Water Pump Flow Rate Calculation With Preprocessed Neural Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating fluid flow rate in water pumps either occupy significant storage space or require high computing power, leading to increased costs and reduced accuracy.
Innovation Solution
A computer-implemented method using a neural network model to determine a flow rate expression characteristic, which is introduced into a water pump control system, allowing for reduced computing load and improved accuracy by preprocessing data on a computer device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the look-up table method is used to calculate fluid flow rate, then the calculation is simpler, but the storage space of MCU is occupied and calculation accuracy is insufficient
Solution Approach 1:
The patent applies preliminary action by pre-training a neural network model offline to generate a simplified flow rate calculation model. This pre-processed model is then deployed on the MCU, allowing the system to perform accurate flow rate calculations with minimal real-time computing resources. The complex neural network training is performed in advance on a more powerful computer, transferring the computational burden from the embedded system to the offline preprocessing stage.
2Measurement precision
If the neural network model is deployed on MCU for real-time calculation, then the flow rate calculation accuracy is improved, but the computing power requirements and cost increase
Solution Approach 1:
The patent extracts the computationally intensive neural network training process from the MCU and performs it offline on a more powerful computer system. Only the essential calculation parameters and simplified model structure are extracted and deployed to the MCU. This separation allows the system to achieve high calculation accuracy without requiring the MCU to have high computing power, as the heavy computational work has been removed and performed in advance.
3Measurement precision
If the neural network model is deployed on MCU, then the flow rate calculation accuracy is improved, but the storage space requirements increase
Solution Approach 1:
The patent applies local quality by optimizing the neural network model to have different characteristics for different parts of the system. The model structure is designed to have high precision in critical calculation areas while using simplified representations in less critical areas. This allows the system to maintain high flow rate calculation accuracy while minimizing the storage space required on the MCU, as only essential model parameters are stored locally.
Data Source
Figure 1~2
AI summary
The embodiments of the present invention relate to a computer-implemented method and system for determining a fluid flow rate of a water pump, wherein the method comprises: determining a flow rate expression characteristic based on revolutions per minute and power according to a flow rate of the water pump under multiple sets of revolutions per minute and powers based on a neural network model through a computer device; wherein, the flow rate expression characteristic comprises an expression term characteristic and a coefficient characteristic corresponding thereto; introducing the flow rate expression characteristic into a water pump control system through the computer device; determining an expression term characteristic value corresponding to the current power and current revolutions per minute of the water pump according to the expression term characteristic through the water pump control system; and determining a current flow rate of the water pump according to the expression term characteristic value and the coefficient characteristic through the water pump control system. The embodiments of the present invention achieve a reduction in the computing load of the MCU in the water pump control system, and an improvement in the calculation accuracy of the flow rate simultaneously.