Energy consumption optimization control method and system for hydraulic system of compression-shear testing machine

Abstract

The invention discloses an energy consumption optimization control method and system for a hydraulic system of a compression-shear testing machine, and relates to the technical field of intelligent control. Pressure, flow, temperature and vibration data of the hydraulic system of the compression-shear testing machine are collected in real time through multiple sensors, and a multi-source heterogeneous data flow is generated; inputting the multi-source heterogeneous data flow into a dynamic energy efficiency evaluation model based on deep learning, outputting an energy efficiency ratio and a key energy consumption node of the current system, generating a pressure-flow-temperature collaborative optimization strategy of the hydraulic system through a reinforcement learning algorithm according to the energy efficiency ratio and the key energy consumption node, and evaluating the energy efficiency of the hydraulic system according to the pressure-flow-temperature collaborative optimization strategy. Converting the collaborative optimization strategy into a control instruction of the hydraulic system, dynamically adjusting the output of an oil pump and the opening degree of a valve group, collecting an instruction execution result, and performing adaptive model optimization of a dynamic energy efficiency evaluation model based on the instruction execution result; control parameters can be adaptively adjusted according to the real-time state of the system, and precise control over complex working conditions is achieved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, in particular to an energy consumption optimization control method and system for a hydraulic system of a compression shear testing machine. Background Art

[0002] Compression-shear testing machines are essential equipment for testing the mechanical properties of materials. Their core power comes from a hydraulic system. Energy consumption during hydraulic system operation has long been a focus of industry attention. Traditional hydraulic systems often utilize fixed parameter control methods during operation, failing to adapt to varying operating conditions and loads, resulting in poor energy efficiency. Especially during compression-shear testing, due to the complex and variable test conditions and large load fluctuations, conventional control methods would cause the system to remain in a high-voltage standby state for extended periods, resulting in significant energy waste.

[0003] Hydraulic systems face a variety of complex operating conditions in actual operation, including varying load conditions such as no load, light load, and heavy load, as well as different operating phases such as startup, stable operation, and braking. The optimal operating parameters for each operating condition vary. Traditional control methods struggle to accurately identify and respond to these complex conditions, resulting in suboptimal system efficiency over time. Furthermore, the performance of hydraulic components gradually degrades over time, and system parameters also change, further complicating energy optimization.

[0004] Traditional control methods, however, often rely on fixed-parameter control strategies that are unable to adapt to the system's real-time state. This "one-size-fits-all" approach struggles to maintain optimal system efficiency under complex and changing operating conditions, leading to significant energy waste.

[0005] Therefore, the hydraulic system of the compression-shear testing machine urgently needs an intelligent energy consumption optimization control method that can perceive the system status in real time, dynamically evaluate the energy efficiency level, and adaptively adjust the control parameters according to different working conditions to achieve efficient energy utilization.

[0006] To this end, the present invention proposes an energy consumption optimization control method and system for a hydraulic system of a compression shear testing machine. Summary of the Invention

[0007] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a method and system for optimizing the energy consumption of the hydraulic system of a compression-shear testing machine. The method and system can adaptively adjust control parameters based on the real-time status of the system, achieving precise control under complex working conditions.

[0008] To achieve the above objectives, a method for optimizing the energy consumption of a hydraulic system of a compression shear testing machine is proposed, which includes the following steps: Step 1: Use multiple sensors to collect real-time pressure, flow, temperature, and vibration data from the hydraulic system of the compression and shear testing machine to generate multi-source heterogeneous data streams; Step 2: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, and output the energy efficiency ratio and key energy consumption nodes of the current system; Step 3: Based on the energy efficiency ratio and key energy consumption nodes, a pressure-flow-temperature collaborative optimization strategy of the hydraulic system is generated through a reinforcement learning algorithm; Step 4: Convert the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjust the oil pump output and valve group opening, and collect the instruction execution results; Step 5: Perform adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results.

[0009] Generating multi-source heterogeneous data streams comprises the following steps: Step 11: Using a pressure sensor to collect pressure data of the hydraulic system of the compression and shear testing machine in real time, and generate a pressure data stream; The method of generating the pressure data stream is: The pressure sensor is installed at the key node of the hydraulic system to monitor the working pressure of the hydraulic system in real time; The pressure sensor uses piezoresistive or piezoelectric sensing technology to convert pressure changes in the hydraulic system into an electrical pressure signal. This electrical pressure signal is amplified and filtered by a signal conditioning circuit to eliminate noise interference, and then converted into a corresponding digital signal by an analog-to-digital converter. This digital signal is transmitted to the data processing unit at a fixed sampling frequency, forming a continuous pressure data stream. Each data point in the pressure data stream includes a timestamp and a pressure value. The timestamp is used to mark the precise moment of data collection, and the pressure value is used to characterize the pressure state of the hydraulic system at that moment.

[0010] Step 12: Using a flow sensor to collect flow data of the hydraulic system of the compression and shear testing machine in real time, and generate a flow data stream; The method of generating the traffic data stream is: The flow sensor is installed in the oil circuit of the hydraulic system and is used to monitor the flow rate of the hydraulic oil in real time.

