Intelligent inversion method for diaphragm head pressure of diaphragm compressor based on strain test

Abstract

The invention discloses a strain test-based diaphragm compressor diaphragm head pressure intelligent inversion method, which comprises the following steps of: firstly, acquiring acquisition data of an internal pressure value of a diaphragm head of a test compressor with the same model as a diaphragm compressor to be tested and a strain value of a set position of the diaphragm head; secondly, in combination with the acquired data, establishing an inversion rule model of film head strain and pressure of compressors of the same type by utilizing an intelligent algorithm; then, acquiring a strain value at the same position as the set position of the diaphragm head of the tested compressor on the diaphragm head of the diaphragm compressor to be tested with the same model; and finally, extracting a strain value, and calculating an internal pressure value of the membrane head in combination with the established inversion rule model so as to obtain the real-time internal pressure of the membrane head. The invention has the advantages of real-time and timely monitoring, high monitoring precision, high clearance volume efficiency, high bearing strength of the membrane head, capability of monitoring the pressure of each working stage, safety and reliability of equipment and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of diaphragm compressor head pressure monitoring. More specifically, the present invention relates to an intelligent inversion method for diaphragm compressor head pressure based on strain testing. Background Art

[0002] Diaphragm compressors are widely used in hydrogen refueling stations, high-purity gases, and chemical industries, and are key dynamic equipment. However, the internal working gas chamber in the compressor head is very small, and at the same time, the diaphragm moves at a high frequency in the working chamber, making it very difficult to obtain the pressure inside the chamber. However, the pressure inside the working chamber is a very important indicator for the compressor, which can evaluate the thermodynamic performance of the compressor and determine whether the air valve and diaphragm have failed.

[0003] In the prior art, pressure tests are generally carried out by the following methods: installing a pressure sensor on the compressor head to directly measure the internal pressure, or installing a pressure sensor on the outlet pipeline of the compressor head to indirectly measure the internal pressure of the compressor head, or calculating the working gas chamber pressure based on the flat plate theory model through radial and circumferential strain values. However, there are also some problems: 1. Installing a pressure sensor on the compressor head increases the clearance volume, reduces the compression efficiency and the pressure-bearing strength of the compressor head. 2. Installing a pressure sensor on the outlet pipeline of the compressor head cannot reflect the actual pressure in the internal working chamber of the compressor head during the suction, compression, exhaust, and expansion stages. 3. The flat plate theory model ignores the influences of the compressor head structure, oil pressure, temperature, etc., and has a large error. Summary of the Invention

[0004] An object of the present invention is to provide an intelligent inversion method for diaphragm compressor head pressure based on strain testing, which has the advantages of real-time and timely monitoring, high monitoring accuracy, high clearance volume efficiency, high pressure-bearing strength of the compressor head, being able to monitor the pressure in each working stage, and the equipment being safe and reliable.

[0005] To solve the above technical problems, the present invention provides an intelligent inversion method for diaphragm compressor head pressure based on strain testing. First, acquisition data of the internal pressure value of the compressor head of a test compressor of the same model as the diaphragm compressor to be tested and the strain value at the set position of the compressor head is obtained; secondly, combining the acquisition data obtained above, an inversion law model of the strain and pressure of the compressor head of the same model is established using an intelligent algorithm; then, the strain value at the same position as the set position of the test compressor head is obtained on the diaphragm compressor head of the same model to be tested; finally, the strain value is extracted, and combined with the established inversion law model, the internal pressure value of the compressor head is calculated, so as to obtain the real-time pressure inside the compressor head.

[0006] Preferably, the acquisition data is obtained through a load test on a compressor test platform, or obtained through factory testing before the same model of diaphragm compressor leaves the factory.

[0007] Preferably, before acquiring the collected data, a pressure sensor is installed on the diaphragm head of the test compressor, and strain gauges are pasted at the set positions. After the pressure sensor is installed on the test diaphragm head, it is directly connected to the working air chamber.

[0008] Preferably, a plurality of strain gauges are respectively pasted radially and circumferentially at the middle part of the air-side end face of the test diaphragm head, and there is a set distance from the exhaust pipe, preferably 5 - 10 cm.

