Vibrating liquid quantity detecting device

since 05 Wang Yuchen

 

One 、 Project overview

   This device is the one I participated in this academic year SRT Project topic , It was completed by me in cooperation with another senior in automation department （ Two people work together ）. This work is based on the common phenomenon of knocking glasses with different liquid volume to produce different pitches in life , Based on this method, a non-invasive liquid measurement device is designed . This work is applicable to some special environments where intrusive measurement methods cannot be used , For example, it is highly toxic 、 Non invasive measurement of highly corrosive liquid storage tanks . The project finally won the third prize in the final evaluation of the school challenge cup .

Two 、 Research ideas and experimental process

   In order to break through the limitations of the existing non-invasive detection device , We have abandoned the traditional method of measuring the liquid storage height 、 Method for calculating liquid volume by weight and other indicators , Analyze the reservoir as a whole with the liquid in it . The core features here use the lessons learned in the course FFT The fundamental frequency of the system determined by the spectral analysis of the algorithm 、 Harmonic frequency . Then the finite element analysis software is used to determine the vibration modes corresponding to these frequencies , Then the excitation point and the receiving position of the sensor are determined , Finally, the neural network is used for data fitting .

3、 ... and 、 Experimental data and feature extraction

   For the 500mL Open glass beaker , Paste the piezoelectric sensor on the cup wall , At the same time, knock and excite the cup wall and bottom , To receive 2s,40kHz The time domain signal with sampling rate is filtered by Gauss filter to eliminate low-frequency noise , Proceed again FFT Identify and extract the characteristic frequency in the spectrum . The data is shown in the figure below , The waveform above is the bottom excitation result , The waveform below is the result of side wall excitation . Time domain results are shown on the left , The right side is FFT The frequency domain results obtained after .

▲ chart 3.1 Test waveform and spectrum

   It can be seen that there are two obvious characteristic frequencies , among 1kHz The left and right frequencies are obvious under both excitation conditions , and 872Hz The frequency of only exists under the side wall excitation . Then the finite element analysis is used to simulate the excitation of these two positions （ The percussion excitation can be approximately one unit impulse ）, The vibration modes corresponding to these two characteristic frequencies are obtained as follows ：

▲ chart 3.2 Two different modes of vibration

   Thus, it can be determined that the vibration modes corresponding to the two striking points are two independent vibration modes . Then change the liquid volume , Yes 0-500mL The liquid level in the range is every 25mL Take one measurement at an interval , Extract the characteristic frequencies of the above two vibration modes , Perform the frequency of the two modes - Liquid level curve drawing , give the result as follows ：

▲ chart 3.3 Relationship between beaker liquid volume and frequency

   You can see , The frequency change rate of bottom mode is higher at low liquid level , At high liquid level, the frequency change rate of sidewall mode is high . Traditional mode based liquid level detection only focuses on the sidewall mode , The range obtained is only the upper half level . And by introducing the bottom mode , Not only will the range be extended to the full level , It is even more difficult to measure the extremely low liquid level in the existing non-invasive liquid level detection device （0-50mL） It has extremely high precision performance .

Four 、 Device implementation

   In order to realize fully automatic excitation 、 testing 、 Liquid volume analysis , We carry out data communication through single-chip microcomputer , Connect the upper computer with the sensor , Excitation device . On the selection of excitation device , In order to obtain more ideal unit impulse excitation , We tried different excitation devices , Finally, the combination of shrapnel and electromagnet is determined , The device has a good acoustic effect on the container used in the experiment , The resulting spectrum is the cleanest . The overall device diagram is as follows ：（ From left to right are the upper computer and its graphical interface , Single chip microcomputer , Interface circuit , Excitation devices and sensors ）

▲ chart 4.1 Experimental test device

   In the actual measurement process , The upper computer will send instructions , The voltage signal is given by the single chip microcomputer to control the relay , And then drive the electromagnet to complete the excitation , Then the data collected by the sensor is transmitted to the single chip microcomputer through the interface circuit , The data structure is adjusted and transmitted to the upper computer , Data analysis 、 Processing and visual display .

   For data fitting methods , We use neural network instead of the traditional exponential fitting method . by comparison , Neural network is closer to its original law , And the data of two channels can be fitted at the same time , Higher accuracy . The structure and training results of the neural network are as follows ：

▲ chart 4.2 Use neural network for data fitting

   You can see , The training result curve is smooth , The projection law on the two planes is consistent with the experiment . Substitute into the test set for testing , give the result as follows ：

▲ chart 4.3 Test set test results

   selection 20 Individual liquid volume （ Non calibration point data ） As a test set . The maximum error on the test set shall not exceed 5mL, The average error does not exceed 2mL. The average accuracy of measurement reaches the average error 0.3%, maximum error 0.68% .

5、 ... and 、 Summary of achievements and prospect

   Compared with traditional non-invasive detection methods , The device realizes high-precision measurement at low liquid volume . This has important application value in the measurement of some expensive solutions . in addition , When the liquid level is inclined 、 In case of instability , This method also shows high robustness . In the works, neural network is used to fit 、 Instead of the traditional fixed function fitting method , It further improves the scientificity and accuracy of prediction . Experimental results show that , When the liquid level is low 、 When high , The measurement accuracy can reach two thousandths ; At the intermediate liquid level, the accuracy can also reach 5 ‰ .

▲ chart 5.1 Testing process ： Capacity 50mL

6、 ... and 、 Experiment and course thoughts

   This project has been studied since the summer vacation when I was a freshman , Until the middle of this semester, I will participate in the final examination of the challenge cup , It spans nearly a year . From the beginning of theoretical research 、 Literature research , To manual knocking test 、 Law exploration , And then to the construction of the device , System debugging 、 improvement , Finally, the experiment on the device 、 adjustment . We met many difficulties along the way , But we adjust step by step for our own goals and enthusiasm for this project 、 improvement , Finally, we got a satisfactory result . This is also my first scientific research experience , Although it is hard, I feel the harvest .

   For the course of signal and system , The core of our whole subject is actually sampling and frequency domain analysis . Therefore, many knowledge in the course of signal and system have important applications in this subject , such as FFT Time-frequency conversion 、 Sampling theorem 、 wave filtering 、 System functions, etc . In the process of learning signals and systems, I have many new ideas about our subject 、 And the continuous advancement of the project also deepened my understanding of the course of signal and system . In the study of almost every new concept , I can find the practical application of this concept in our project , So as to help me better remember and understand .

▲ chart 6.1 Testing process ： Capacity 300mL

   This also makes me understand that this course is a combination of theory and practice , And the importance of learning in practice . The achievement of our subject can not be achieved without the wonderful interpretation of the course of signal and system by Mr. Zhuo . so to speak , This course is the most practical one in my college , The most interesting course . Thank you for your hard work in this semester .

attach ： Challenge Cup Award certificate

▲ chart 6.2 Challenge Cup Award certificate

---

● Related chart Links :

• chart 3.1 Test waveform and spectrum
• chart 3.2 Two different modes of vibration
• chart 3.3 Relationship between beaker liquid volume and frequency
• chart 4.1 Experimental test device
• chart 4.2 Use neural network for data fitting
• chart 4.3 Test set test results
• chart 6.2 Challenge Cup Award certificate