[0011] The flow sensor uses turbine or ultrasonic sensing technology to convert the flow rate of the hydraulic oil into an electrical flow signal. This electrical flow signal is linearized and filtered by a signal conditioning circuit to eliminate noise caused by flow rate fluctuations. It is then converted into a corresponding digital signal by an analog-to-digital converter. This digital signal is transmitted to the data processing unit at a fixed sampling frequency, forming a continuous flow data stream. Each data point in the flow data stream includes a timestamp and a flow value. The timestamp is used to mark the precise moment of data collection, and the flow value is used to characterize the flow state of the hydraulic system at that moment.

[0012] Step 13: Using a temperature sensor to collect temperature data of the hydraulic system of the compression and shear testing machine in real time, and generate a temperature data stream; The method of generating the temperature data stream is: The temperature sensor is installed at a key position of the hydraulic system to monitor the temperature change of the hydraulic oil in real time; The temperature sensor uses thermocouple or thermistor sensing technology to convert hydraulic oil temperature changes into a temperature electrical signal. This temperature electrical signal is also linearized and filtered by a signal conditioning circuit to eliminate noise caused by ambient temperature fluctuations before being converted into a digital signal by an analog-to-digital converter. The digital signal is transmitted to the data processing unit at a fixed sampling frequency to form a continuous temperature data stream; Each data point in the temperature data stream includes a timestamp and a temperature value. The timestamp is used to mark the precise moment of data collection, and the temperature value is used to characterize the temperature state of the hydraulic system at that moment.

[0013] Step 14: Using a vibration sensor to collect vibration data of the hydraulic system of the compression and shear testing machine in real time, and generate a vibration data stream; The method of generating the vibration data stream is: The vibration sensor is installed on the mechanical components of the hydraulic system and is used to monitor the mechanical vibration status of the hydraulic system in real time; The vibration sensor uses piezoelectric or acceleration sensing technology to convert mechanical vibration into a vibration electrical signal; the vibration electrical signal is amplified and filtered by a signal conditioning circuit to eliminate high-frequency noise interference, and then converted into a corresponding digital signal by an analog-to-digital converter. The digital signal is transmitted to a data processing unit at a fixed sampling frequency to form a continuous vibration data stream; Each data point in the vibration data stream includes a timestamp and a vibration amplitude. The timestamp is used to mark the precise moment of data collection, and the vibration amplitude is used to characterize the vibration state of the hydraulic system at that moment.

[0014] Step 15: Integrate the pressure data stream, the flow data stream, the temperature data stream, and the vibration data stream into a multi-source heterogeneous data stream; The input is fed into a dynamic energy efficiency evaluation model based on deep learning, and outputting the energy efficiency ratio and key energy consumption nodes of the current system includes the following steps: Step 21: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, extract the dynamic features of the pressure, flow, temperature and vibration data through the feature extraction layer of the model, and generate a dynamic feature vector; The dynamic energy efficiency evaluation model adopts a hybrid architecture combining convolutional neural network and long short-term memory network; The feature extraction layer consists of multiple convolutional layers and pooling layers, and each convolutional layer uses a different convolution kernel size; During the feature extraction process, the pressure data is first normalized. The normalized pressure data and flow data are then feature mapped using a convolutional layer to generate a preliminary feature map. The temperature and vibration data are then processed using separate convolutional layers to extract their dynamic features. Subsequently, the feature maps and dynamic features are reduced in dimension through a pooling layer. Finally, all feature maps and dynamic features are aligned in the time dimension and time series modeled through an LSTM layer to generate a dynamic feature vector. Step 22: Input the dynamic feature vector into the fully connected layer of the dynamic energy efficiency evaluation model, calculate the energy efficiency ratio of the current system and the weight distribution of key energy consumption nodes, and generate energy efficiency ratio and key energy consumption node data; The fully connected layer consists of multiple hidden layers, each hidden layer uses the ReLU activation function, and the number of neurons in the input layer is consistent with the dimension of the dynamic feature vector; When calculating the energy efficiency ratio (EER), the fully connected layer first maps the dynamic feature vector to a low-dimensional space to generate a preliminary estimate of the EER. The back-propagation algorithm then optimizes the weight parameters to minimize the error between the predicted EER and the actual EER. The EER is defined as the ratio of the hydraulic system's effective output power to its input power. The weight distribution of the key energy consumption nodes is achieved through an attention mechanism, which assigns different weights to each time step in the dynamic feature vector to identify the operation node with the greatest impact on energy consumption; Step 23: Outputting the energy efficiency ratio and key energy consumption node data as the evaluation result of the dynamic energy efficiency evaluation model; The method for generating a pressure-flow-temperature collaborative optimization strategy for a hydraulic system using a reinforcement learning algorithm includes the following steps: Step 31: Input the energy efficiency ratio and key energy consumption nodes into the reinforcement learning algorithm as basic data of the state space; Step 32: Generate the adjustable ranges of pressure, flow, and temperature of the hydraulic system through the action space definition module in the reinforcement learning algorithm; Step 33: Based on the state space and action space, iteratively calculate the pressure-flow-temperature collaborative parameter combination through the strategy optimization module of the reinforcement learning algorithm; Step 34: Outputting the pressure-flow-temperature collaborative parameter combination as a pressure-flow-temperature collaborative optimization strategy for the hydraulic system; the collaborative optimization strategy includes a pressure adjustment value, a flow adjustment value, and a temperature adjustment value; The dynamic adjustment of the oil pump output and the valve group opening comprises the following steps: Step 41: parsing the pressure-flow-temperature collaborative optimization strategy into an executable control instruction sequence through a control instruction conversion module; The parsing process of the control instruction conversion module is as follows: The pressure adjustment value in the collaborative optimization strategy is first mapped to a corresponding motor speed command value through an oil pump characteristic curve, and the oil pump characteristic curve mapping is pre-established based on the oil pump's displacement-pressure-speed characteristic curve; The flow adjustment value is converted into a corresponding valve core displacement instruction value through the flow characteristic equation of the valve group of the proportional valve; The temperature adjustment value is converted into a regulation instruction of the cooler fan speed or heater power according to the system heat balance equation; Step 42: Sending the control instruction sequence to the actuator of the hydraulic system via the real-time control bus; The method of collecting the instruction execution results is: Real-time recording of the actual energy efficiency ratio of the motor after the control instruction is executed, forming the instruction execution result; The adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution result includes the following steps: Step 51: Compare and analyze the instruction execution result with the expected optimization target to calculate the actual energy efficiency improvement rate; Step 52: Based on the actual energy efficiency improvement rate, update the parameter weights of the dynamic energy efficiency evaluation model; Step 53: Apply the updated model parameters to the next round of energy efficiency evaluation.