[0009] Preferably, an inversion law model is established, including but not limited to BP neural network, SVM model.

[0010] Preferably, establishing an inversion law model specifically includes: Dividing the collected strain and pressure value data into a training sample set and a test sample set. Each sample set contains strain and pressure data, where the strain is used as the input value and the pressure is used as the target value; Then using the training sample set to train the intelligent algorithm model to obtain the model parameters; Using the test sample set to test the intelligent algorithm model to check whether the calculation error of the intelligent algorithm model meets the expected level; if it meets the expected level, the model training is completed; if it does not meet the expected level, adjust the model training parameters, retrain, and check again until the calculation error of the intelligent algorithm model meets the expected level.

[0011] Preferably, there are three ways to select the strain value. The value measured by the radial strain gauge is used as the input value, or the value measured by the circumferential strain gauge is used as the input, or the value obtained by comprehensively calculating the values measured by the radial and circumferential strain gauges is used as the input value.

[0012] Preferably, establish inversion law models of strain and pressure for the diaphragm heads of multiple different models of compressors to realize the inversion of the internal pressure of the diaphragm head based on the strain values of the diaphragm heads of different models.

[0013] The present invention has at least the following beneficial effects: 1. Compared with the calculation formula derived from the diaphragm head flat plate theory, the intelligent inversion method of the present invention inverses the internal pressure of the diaphragm head through an artificial intelligence algorithm model, can inversely calculate the internal pressure of the diaphragm head in real time, and has higher accuracy.

[0014] 2. Compared with indirectly measuring the internal pressure of the diaphragm head by installing a pressure sensor on the outlet pipeline of the diaphragm head, the intelligent inversion method of the present invention can measure and inverse the pressure in all working stages inside the diaphragm head, and can fully reflect the pressure in the working chamber of the diaphragm head during the suction, compression, exhaust, and expansion stages.

[0015] 3. Compared with installing a pressure sensor on the diaphragm head, the intelligent inversion method of the present invention is pre-tested on diaphragm compressors of the same model. There is no need to drill additional measurement holes on the working chamber of the diaphragm head, ensuring volumetric efficiency, higher clearance volumetric efficiency, and higher pressure-bearing strength of the diaphragm head.

[0016] 4. The present invention can be applied in the monitoring system of diaphragm compressors. Intelligent algorithm models are established and trained for different models of compressors to monitor the internal pressure of the diaphragm head of diaphragm hydrogen compressors, and timely evaluate the performance status of the air valves and diaphragms of the diaphragm head to ensure the safe and reliable operation of the equipment.

[0017] Other advantages, objectives, and features of the present invention will be partially reflected by the following description and partially understood by those skilled in the art through the research and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the installation of one of the pressure sensors and strain gauges of the present invention; Figure 2 It is a schematic diagram of the principle of the BP neural network of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0019] In order to better understand the purpose, structure, and function of the present invention, the following further detailed description of the present invention with reference to the drawings is provided for those skilled in the art to implement according to the description in the specification.

[0020] It should be noted that the experimental methods described in the following embodiments are all conventional methods unless otherwise specified, and the reagents and materials can be obtained from commercial channels unless otherwise specified; in the description of the present invention, the orientation or positional relationship indicated by terms such as "horizontal", "vertical", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0021] The present invention provides an intelligent inversion method for the diaphragm head pressure of a diaphragm compressor based on strain testing. The principle is as follows: Inside the laboratory or before leaving the factory, the internal pressure value of the diaphragm head of compressors of the same model and the strain value at specific parts of the diaphragm head are collected, and then an inversion law model of the strain and pressure of the diaphragm head of compressors of the same model is established using an intelligent algorithm. During on-site application, strain gauges are pasted at specific positions specified on the diaphragm head of the same model, and the internal pressure value of the diaphragm head is calculated by using the extracted strain value and the established inversion law model of this type of compressor, so as to obtain the real-time pressure inside the diaphragm head.

