represents a thermal inertia compensation factor, preferably 0.5.

[0047] Therefore, the thermal delay compensation layer can compensate for the delay of heat propagation.

[0048] Further, the loading head resonance suppression sub-model comprises a differential feature extraction layer and a resonance suppression operation layer, and the resonance suppression unit comprises: a differential enhancement extraction sub-unit, configured to reduce fixed fluctuations of the loading head motion input vector through differential operation to generate loading head timing features; a resonance suppression sub-unit, configured to pass the mechanical timing features through the resonance suppression operation layer to generate the loading head features.

[0049] In particular, the resonance suppression operation layer can actively cancel the resonance energy through an amplitude attenuation algorithm to avoid load output distortion caused by resonance, and ensure that the actual application value of the mechanical load parameter is consistent with the set value.

[0050] Specifically, the loading head motion input vector further comprises at least wherein D represents the position of the loading head, represents the loading head position set value, V represents the loading head speed, and F represents the force sequence, represents the cable material cooling rate, and k represents the cable material stiffness, represents the force displacement sequence of the loading head, represents the acceleration of the loading head.

[0051] Specifically, the differential operation can be represented as: ; In the formula, represents the primary feature, represents the ReLU activation function, represents the loading head motion input vector, respectively represent the convolution weight matrix and the bias, represents the first-order difference, represents the difference between adjacent feature points along the time dimension for the primary feature, represents the difference between adjacent feature points along the time dimension for the first-order difference, see the form of the loading head motion input vector, the difference represents t2-t1, t3-t2, … for the time dimension, represents the second-order difference.

[0052] Specifically, the resonance suppression operation layer can be represented as: ; In the formula, represents a fast Fourier transform (RFT) that converts a second-order differential time-domain signal into a frequency-domain representation, represents taking a modulus M of the frequency-domain signal, represents locating a maximum value point in the amplitude spectrum, corresponding to the main resonance frequency of the signal , represents a point multiplication operation, represents a Dirac function, represents an attenuation coefficient, preferably 0.85, represents the suppressed frequency spectrum, represents an inverse Fourier transform (IRFT) that restores the suppressed frequency spectrum to the time-domain signal of the loading head feature .

[0053] Specifically, the gating fusion layer can be represented as: ; In the formula, represents a gating weight, represents a one-dimensional convolution operation, represents a thermodynamic feature, represents a loading head feature, represents a Sigmoid activation function, represents a comprehensive physical feature.

[0054] Further, the convolution pooling layer includes a first convolution pooling layer, a second convolution pooling layer, and a flattening layer, and a convolution kernel size of the first convolution pooling layer is smaller than a convolution kernel size of the second convolution pooling layer, and the convolution unit includes: a first convolution pooling subunit, configured to pass the basic input vector through the first convolution pooling layer to generate a local time sequence feature; a second convolution pooling subunit, configured to pass the local time sequence feature through the second convolution pooling layer to generate a global time sequence feature; a flattening layer subunit, configured to pass the global feature through the flattening layer to generate the comprehensive time sequence feature.

[0055] In particular, for the multi-time scale characteristics of the load and temperature parameters in the cable test, such as millisecond-level load impact and minute-level temperature conduction, the small kernel convolution quickly responds to high-frequency dynamics, and the large kernel convolution captures slow-changing trends, and the two work together to enable the model to maintain high-precision feature extraction capability in both transient and steady-state working conditions, providing comprehensive time sequence basis for control parameter generation.

[0056] Specifically, the first convolutional pooling layer adopts 32 convolution kernels, a convolution size of 5, a step size of 1, a ReLU activation function, and a MaxPooling pooling size of 2. The second convolutional layer adopts 64 convolution kernels, a convolution size of 3, a step size of 1, a ReLU activation function, and a MaxPooling pooling size of 2. The flatten layer is used to convert the above multi-dimensional global time sequence features into a one-dimensional vector to adapt to the subsequent fully connected layer processing. The fully connected layer is set to 128 nodes to map the features of the integrated time sequence features and the integrated physical features.

[0057] Further, the control parameters include temperature control parameters and force control parameters, and the constraint unit includes: a thermodynamic constraint subunit for thermodynamically constraining the temperature control parameters based on the temperature parameters and the maximum thermal energy of the heater; a load head constraint subunit for force-constraining the force control parameters based on the cable material stiffness and the maximum stress of the load head.