[0015] An energy consumption optimization control system for the hydraulic system of a compression shear testing machine is proposed, which includes a data stream collection module, an evaluation model construction module, a collaborative strategy generation module, and an execution feedback module. The modules are electrically connected to each other. The data stream collection module collects the pressure, flow, temperature and vibration data of the hydraulic system of the compression and shear testing machine in real time through multiple sensors, generates multi-source heterogeneous data streams, and sends the multi-source heterogeneous data streams to the evaluation model construction module; An evaluation model construction module inputs the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, outputs the energy efficiency ratio and key energy consumption nodes of the current system, and sends the energy efficiency ratio and key energy consumption nodes to a collaborative strategy generation module; A collaborative strategy generation module generates a pressure-flow-temperature collaborative optimization strategy for the hydraulic system based on the energy efficiency ratio and key energy consumption nodes through a reinforcement learning algorithm, and sends the collaborative optimization strategy to the execution feedback module; The execution feedback module converts the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjusts the oil pump output and valve group opening, collects the instruction execution results, performs adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results, and returns the optimized dynamic energy efficiency evaluation model to the evaluation model construction module.

[0016] Compared with the prior art, the present invention has the following beneficial effects: A method for optimizing the energy consumption of a hydraulic system for a compression-shear testing machine based on a dynamic energy efficiency evaluation model and reinforcement learning. This method first uses multiple sensors to collect system operating data, including parameters such as pressure, flow, temperature, and displacement, to construct a system state vector. Then, based on the collected data, a dynamic energy efficiency evaluation model is established. This model comprehensively considers energy conversion efficiency, energy loss rate, and system response performance to evaluate the system's energy efficiency status in real time. This model can comprehensively reflect the system's energy efficiency status and provide an accurate evaluation basis for optimized control. Based on the dynamic energy efficiency evaluation, a deep reinforcement learning method is further used to construct an intelligent control model. This model takes the energy efficiency status as input, learns the dynamic characteristics of the system through a deep neural network, and outputs an appropriate control strategy. This allows the control parameters to be adaptively adjusted according to the system's real-time status, achieving precise control of complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the energy consumption optimization control method for the hydraulic system of the pressure shear testing machine in Example 1 of the present invention; Figure 2 This is a model structure diagram of the dynamic energy efficiency evaluation model in Example 1 of the present invention; Figure 3 This is a module connection diagram of the energy consumption optimization control system of the hydraulic system of the medium-pressure shear testing machine in Example 2 of the present invention. DETAILED DESCRIPTION

[0018] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] Example 1

[0020] like Figure 1 As shown, the energy consumption optimization control method of the hydraulic system of the compression shear testing machine includes the following steps: Step 1: Use multiple sensors to collect real-time pressure, flow, temperature, and vibration data from the hydraulic system of the compression and shear testing machine to generate multi-source heterogeneous data streams; Step 2: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, and output the energy efficiency ratio and key energy consumption nodes of the current system; Step 3: Based on the energy efficiency ratio and key energy consumption nodes, a pressure-flow-temperature collaborative optimization strategy of the hydraulic system is generated through a reinforcement learning algorithm; Step 4: Convert the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjust the oil pump output and valve group opening, and collect the instruction execution results; Step 5: Perform adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results.