[0022] In actual application, it is necessary to establish the strain-pressure inversion law models for compressor diaphragm heads of various models in the laboratory or before leaving the factory, establish and train different intelligent algorithm models for different models of compressors to achieve the inversion of the internal pressure of the diaphragm head from the strain value. Therefore, usually compressor manufacturers have a better implementation foundation. Or it can be carried out on a compressor test platform in the laboratory.

[0023] The specific implementation plan is as follows.

[0024] Obtain the direct internal pressure and external strain values of the diaphragm head in the laboratory through the test diaphragm head. Then establish an artificial intelligence algorithm model, use the strain value and pressure value as the training sample set and target sample set to train and test the intelligent algorithm model, and obtain an intelligent algorithm model that meets the requirements. Finally, use this model to inversely calculate the internal pressure of the diaphragm head from the measured strain value.

[0025] Collect the strain value of the diaphragm head from the field without collecting the pressure value. The diaphragm head on-site is not a test diaphragm head and a pressure sensor directly connected to the air chamber cannot be installed. Use the strain value collected on-site as the input of the intelligent algorithm model to inversely calculate the internal pressure value of the diaphragm head.

[0026] 1. Collect the strain value and internal pressure value of the diaphragm head. Obtain the collection data of the internal pressure value of the test compressor diaphragm head and the strain value at the set position of the diaphragm head of the same model as the diaphragm compressor to be tested.

[0027] a. Prepare the test diaphragm head, install a pressure sensor on the diaphragm head, and paste strain gauges at specific positions (as Figure 1 shown). Compared with the conventional diaphragm head, the test diaphragm head can install a pressure sensor directly connected to the working air chamber.

[0028] When the diaphragm compressor is in operation, the piston inside it is pushed by the piston rod to perform a linear reciprocating motion. Therefore, the force generated by the piston pushing the gas in the reciprocating direction will be transmitted to the diaphragm head side. The stress is relatively concentrated at the middle part of the gas-side end face of the diaphragm head and the edge part fastened by bolts. In order to monitor the deformation of the diaphragm head, radial and circumferential strain gauges can be pasted at the middle part of the gas-side end face of the diaphragm head.

[0029] The specific positions where the strain gauges are pasted, such as Figure 1 shows the schematic diagrams of the pressure sensor and the strain gauge. Paste the strain gauge at a position with a distance d from the exhaust pipe. The size of d is specified as a unified value, such as 5 cm, 10 cm. Theoretically, the smaller d is, the more appropriate it is. The closer to the center of the cylinder head (exhaust pipe), the greater the strain deformation, and the more conducive it is to measuring the strain value. Specify a unified distance d to ensure that all measurement points are compared on a single benchmark.

[0030] b. On the compressor test platform, conduct a compressor load test, and use a data collector to measure the high-frequency pressure value and strain value.

[0031] 2. Establish an inversion law model of diaphragm head strain and pressure using intelligent algorithms. Combining the acquisition data obtained above, use intelligent algorithms to establish an inversion law model of diaphragm head strain and pressure for compressors of the same model; obtain the strain value at the same position as the set position of the diaphragm head of the test compressor on the diaphragm compressor of the same model to be tested; extract the strain value, and combine it with the established inversion law model above to calculate the internal pressure value of the diaphragm head, so as to obtain the real-time pressure inside the diaphragm head.

[0032] a. Establish an intelligent algorithm model, including but not limited to, such as BP neural network, SVM model, etc.

[0033] As Figure 2 shown, the principle of the BP neural network is a multi-layer feedforward network that evaluates the weights of each layer according to error backpropagation. Its topological structure can be divided into three layers, namely the input layer, the output layer, and the hidden layer. Usually, the BP model needs to be trained before it can be used for identification and classification. During training, the training sample feature values are passed to the network model, and then the neurons in the input layer, hidden layer, and output layer of the model are activated in sequence and calculated according to the neuron activation function; then, according to the error between the target output and the actual output, the connection weights are adjusted layer by layer in the reverse direction, so that the network finally converges to meet the error requirements or the number of iterations.