[0058] Specifically, the thermodynamic constraint can be expressed as: ; In the formula, respectively represent the i control parameter and the d control parameter of the temperature T control parameter of the temperature PID controller, respectively represent the i control parameter and the d control parameter of the temperature T control in the original control parameter, represents the maximum thermal energy of the heater, wherein represents the maximum temperature allowed by the heater, represents the temperature parameter, represents the maximum heat dissipation power available for the test platform, represents the time step of the control cycle, represents the minimum action threshold.

[0059] Specifically, the force constraint can be expressed as: ; In the formula, respectively represent the p control parameter of the displacement D control parameter of the load head displacement controller and the d control parameter of the force F control parameter of the load head force controller, represents the p control parameter of the force displacement D control and the d control parameter of the force F parameter size control in the original control parameter, represents the maximum force of the load head, represents the maximum displacement of the load head, and k represents the cable material stiffness, represents the minimum speed threshold required by the current set working condition.

[0060] More specifically, the control module determines the process of executing temperature as: controlling the infrared heating plate of the heater by heater_power = min(0.6*P + 0.4*I, MAX_HEATER_POWER), controlling the semiconductor refrigerating sheet of the heater by tec_power = min(0.4*abs(P) + 0.6*abs(I), MAX_TEC_POWER), and controlling the convection fan of the heater by fan_speed = int(MAX_FAN_SPEED*(0.3 + 0.7*abs(D) / MAX_DERIVATIVE)), wherein MAX_HEATER_POWER represents the maximum power of the infrared heating plate, MAX_TEC_POWER represents the maximum power of the semiconductor refrigerating sheet, MAX_FAN_SPEED represents the maximum rotation speed of the convection fan, and MAX_DERIVATIVE represents the maximum acceleration of the convection fan, wherein is a set value, e_T is an error of a set working condition temperature and an actual temperature, respectively represent an i control parameter and a d control parameter of a temperature control parameter of a temperature PID controller, and integral and derivative are respectively an integral and a differential of the error. The control of the multi-degree-of-freedom loading head is determined in the same way or based on an existing PID gimbal attitude solving algorithm and a PID hydraulic cylinder force control algorithm, and will not be described here.

[0061] Especially, the thermodynamic compensation mechanism and the loading head resonance suppression sub-model respectively learn the temperature field hysteresis effect and the mechanical system resonance interference, and the gating fusion layer realizes adaptive coupling of physical field features through dynamic weight distribution. This hierarchical learning architecture enables the BP neural network to accurately fit the strong non-linear relationship in extreme scenarios, reduces the adjustment error of the control parameters, and realizes stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater corresponding to complex extreme working conditions of the test platform.

[0062] Specifically, the full connection layer can be represented as: ; In the formula, y represents an original control parameter, ReLU represents a ReLU activation function, respectively represent a learnable convolution matrix and a bias vector of a convolution full connection layer, and x represents a splicing vector of the integrated time sequence features and the integrated physical features.

[0063] As Figure 4 shown, further, the neural network back propagation sub-module comprises: a loss function construction unit, configured to construct the test platform loss function based on a control parameter error term and a control parameter change rate; a back propagation unit configured to perform back propagation adjustment of the control parameter fitting model in a control parameter direction, a convolution direction, a thermodynamic direction, and a resonance suppression direction in sequence based on the test stationary loss function.

[0064] In particular, the ordered back propagation path of the control parameter, the convolution, the thermodynamic, and the resonance suppression is adopted to realize directional gradient transmission from the output layer to the physical field sub-module: the control parameter direction is used to preferentially correct the mapping weight of the full connection layer, to ensure that the deviation of the output parameter from the set value converges rapidly; the convolution direction is used to specifically optimize the convolution kernel parameter of the time sequence feature extraction, to improve the modeling accuracy of the dynamic load temperature sequence; the thermodynamic and resonance suppression directions are used to respectively adjust the internal parameters of the compensation mechanism and the suppression model, to avoid gradient interference across the physical field, and to make the optimization more specific.