[0021] In an embodiment of the present invention, generating a multi-source heterogeneous data stream includes the following steps: Step 11: Using a pressure sensor to collect pressure data of the hydraulic system of the compression and shear testing machine in real time, and generate a pressure data stream; Specifically, the method of generating the pressure data stream is: The pressure sensor is installed at a key node of the hydraulic system, such as the oil pump outlet, valve group inlet or actuator inlet, and is used to monitor the working pressure of the hydraulic system in real time.

[0022] The pressure sensor uses piezoresistive or piezoelectric sensing technology to convert pressure changes in the hydraulic system into an electrical pressure signal. This pressure signal is amplified and filtered by a signal conditioning circuit to eliminate noise interference. It is then converted into a corresponding digital signal by an analog-to-digital converter. This digital signal is transmitted to a data processing unit at a fixed sampling frequency, forming a continuous pressure data stream.

[0023] Each data point in the pressure data stream includes a timestamp and a pressure value. The timestamp is used to mark the precise moment of data collection, and the pressure value is used to characterize the pressure state of the hydraulic system at that moment.

[0024] Step 12: Using a flow sensor to collect flow data of the hydraulic system of the compression and shear testing machine in real time, and generate a flow data stream; Specifically, the method of generating the traffic data stream is: The flow sensor is installed in the oil circuit of the hydraulic system, such as the oil pump outlet or the actuator inlet, and is used to monitor the flow rate of the hydraulic oil in real time.

[0025] The flow sensor uses turbine or ultrasonic sensing technology to convert the hydraulic oil flow rate into an electrical flow rate signal. This signal is linearized and filtered by a signal conditioning circuit to eliminate noise caused by flow rate fluctuations. It is then converted into a corresponding digital signal by an analog-to-digital converter. This digital signal is transmitted to a data processing unit at a fixed sampling frequency, forming a continuous flow rate data stream.

[0026] Each data point in the flow data stream includes a timestamp and a flow value. The timestamp is used to mark the precise moment of data collection, and the flow value is used to characterize the flow state of the hydraulic system at that moment.

[0027] Step 13: Using a temperature sensor to collect temperature data of the hydraulic system of the compression and shear testing machine in real time, and generate a temperature data stream; Specifically, the method of generating the temperature data stream is: The temperature sensor is installed at a key position of the hydraulic system, such as inside the oil tank or at the oil pump outlet, to monitor the temperature change of the hydraulic oil in real time.

[0028] The temperature sensor uses thermocouple or thermistor sensing technology to convert hydraulic oil temperature changes into an electrical temperature signal. This signal is also linearized and filtered by a signal conditioning circuit to eliminate noise caused by ambient temperature fluctuations before being converted into a digital signal via an analog-to-digital converter.

[0029] The digital signal is transmitted to the data processing unit at a fixed sampling frequency to form a continuous temperature data stream.

[0030] Each data point in the temperature data stream includes a timestamp and a temperature value. The timestamp is used to mark the precise moment of data collection, and the temperature value is used to characterize the temperature state of the hydraulic system at that moment.

[0031] Step 14: Using a vibration sensor to collect vibration data of the hydraulic system of the compression and shear testing machine in real time, and generate a vibration data stream; Specifically, the vibration data stream is generated in the following manner: Specifically, the vibration sensor is installed on a mechanical component of the hydraulic system, such as an oil pump housing or a valve group bracket, and is used to monitor the mechanical vibration state of the hydraulic system in real time.

[0032] The vibration sensor uses piezoelectric or acceleration sensing technology to convert mechanical vibrations into electrical vibration signals. These signals are amplified and filtered by a signal conditioning circuit to eliminate high-frequency noise interference. Then, an analog-to-digital converter converts them into digital signals. These digital signals are transmitted to a data processing unit at a fixed sampling frequency, forming a continuous vibration data stream.

[0033] Each data point in the vibration data stream includes a timestamp and a vibration amplitude. The timestamp is used to mark the precise moment of data collection, and the vibration amplitude is used to characterize the vibration state of the hydraulic system at that moment.

[0034] Step 15: Integrate the pressure data stream, the flow data stream, the temperature data stream, and the vibration data stream into a multi-source heterogeneous data stream; Specifically, the integration process of the multi-source heterogeneous data stream is as follows: First, each data stream is time-aligned to ensure that the timestamps of all data points are consistent. The data integration module then merges the aligned data streams in time sequence to form a unified data structure. Each data point in the multi-source heterogeneous data stream contains a timestamp, pressure value, flow value, temperature value, and vibration amplitude, which are used to fully characterize the working status of the hydraulic system at that moment. Furthermore, the input to the dynamic energy efficiency evaluation model based on deep learning and the output of the energy efficiency ratio and key energy consumption nodes of the current system include the following steps: Step 21: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, extract the dynamic features of the pressure, flow, temperature and vibration data through the feature extraction layer of the model, and generate a dynamic feature vector; Specifically, if Figure 2 As shown, the dynamic energy efficiency evaluation model adopts a hybrid architecture combining convolutional neural network and long short-term memory network; The CNN part is used to extract local features of the data, while the LSTM part captures long-term dependencies in the time series; The feature extraction layer consists of multiple convolutional and pooling layers, each of which uses a different kernel size to cover dynamic changes at different time scales. For example, smaller kernels are used to capture subtle changes in high-frequency vibration signals, while larger kernels are used to analyze slow fluctuations in pressure and flow.