[0034] b. Divide the strain and pressure value data collected on the test platform into a training sample set and a test sample set. Each sample set contains strain and pressure data, where the strain is used as the input value and the pressure is used as the target value. There are three ways to select the strain value. The value measured by the radial strain gauge is selected as the input value, or the value measured by the circumferential strain gauge is selected as the input, or the value obtained by comprehensively calculating the values measured by the radial and circumferential strain gauges is used as the input value. The comprehensive calculation is such as removing the mean value, or taking different weight coefficients, or calculating according to a set formula, that is, the input value is obtained after a certain processing calculation of the two data.

[0035] c. Use the training sample set to train the intelligent algorithm model to obtain the model parameters and clarify the model.

[0036] d. Use the test sample set to test the intelligent algorithm model to check whether the calculation error of the intelligent algorithm model meets the expected level. If it meets the expectation, the model training is completed. If it does not meet the expectation, adjust the model training parameters and retrain until the calculation error of the test meets the expected level.

[0037] It will be understood that the present invention is described by way of some embodiments, and those skilled in the art will know that various changes or equivalent replacements can be made to these features and embodiments without departing from the spirit and scope of the present invention. Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily achieved. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to the specific details and the illustrated examples here.

Claims

1. An intelligent inversion method for the diaphragm compressor head pressure based on strain testing, characterized in that, First, obtain the acquisition data of the internal pressure value of the diaphragm head and the strain value at the set position of the diaphragm head of the test compressor of the same model as the diaphragm compressor to be tested; secondly, combine the acquisition data obtained above and use an intelligent algorithm to establish an inversion law model of the strain and pressure of the diaphragm head of the compressor of the same model. Then, obtain the strain value at the same position as the set position of the diaphragm head of the above test compressor on the diaphragm head of the diaphragm compressor of the same model to be tested; finally, extract the strain value, and combine it with the established inversion law model above to calculate the internal pressure value of the diaphragm head, so as to obtain the real-time pressure inside the diaphragm head.

2. The intelligent inversion method for the diaphragm compressor head pressure based on strain measurement according to claim 1, wherein The acquisition data is obtained through a load test on the compressor test platform, or obtained through factory testing before the diaphragm compressor of the same model leaves the factory.

3. The intelligent inversion method for the diaphragm compressor head pressure based on strain test according to claim 1, characterized in that, Before obtaining the acquisition data, install a pressure sensor on the diaphragm head of the test compressor, and paste strain gauges at the set positions. After the pressure sensor on the test diaphragm head is installed, it is directly connected to the working air chamber.

4. The intelligent inverse calculation method for the diaphragm compressor head pressure based on strain measurement according to claim 3, characterized in that, Paste multiple strain gauges radially and circumferentially at the middle part of the gas-side end face of the test diaphragm head, and there is a set distance between them and the exhaust pipe, preferably 5-10 cm.

5. The intelligent inversion method for the diaphragm compressor head pressure based on strain test according to claim 1, characterized in that Establish an inversion law model, including but not limited to BP neural network, SVM model.

6. The intelligent inversion method for the diaphragm compressor head pressure based on strain test according to claim 4, wherein Establishing an inversion law model specifically includes: Divide the collected strain and pressure value data into a training sample set and a test sample set. Each sample set contains strain and pressure data, where the strain is used as the input value and the pressure is used as the target value. Then use the training sample set to train the intelligent algorithm model to obtain the model parameters. Use the test sample set to test the intelligent algorithm model to check whether the calculation error of the intelligent algorithm model meets the expected level; if it meets the expected level, the model training is completed; if it does not meet the expected level, adjust the model training parameters, retrain, and check again until the calculation error of the intelligent algorithm model meets the expected level.

7. The intelligent inverse method for the diaphragm compressor head pressure based on strain measurement according to claim 6, wherein, There are three ways to select the strain value. The value measured by the radial strain gauge is selected as the input value, or the value measured by the circumferential strain gauge is selected as the input, or the value obtained by comprehensively calculating the values measured by the radial and circumferential strain gauges is selected as the input value.

8. The intelligent inversion method for the diaphragm compressor head pressure based on strain test according to claim 1, wherein Establish inversion law models of the strain and pressure of the diaphragm heads of multiple different models of compressors to realize the inversion of the internal pressure of the diaphragm heads from the strain values of the diaphragm heads of different models of compressors.

Images