[0065] Specifically, the test stationary loss function can be expressed as: ; In the formula, the test stationary loss function is represented by L, and a and b respectively represent two weighting coefficients, the time t weighted control parameter absolute error is represented by e, the control parameter error term of the first three periods is represented by e1, the control parameter change rate term of the comprehensive fluctuation amplitude of the minimum control amount of the L2 norm calculated for the heater power change, the loading head displacement rate, and the loading head acting force change rate is represented by e2.

[0066] Specifically, the control parameter direction is , and the convolution direction is .

[0067] Specifically, the process of back propagation adjustment of the thermodynamic compensation mechanism can be expressed as , and the process of back propagation adjustment of the loading head resonance suppression sub-model can be expressed as , the mechanical flow gradient filter is represented by H, the weight matrix of the loading head resonance suppression sub-model is represented by W, the cutoff frequency (typical value 100 Hz) is represented by f, and high-frequency noise is suppressed. The above formula represents the solving process of the loss function gradient value.

[0068] In particular, through real-time feedback of the load parameter sequence, the network weight is dynamically corrected along the gradient path in the control parameter direction, the convolution direction, the thermodynamic direction, and the resonance suppression direction. Compared with the traditional single-direction gradient descent, the parameter convergence speed under extreme working conditions is improved.

[0069] In this embodiment, the temperature and multi-degree-of-freedom mechanical load are applied cooperatively through the cable test module, which can simulate the complex extreme working conditions of temperature mechanics that the cable faces in actual operation. Through the control parameter fitting model based on the BP neural network architecture, the nonlinear relationship between the temperature and the load parameters can be accurately captured, so that the multifunctional test platform does not need to rely on artificial preset models and can autonomously learn the control parameters corresponding to complex extreme working conditions, thereby improving the adaptability of the test platform to complex working condition tests and realizing stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the platform. Through the thermodynamic compensation mechanism and the loading head resonance suppression sub-model, the temperature field hysteresis effect and mechanical system resonance interference are respectively learned. The gating fusion layer realizes adaptive coupling of physical field characteristics through dynamic weight distribution. This hierarchical learning architecture enables the BP neural network to accurately fit the strong nonlinear relationship in extreme scenarios and reduce the adjustment error of the control parameters, so as to realize stable and accurate parameter control of the multi-degree-of-freedom loading head and the heater of the test platform corresponding to complex extreme working conditions. Through real-time feedback of the load parameter sequence, the network weight is dynamically corrected along the gradient path in the control parameter direction, the convolution direction, the thermodynamic direction, and the resonance suppression direction. Compared with the traditional single-direction gradient descent, the parameter convergence speed in extreme working conditions is improved.

[0070] Those skilled in the art can appreciate that the modules and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0071] So far, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the drawings, but those skilled in the art can easily understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to related technical features without departing from the principles of the present application, and the technical solutions after such changes or replacements will fall within the protection scope of the present application.

[0072] The above description is only the preferred embodiments of the present application and is not intended to limit the present application; for those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A multifunctional integrated cable simulation working condition test platform, characterized in that: include: Cable testing module, used to collect temperature parameters and load parameters of the test cable when the heater and multi-degree-of-freedom loading head are tested; A neural network module is used to determine the control parameters by fitting the temperature parameters and the load parameters into a control parameter fitting model based on a BP neural network architecture, and to perform back propagation adjustment of the model parameters of the control parameter fitting model by testing a stationary loss function; The control module is used to generate the execution temperature of the heater and the execution force of the multi-degree-of-freedom loading head through a control algorithm according to the control parameters.

2. The multifunctional integrated cable simulation working condition test platform according to claim 1, characterized in that: The neural network module includes: A neural network forward propagation submodule, configured to generate the control parameters by fitting the temperature parameters and the load parameters through a control parameter fitting model based on a BP neural network architecture; The neural network back propagation submodule is used to adjust the model parameters of the control parameter fitting model by back propagation of the load parameters through the test steady loss function.