[0035] During the feature extraction process, the pressure data is first normalized to eliminate dimensional differences. The normalized pressure data and flow data are then mapped using a convolutional layer to generate a preliminary feature map. The temperature and vibration data are then processed using separate convolutional layers to extract their unique dynamic features. Subsequently, the feature maps and dynamic features are reduced in dimension through a pooling layer to reduce computational complexity. Finally, all feature maps and dynamic features are aligned in the time dimension and modeled as time series through an LSTM layer to generate a dynamic feature vector.

[0036] It can be understood that the dynamic feature vector is a high-dimensional vector, which includes the comprehensive dynamic characteristics of the hydraulic system within a specific time window.

[0037] Step 22: Input the dynamic feature vector into the fully connected layer of the dynamic energy efficiency evaluation model, calculate the energy efficiency ratio of the current system and the weight distribution of key energy consumption nodes, and generate energy efficiency ratio and key energy consumption node data; Specifically, the fully connected layer consists of multiple hidden layers, each hidden layer uses the ReLU activation function to avoid the gradient vanishing problem, and the number of neurons in the input layer is consistent with the dimension of the dynamic feature vector to ensure that all feature information is fully transmitted.

[0038] When calculating the energy efficiency ratio (EER), the fully connected layer first maps the dynamic feature vector to a low-dimensional space to generate a preliminary estimate of the EER. The back-propagation algorithm then optimizes the weight parameters to minimize the error between the predicted and actual EER values. The EER is defined as the ratio of the hydraulic system's effective output power to its input power, reflecting the system's energy efficiency. The weight distribution of the key energy consumption nodes is achieved through an attention mechanism. The attention mechanism assigns different weights to each time step in the dynamic feature vector to identify the operation node with the greatest impact on energy consumption. For example, when the pressure of the hydraulic system fluctuates greatly, the attention mechanism will assign a higher weight to the pressure data, indicating that pressure regulation is the key node of the current energy consumption. The final energy efficiency ratio and key energy consumption node data are output in a structured format, including the energy efficiency ratio value and the weight distribution matrix of the key nodes; Step 23: Outputting the energy efficiency ratio and key energy consumption node data as the evaluation result of the dynamic energy efficiency evaluation model; Specifically, the evaluation results are stored in JSON format for subsequent module parsing and processing. The energy efficiency ratio value is expressed in floating point form and is accurate to two decimal places to ensure accuracy. The key energy consumption node data includes node name, weight value and timestamp information to identify energy consumption hotspots and their duration.

[0039] Furthermore, the generation of a pressure-flow-temperature collaborative optimization strategy for a hydraulic system by a reinforcement learning algorithm includes the following steps: Step 31: Input the energy efficiency ratio and key energy consumption nodes into the reinforcement learning algorithm as basic data of the state space; Specifically, the state space is a core concept in the reinforcement learning algorithm, which is used to describe the current state characteristics of the system.

[0040] In the specific implementation of the present invention, the specific construction process of the state space includes: The energy efficiency ratio (EER) output by the dynamic energy efficiency assessment model is a quantitative indicator of the hydraulic system's current energy efficiency. A higher EER value indicates a more efficient system. Key energy consumption node data identifies specific components or operating parameters in the hydraulic system that consume significant energy, such as the oil pump, valve block, or specific temperature range.

[0041] When the energy efficiency ratio and key energy consumption nodes are input into the reinforcement learning algorithm, the state space is defined to include the key operating parameters of the hydraulic system. Specifically, the state space consists of the following dimensions: energy efficiency ratio, weight distribution of key energy consumption nodes, current pressure value, current flow value, and current temperature value.

[0042] The data in each dimension is normalized to ensure that parameters of different dimensions can be compared and calculated on the same scale. The normalization process uses a linear transformation method to map the original data to the interval [0,1].

[0043] In this embodiment of the present invention, the state space construction also considers the hydraulic system's historical operating data. By introducing time series analysis techniques, the state space incorporates the energy efficiency ratio variation trends and the dynamic distribution of key energy consumption nodes within several recent time windows. This allows the reinforcement learning algorithm to capture the dynamic characteristics of the hydraulic system, thereby more accurately assessing the impact of the current state on the hydraulic system's energy efficiency.

[0044] Step 32: Generate the adjustable ranges of pressure, flow, and temperature of the hydraulic system through the action space definition module in the reinforcement learning algorithm; Specifically, action space is another core concept in reinforcement learning algorithms, which is used to describe the actions or adjustments that a system can take.

