“…Based to the pump affinity laws [15,27], it can be inferred that the flow rate Q is linearly related to the rotational frequency f.…”
Section: Discussionmentioning
confidence: 99%
“…In recent years, data-driven models based on historical data features have become a popular alternative method in centrifugal pump research [10][11][12][13][14]. These models can be designed even if the design team lacks a complete understanding of the internal flow field of the centrifugal pump, relying solely on a large amount of data and mature development experience [15][16][17][18]. However, the reliability of data-driven empirical models largely depends on the accuracy of the modeling data.…”
The present paper presents a multilayer hybrid model for sensorless measurement of pump operating status, with the objective of enabling safe and stable operations while reducing energy losses. The model takes easily measurable variables such as rotational frequency and valve opening as input features to predict the performance parameters of the centrifugal pump. By integrating Just-in-time learning (JITL) with Gaussian process regression (GPR) and leveraging the unique probability features of GPR, a JGPR is developed to extract valuable feature information. The JGPR sequentially predicts the flow rate, coefficient for dynamic head, and shaft power. The predicted values are extended to other input features, which can accurately capture the characteristics of the centrifugal pump and effectively replace the process of acquiring process parameters through sensors. Finally, the mechanism model is integrated into the multilayer JGPR model to calculate the performance parameters of centrifugal pump. The validation results indicate a strong agreement between predicted and experimental results, with predicted meeting performance parameters all engineering requirements. Compared to a single model, the multilayer hybrid model significantly improves the reliability of predictions, demonstrating the feasibility of using this approach to predict performance parameters. This research provides valuable insights into the measurement of sensorless pump operating states, enabling safe and efficient operation in complex conditions.
“…Based to the pump affinity laws [15,27], it can be inferred that the flow rate Q is linearly related to the rotational frequency f.…”
Section: Discussionmentioning
confidence: 99%
“…In recent years, data-driven models based on historical data features have become a popular alternative method in centrifugal pump research [10][11][12][13][14]. These models can be designed even if the design team lacks a complete understanding of the internal flow field of the centrifugal pump, relying solely on a large amount of data and mature development experience [15][16][17][18]. However, the reliability of data-driven empirical models largely depends on the accuracy of the modeling data.…”
The present paper presents a multilayer hybrid model for sensorless measurement of pump operating status, with the objective of enabling safe and stable operations while reducing energy losses. The model takes easily measurable variables such as rotational frequency and valve opening as input features to predict the performance parameters of the centrifugal pump. By integrating Just-in-time learning (JITL) with Gaussian process regression (GPR) and leveraging the unique probability features of GPR, a JGPR is developed to extract valuable feature information. The JGPR sequentially predicts the flow rate, coefficient for dynamic head, and shaft power. The predicted values are extended to other input features, which can accurately capture the characteristics of the centrifugal pump and effectively replace the process of acquiring process parameters through sensors. Finally, the mechanism model is integrated into the multilayer JGPR model to calculate the performance parameters of centrifugal pump. The validation results indicate a strong agreement between predicted and experimental results, with predicted meeting performance parameters all engineering requirements. Compared to a single model, the multilayer hybrid model significantly improves the reliability of predictions, demonstrating the feasibility of using this approach to predict performance parameters. This research provides valuable insights into the measurement of sensorless pump operating states, enabling safe and efficient operation in complex conditions.
“…BRBP neural network introduces Bayesian regularization rules based on the BP neural network to overcome the above problems of the BP network. Thus, the network's training speed and generalization performance are improved [22]. The modified objective function is added to the neural network weights, as shown in Eq.…”
As a typical welding structure, the girth weld of high-grade pipelines has obvious heterogeneity. Therefore, the axial mechanical properties of girth weld materials cannot be accurately tested, and the safety evaluation of the girth weld of the pipeline is seriously affected. Based on MATLAB-PYTHON-ABAQUS co-simulation, an optimization inversion method of the material stress-strain constitutive relationship in the weld zone of the high-grade pipeline is proposed in this paper. Four groups of uniaxial tensile tests with different notch sizes are carried out, and the load-displacement curves of each specimen are obtained. The true stress-strain constitutive relationship of the weld zone material is obtained by the Bayesian regularization back propagation (BRBP) neural network and the Grey wolf optimizer (GWO). The accuracy of the constitutive relationship is fully verified by the test data, and the relative error is less than 1%. In addition, taking the weld material of the oil and gas pipeline as an example, this paper proposes a universal design method of specimen size with a notch. The proposed method is convenient for researchers in different fields to measure the stress-strain constitutive relationship of different materials. It is worth mentioning that the inversion method proposed in this paper is also applicable to the measurement of the stress-strain constitutive relationship of homogeneous metallic materials in a large strain range. The proposed inversion method can provide an accurate stress-strain constitutive relationship for the safety evaluation of girth welds of high-grade pipelines.
“…The standard back-propagation neural network uses the BP algorithm which has three types of layers, input layer, hidden layer, and output layer (Figure 3). The training data entered the neural network data through the input layer, after that the hidden layer represents as processing layer, and the output layer is the last stage that produces the decision of the module [24]. The results of the output layer are compared with target data and find the error between them.…”
Section: Neural Network For Obstacle Classificationmentioning
When a mobile robot has the ability to avoid obstacles while traveling is called an autonomous robot. There are various methods and techniques used to get a collision-free path until gets to the target point. The dynamic obstacle problems are handled by reactive mobile robot navigation techniques. In this paper, the problem of dynamic obstacle avoidance has been addressed by proposing a combination between an Adaptive Neuro-Fuzzy inference system and a Neural network. The proposed system consists of three main parts. The first part was abstracted by using A* algorithm to get the initial path from the start to the goal point. The second part of the system classifies Obstacle(s). The classification estimate whether the obstacle is dangerous and may collide with the mobile robot or not. The relative velocity and distance between the mobile robot and obstacle (s) determine whether the obstacle(s) are dangerous or not. Bayesian regularization Back-Propagation Neural Network is used to train the data for obstacle severity classification. Where obstacle is divided into five zones where zone 1 is dangerous and zone 5 is safe. When obstacle gets into critical regions classified as dangerous. The third part of the system is related to avoiding obstacles if these obstacles are classified as a danger to the mobile robot. The Adaptive Neuro-Fuzzy Inference System has been adopted in the process of avoiding obstacles during the mobile robot motion. Obstacle avoidance is a reaction taken by the robot to avoid collision with dynamic obstacles around it, which are classified as dangerous obstacles by the neural networks. Three important criteria were used as input to the Adaptive Neuro-Fuzzy Inference System, which are the relative speed, distance, and angle between the robot and the obstacle, the output was a suggested steering angle and speed for the mobile robot. The simulation results for the tested cases show the capability of the proposed controller for avoiding static and dynamic obstacles in a fully known environment. The Adaptive Neuro-Fuzzy Inference System enhances the performance of the proposed controller resulting in the reduction of path length, processing time, and the number of iterations.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.