3. The multifunctional integrated cable simulation working condition test platform according to claim 2, characterized in that: The load parameters include a force displacement sequence and a force parameter sequence. The control parameter fitting model includes a thermodynamic compensation mechanism, a loading head resonance suppression sub-model, a gated fusion layer, a convolutional pooling layer, a fully connected layer, and a constraint layer. The neural network forward propagation sub-module includes: an input integration unit, configured to integrate the temperature parameter, the force-displacement sequence, the force parameter sequence, the temperature setting value, the displacement setting value, and the force setting value into channels of a basic input vector, generate a thermodynamic input vector based on the basic input vector and the thermal conductivity of the cable material, and generate a loading head motion input vector based on the basic input vector and the stiffness of the cable material; a thermodynamic processing unit for passing the thermodynamic input vector through a thermodynamic compensation mechanism to generate a thermodynamic signature; a resonance suppression unit, configured to pass the loading head motion input vector through a loading head resonance suppression sub-model to generate a loading head feature; a gated fusion unit for passing the thermodynamic characteristics and the loading head characteristics through a gated fusion layer to generate a comprehensive physical characteristic; A convolution unit, configured to pass the basic input vector through a convolutional pooling layer to generate a comprehensive temporal feature; a mapping unit, configured to perform feature mapping on the comprehensive temporal features and the comprehensive physical features through a fully connected layer to generate original control parameters; A constraint unit is used to perform output constraints on the original control parameters through a constraint layer based on the thermal conductivity of the cable material and the stiffness of the cable material, so as to generate the control parameters.

4. The multifunctional integrated cable simulation working condition test platform according to claim 3 is characterized in that: The thermodynamic compensation mechanism includes an expanded convolution layer and a thermal delay compensation layer, and the thermodynamic processing unit includes: a dilated convolution subunit, configured to perform time series extraction on the thermodynamic flow input vector through a dilated convolution layer to generate a thermodynamic time series feature; The thermal delay compensation subunit is used to pass the current thermal time series characteristics and the historical thermal time series characteristics through the thermal delay compensation layer to generate the thermodynamic characteristics.

5. The multifunctional integrated cable simulation working condition test platform according to claim 3 is characterized in that: The loading head resonance suppression sub-model includes a differential feature extraction layer and a resonance suppression operation layer, and the resonance suppression unit includes: A differential enhancement extraction subunit, configured to reduce fixed fluctuations of the loading head motion input vector by differential operation to generate a loading head timing feature; The resonance suppression subunit is used to pass the mechanical timing characteristics through the resonance suppression operation layer to generate the loading head characteristics.

6. The multifunctional integrated cable simulation working condition test platform according to claim 3, characterized in that: The convolution pooling layer includes a first convolution pooling layer, a second convolution pooling layer and a flattening layer, and the convolution kernel size of the first convolution pooling layer is smaller than the convolution kernel size of the second convolution pooling layer. The convolution unit includes: a first convolutional pooling subunit, configured to pass the basic input vector through a first convolutional pooling layer to generate local temporal features; A second convolutional pooling subunit, configured to pass the local temporal features through a second convolutional pooling layer to generate global temporal features; The flattening layer subunit is used to pass the global feature through the flattening layer to generate the comprehensive temporal feature.

7. The multifunctional integrated cable simulation working condition test platform according to claim 3, characterized in that: The control parameters include temperature control parameters and force control parameters, and the constraint unit includes: a thermodynamic constraint subunit, configured to thermodynamically constrain the temperature control parameter based on the temperature parameter and the maximum heat energy of the heater; The loading head constraint subunit is used to perform force constraints on the force control parameters based on the cable material stiffness and the maximum stress of the loading head.

8. The multifunctional integrated cable simulation working condition test platform according to claim 2, characterized in that: The neural network back propagation submodule includes: A loss function construction unit is used to construct the test steady loss function based on the control parameter error term and the control parameter change rate.

9. The multifunctional integrated cable simulation working condition test platform according to claim 2, characterized in that: The neural network back propagation submodule also includes: A back propagation unit is used to perform back propagation adjustments on the control parameter fitting model in a control parameter direction, a convolution direction, a thermodynamic direction and a resonance suppression direction based on the test stationary loss function.

10. The multifunctional integrated cable simulation working condition test platform according to any one of claims 1 to 9, characterized in that: Integrating the heater and the multi-degree-of-freedom loading head in an environmental simulation chamber; The multi-degree-of-freedom loading head includes a hydraulic cylinder array and a universal joint for controlling the hydraulic cylinder array; The heater at least includes a semiconductor refrigeration sheet, an infrared heating plate and a convection fan.

Images