[0045] In an embodiment of the present invention, the specific process of constructing the action space includes: The adjustable ranges of pressure, flow, and temperature are determined based on the design parameters and operating limitations of the hydraulic system. For example, the pressure adjustment range is limited by the maximum output capacity of the oil pump, the flow adjustment range is limited by the valve group opening, and the temperature adjustment range is limited by the cooling system performance. The adjustable ranges are predetermined based on the above limitations through experimental or simulation data and serve as boundary conditions for the action space.

[0046] Each action in the action space corresponds to a specific set of parameter adjustments. For example, an action might be "increase pressure by a certain percentage, decrease flow by a certain percentage, and maintain temperature constant." Combinations of these actions can be represented in either a discrete or continuous manner. A discrete action space divides the adjustment amount for each parameter into several levels, while a continuous action space allows parameters to take any value within a certain range.

[0047] In embodiments of the present invention, the design of the action space further considers the dynamic response characteristics of the system. Specifically, by incorporating a dynamic model of the hydraulic system, each action in the action space is assigned a predicted response time. For example, adjusting pressure may take time to reach the target value, while adjusting flow rate may respond more quickly. This allows the reinforcement learning algorithm to more rationally plan action sequences and avoid system instability caused by frequent adjustments.

[0048] Step 33: Based on the state space and action space, iteratively calculate the pressure-flow-temperature collaborative parameter combination through the strategy optimization module of the reinforcement learning algorithm; Specifically, the policy optimization module is the core part of the reinforcement learning algorithm, which is used to learn the optimal mapping relationship from state to action.

[0049] In an embodiment of the present invention, the policy optimization module adopts a deep deterministic policy gradient algorithm, which combines a deep neural network and a deterministic policy gradient method and is applicable to problems in continuous action space.

[0050] Specifically, the specific implementation process of the strategy optimization module includes: Construct an actor network and a critic network to form a strategy optimization module; The actor network is responsible for generating actions based on the current state, while the critic network evaluates the value of the action. Both the actor network and the critic network adopt a multi-layer perceptron structure, with the input layer receiving state space data and the output layer generating actions or value evaluations.

[0051] During training, the policy optimization module collects empirical data through interactions with the environment, namely the hydraulic system. Each interaction includes the current state, the action taken, the reward obtained, and the next state. In embodiments of the present invention, the reward obtained is represented by a reward function. For example, this reward function can be used to determine positive rewards for improved energy efficiency or reduced energy consumption at key energy-consuming nodes, while negative rewards are awarded for system instability or exceeding operating limits.

[0052] The policy optimization module iteratively updates the parameters of the actor and critic networks using gradient descent. Each update is based on a batch of empirical data, calculating the loss function and backpropagating the gradients. After multiple iterations, the actor network generates a near-optimal sequence of appropriate actions, thereby optimizing the pressure-flow-temperature synergistic parameter combination. Step 34: Outputting the pressure-flow-temperature collaborative parameter combination as a pressure-flow-temperature collaborative optimization strategy for the hydraulic system; It is understood that the collaborative optimization strategy includes the adjustment values ​​of pressure, flow rate and temperature output by the reinforcement learning algorithm; Furthermore, the dynamic adjustment of the oil pump output and the valve group opening comprises the following steps: Step 41: parsing the pressure-flow-temperature collaborative optimization strategy into an executable control instruction sequence through a control instruction conversion module; Specifically, the parsing process of the control instruction conversion module is: The pressure adjustment value in the collaborative optimization strategy is first mapped to a corresponding motor speed command value through an oil pump characteristic curve, and the oil pump characteristic curve mapping is pre-established based on the oil pump's displacement-pressure-speed characteristic curve; The flow adjustment value is converted into a corresponding valve core displacement instruction value through the flow characteristic equation of the valve group of the proportional valve; The temperature adjustment value is converted into a regulation instruction of the cooler fan speed or heater power according to the system heat balance equation; Step 42: Sending the control instruction sequence to the actuator of the hydraulic system via the real-time control bus; The method of collecting the instruction execution results is: Real-time recording of the actual energy efficiency ratio of the motor after the control instruction is executed, forming the instruction execution result; Furthermore, the adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution result includes the following steps: Step 51: Compare and analyze the instruction execution result with the expected optimization target to calculate the actual energy efficiency improvement rate; Specifically, the comparative analysis method is: Pair and compare the energy efficiency data before and after instruction execution to calculate the actual energy efficiency improvement percentage; The actual energy efficiency improvement rate is calculated by the following formula: Energy efficiency improvement rate = (energy efficiency ratio after execution - energy efficiency ratio before execution) / energy efficiency ratio before execution × 100%, to reflect the actual effect of the collaborative optimization strategy.

[0053] Step 52: Based on the actual energy efficiency improvement rate, update the parameter weights of the dynamic energy efficiency evaluation model; Specifically, the method of updating parameter weights is: A gradient descent method is used to fine-tune the parameters of the dynamic energy efficiency evaluation model. When the actual energy efficiency improvement rate is lower than expected, the weights of the corresponding key energy consumption node characteristics are increased by a preset ratio, so that the model pays more attention to these nodes in subsequent evaluations. When the control deviation is large, the weights of the corresponding dynamic characteristics in the model are adjusted to improve the model's prediction accuracy of the system's dynamic response.

[0054] The parameter update process uses an adaptive learning rate strategy, which dynamically adjusts the learning rate based on historical update results. For well-performing parameter regions, a smaller learning rate is used for fine-tuning; for poorly performing regions, a larger learning rate is used to accelerate convergence.

[0055] In an embodiment of the present invention, parameter updating also introduces regularization technology, which prevents the model from overfitting to specific working conditions and maintains the generalization ability of the model by adding an L2 regularization term to the loss function.

[0056] Step 53: Apply the updated model parameters to the next round of energy efficiency evaluation; Specifically, the application process is: The updated model parameters are loaded into the dynamic energy efficiency evaluation model. During the next round of operation, the system will perform energy efficiency evaluation and optimization control based on these updated parameters and functions, forming a closed-loop adaptive optimization mechanism.

[0057] Example 2

[0058] like Figure 3 As shown, the energy consumption optimization control system of the hydraulic system of the compression and shear testing machine includes a data stream collection module, an evaluation model construction module, a collaborative strategy generation module, and an execution feedback module; wherein each module is electrically connected; The data stream collection module collects the pressure, flow, temperature and vibration data of the hydraulic system of the compression and shear testing machine in real time through multiple sensors, generates multi-source heterogeneous data streams, and sends the multi-source heterogeneous data streams to the evaluation model construction module; An evaluation model construction module inputs the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, outputs the energy efficiency ratio and key energy consumption nodes of the current system, and sends the energy efficiency ratio and key energy consumption nodes to a collaborative strategy generation module; A collaborative strategy generation module generates a pressure-flow-temperature collaborative optimization strategy for the hydraulic system based on the energy efficiency ratio and key energy consumption nodes through a reinforcement learning algorithm, and sends the collaborative optimization strategy to the execution feedback module; The execution feedback module converts the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjusts the oil pump output and valve group opening, collects the instruction execution results, performs adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results, and returns the optimized dynamic energy efficiency evaluation model to the evaluation model construction module.

[0059] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0060] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0061] The above preset parameters or preset thresholds are all set by those skilled in the art according to actual conditions or obtained through large amounts of data simulation.

[0062] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine is characterized in that: The following steps are involved: Step 1: Use multiple sensors to collect real-time pressure, flow, temperature, and vibration data from the hydraulic system of the compression and shear testing machine to generate multi-source heterogeneous data streams; Step 2: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, and output the energy efficiency ratio and key energy consumption nodes of the current system; Step 3: Based on the energy efficiency ratio and key energy consumption nodes, a pressure-flow-temperature collaborative optimization strategy of the hydraulic system is generated through a reinforcement learning algorithm; Step 4: Convert the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjust the oil pump output and valve group opening, and collect the instruction execution results; Step 5: Perform adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results.

2. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 1 is characterized in that: Generating multi-source heterogeneous data streams comprises the following steps: Step 11: Using a pressure sensor to collect pressure data of the hydraulic system of the compression and shear testing machine in real time, and generate a pressure data stream; Step 12: Using a flow sensor to collect flow data of the hydraulic system of the compression and shear testing machine in real time, and generate a flow data stream; Step 13: Using a temperature sensor to collect temperature data of the hydraulic system of the compression and shear testing machine in real time, and generate a temperature data stream; Step 14: Using a vibration sensor to collect vibration data of the hydraulic system of the compression and shear testing machine in real time, and generate a vibration data stream; Step 15: Integrate the pressure data stream, the flow data stream, the temperature data stream, and the vibration data stream into a multi-source heterogeneous data stream.

3. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 2 is characterized in that: The input is fed into a dynamic energy efficiency evaluation model based on deep learning, and outputting the energy efficiency ratio and key energy consumption nodes of the current system includes the following steps: Step 21: Input the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, extract the dynamic features of the pressure, flow, temperature and vibration data through the feature extraction layer of the model, and generate a dynamic feature vector; Step 22: Input the dynamic feature vector into the fully connected layer of the dynamic energy efficiency evaluation model, calculate the energy efficiency ratio of the current system and the weight distribution of key energy consumption nodes, and generate energy efficiency ratio and key energy consumption node data; Step 23: Output the energy efficiency ratio and key energy consumption node data as the evaluation result of the dynamic energy efficiency evaluation model.

4. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 3 is characterized in that: The dynamic energy efficiency evaluation model adopts a hybrid architecture combining convolutional neural network and long short-term memory network; The feature extraction layer consists of multiple convolutional layers and pooling layers, and each convolutional layer uses a different convolution kernel size; During the feature extraction process, the pressure data is first normalized. The normalized pressure data and flow data are then feature mapped using a convolutional layer to generate a preliminary feature map. The temperature and vibration data are then processed using separate convolutional layers to extract their dynamic features. Subsequently, the feature maps and dynamic features are reduced in dimension through a pooling layer; finally, all feature maps and dynamic features are aligned in the time dimension and time series modeled through an LSTM layer to generate a dynamic feature vector.

5. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 4 is characterized in that: The fully connected layer consists of multiple hidden layers, each hidden layer uses the ReLU activation function, and the number of neurons in the input layer is consistent with the dimension of the dynamic feature vector; When calculating the energy efficiency ratio (EER), the fully connected layer first maps the dynamic feature vector to a low-dimensional space to generate a preliminary estimate of the EER. The back-propagation algorithm then optimizes the weight parameters to minimize the error between the predicted EER and the actual EER. The EER is defined as the ratio of the hydraulic system's effective output power to its input power. The weight distribution of the key energy consumption nodes is achieved through an attention mechanism, which assigns different weights to each time step in the dynamic feature vector to identify the operation node with the greatest impact on energy consumption.

6. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 5 is characterized in that: The method for generating a pressure-flow-temperature collaborative optimization strategy for a hydraulic system using a reinforcement learning algorithm includes the following steps: Step 31: Input the energy efficiency ratio and key energy consumption nodes into the reinforcement learning algorithm as basic data of the state space; Step 32: Generate the adjustable ranges of pressure, flow, and temperature of the hydraulic system through the action space definition module in the reinforcement learning algorithm; Step 33: Based on the state space and action space, iteratively calculate the pressure-flow-temperature collaborative parameter combination through the strategy optimization module of the reinforcement learning algorithm; Step 34: Output the pressure-flow-temperature collaborative parameter combination as a pressure-flow-temperature collaborative optimization strategy for the hydraulic system; the collaborative optimization strategy includes a pressure adjustment value, a flow adjustment value, and a temperature adjustment value.

7. The energy consumption optimization control method of the hydraulic system of the compression shear testing machine according to claim 6 is characterized in that: The dynamic adjustment of the oil pump output and the valve group opening comprises the following steps: Step 41: parsing the pressure-flow-temperature collaborative optimization strategy into an executable control instruction sequence through a control instruction conversion module; Step 42: Send the control instruction sequence to the actuator of the hydraulic system via the real-time control bus.

8. The energy consumption optimization control method for the hydraulic system of the compression shear testing machine according to claim 7 is characterized in that: The parsing process of the control instruction conversion module is as follows: The pressure adjustment value in the collaborative optimization strategy is first mapped to a corresponding motor speed command value through an oil pump characteristic curve, and the oil pump characteristic curve mapping is pre-established based on the oil pump's displacement-pressure-speed characteristic curve; The flow adjustment value is converted into a corresponding valve core displacement instruction value through the flow characteristic equation of the valve group of the proportional valve; The temperature adjustment value is converted into an adjustment instruction for the cooler fan speed or heater power according to the system heat balance equation.

9. The energy consumption optimization control method for the hydraulic system of the compression shear testing machine according to claim 8, characterized in that: The method of collecting instruction execution results is: The actual energy efficiency ratio of the motor after the control instruction is executed is recorded in real time to form the instruction execution result.

10. The energy consumption optimization control method for the hydraulic system of the compression shear testing machine according to claim 9, characterized in that: The adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution result includes the following steps: Step 51: Compare and analyze the instruction execution result with the expected optimization target to calculate the actual energy efficiency improvement rate; Step 52: Based on the actual energy efficiency improvement rate, update the parameter weights of the dynamic energy efficiency evaluation model; Step 53: Apply the updated model parameters to the next round of energy efficiency evaluation.

11. An energy consumption optimization control system for a hydraulic system of a compression shear testing machine, which is used to implement the energy consumption optimization control method for a hydraulic system of a compression shear testing machine according to any one of claims 1 to 10, characterized in that: It includes a data flow collection module, an evaluation model building module, a collaborative strategy generation module, and an execution feedback module; wherein each module is electrically connected; The data stream collection module collects the pressure, flow, temperature and vibration data of the hydraulic system of the compression and shear testing machine in real time through multiple sensors, generates multi-source heterogeneous data streams, and sends the multi-source heterogeneous data streams to the evaluation model construction module; An evaluation model construction module inputs the multi-source heterogeneous data stream into a dynamic energy efficiency evaluation model based on deep learning, outputs the energy efficiency ratio and key energy consumption nodes of the current system, and sends the energy efficiency ratio and key energy consumption nodes to a collaborative strategy generation module; A collaborative strategy generation module generates a pressure-flow-temperature collaborative optimization strategy for the hydraulic system based on the energy efficiency ratio and key energy consumption nodes through a reinforcement learning algorithm, and sends the collaborative optimization strategy to the execution feedback module; The execution feedback module converts the collaborative optimization strategy into control instructions for the hydraulic system, dynamically adjusts the oil pump output and valve group opening, collects the instruction execution results, performs adaptive model optimization of the dynamic energy efficiency evaluation model based on the instruction execution results, and returns the optimized dynamic energy efficiency evaluation model to the evaluation model construction module.

Images