1. The marine fuel sulfur content detector is characterized by comprising a detector body (7) and a shell, wherein a containing cavity is formed in the detector body (7), a light source (8), a first sample cell (4), a second sample cell (9), a spectrometer (5), an LED display screen (6), a data analysis module (3), a peristaltic pump (2) and a power supply (1) are arranged in the containing cavity, the power supply (1) is arranged on a bottom plate of the containing cavity, and the peristaltic pump (2) is arranged on the power supply (1);

one side of the peristaltic pump (2) is provided with the data analysis module (3), and the other side is provided with the second sample cell (9);

the data analysis module (3) is provided with the light source (8), the light source (8) is provided with the first sample cell (4), and one side of the first sample cell (4) is provided with the spectrometer (5);

the LED display screen (6) is arranged on the shell.

2. The marine fuel sulfur content detector according to claim 1, wherein a liquid inlet of the peristaltic pump (2) is connected to the second sample cell (9), and a liquid outlet of the peristaltic pump (2) is connected to the first sample cell (4).

3. The marine fuel sulfur content detector according to claim 1, wherein said first sample cell (4) is connected to said light source (8) and said spectrometer (5), respectively.

4. The marine fuel sulfur content detector according to claim 1, wherein the second sample cell (9) is a detachable structure.

5. The detector for detecting the sulfur content of the marine fuel oil as claimed in claim 1, wherein an algorithm for analyzing the trace elements of the marine fuel oil is loaded in the data analysis module (3).

6. The method for detecting the sulfur content in the marine fuel oil based on any one of claims 1 to 5 is characterized by comprising the following steps of:

s1: starting a light source (8) to irradiate the solution in the first sample cell (4), starting a peristaltic pump to dropwise add a sulfur-containing calibration material to the solution in the first sample cell (4), and simultaneously starting a spectrometer (5) to collect data in the first sample cell in real time;

s2: storing the data in the first sample pool collected in the spectrometer (5) in S1 in real time through a data analysis module (3);

s3: data preprocessing is carried out on the data stored in the data analysis module (3) in the S2, and preprocessed data are obtained;

s4: classifying and testing the preprocessed data in the S3 by using a convolutional neural network to obtain classified and tested data;

s5: outputting the data of the classification test in the S4 by using a convolutional neural network to obtain a classification probability value;

s6: and (5) weighting the probability value classified in the S5 and the real concentration value, and outputting the category serial number and the element content to obtain the sulfur content in the fuel oil to be detected.

7. The method for detecting the sulfur content in the marine fuel oil as claimed in claim 6, wherein said preprocessing of the data in S3 uses a dimensionality reduction process.

8. The method for detecting the sulfur content in the marine fuel oil as claimed in claim 7, wherein the solution in the first sample cell (4) in the step S1 is a mixed solution of tin dioxide quantum dots and a diluted fuel oil solution.

9. The method for detecting the sulfur content in the marine fuel oil detector as claimed in claim 8, wherein the diluted fuel oil solution is diluted with an alcohol solution.

Background

The shipping industry is vital to the role of international trade, and a large number of international goods are shipped each year. But the atmospheric pollution is also aggravated when the ship sails, and a large number of research results show that SO in the exhaust gas discharged by the ship₂About 4 to 9 percent. This is because the fuel oil used in ships contains many sulfur components, such as mercaptan, thioether, disulfide, thiophene and its derivatives, among which, mainly, organic sulfide, which is harmful to human body and atmosphere environment, and most of the sulfur dioxide formed during combustion is discharged with the exhaust gas. In order to effectively control the emission of pollutants in the tail gas of ships, the international maritime organization regulates that ships with sulfur content higher than 0.5% are prohibited from sailing in 2020, and simultaneously, the maritime supervision department gradually strengthens the supervision of ships with excessive emission.

At present, the conventional fuel oil sulfur content detection methods commonly used by maritime supervision departments comprise third-party detection and portable sulfur content detector detection. Conventional third party verification steps include: and manually judging suspicious ships, and collecting oil samples when supervisors board the ships, and sending the oil samples to a third-party inspection mechanism for detection to obtain a detection result. The whole detection period generally needs more than 3 days, and from the viewpoint of real-time performance, the method is difficult to meet the actual supervision requirement. Therefore, the marine supervision department usually adopts a portable sulfur content detector to measure the sulfur content in an oil sample, and the main principle is that a primary ray emitted by a ray tube is utilized to excite a sample, the sulfur content in the sample emits a characteristic ray, the characteristic ray is measured by a ray detector and the ray intensity is recorded, the ray intensity is in direct proportion to the sulfur content, and the sulfur content in various oil products can be measured by a curve calibrated in advance. However, when a fluorescence sulfur detector is used for measuring a sample, particularly a sample containing trace sulfur and sulfur compounds, the proportion of sulfur rays absorbed by air is large, and a detector receives sulfur rays with certain errors, so that the measurement accuracy is low, the stability is low, the measurement effect is influenced, and the price is high.

Disclosure of Invention

The invention provides a detector and a detection method for sulfur content in marine fuel oil, and aims to solve the problems of low measurement accuracy, low stability and high price of the conventional detector.

In order to achieve the purpose, the technical scheme of the invention is as follows:

a marine fuel sulfur content detector comprises a detector body and a shell, wherein a containing cavity is arranged in the detector body, a light source, a first sample cell, a second sample cell, a spectrometer, an LED display screen, a data analysis module, a peristaltic pump and a power supply are arranged in the containing cavity, the power supply is arranged on a bottom plate of the containing cavity, and the peristaltic pump is arranged on the power supply;

the data analysis module is arranged on one side of the peristaltic pump, and the second sample pool is arranged on the other side of the peristaltic pump;

the data analysis module is provided with the light source, the light source is provided with the first sample cell, and one side of the first sample cell is provided with the spectrometer;

the LED display screen is arranged on the shell.

Further, a liquid inlet of the peristaltic pump is connected with the second sample cell, and a liquid outlet of the peristaltic pump is connected with the first sample cell.

Further, the first sample cell is respectively connected with the light source and the spectrometer.

Further, the second sample cell is of a detachable structure.

Furthermore, an algorithm for analyzing the trace elements of the marine fuel oil is carried in the data analysis module.

A detection method of a detector for detecting the sulfur content of marine fuel oil comprises the following steps:

s1: starting a light source to irradiate the solution in the first sample pool, starting a peristaltic pump to dropwise add a sulfur-containing calibration material to the solution in the first sample pool, and simultaneously starting a spectrometer to collect data in the first sample pool in real time;

s2: storing the data in the first sample pool collected in the spectrometer in S1 in real time through a data analysis module;

s3: data preprocessing is carried out on the data stored in the data analysis module in the S2, and preprocessed data are obtained;

s4: classifying and testing the preprocessed data in the S3 by using a convolutional neural network to obtain classified and tested data;

s5: outputting the data of the classification test in the S4 by using a convolutional neural network to obtain a classification probability value;

s6: and (5) weighting the probability value classified in the S5 and the real concentration value, and outputting the category serial number and the element content to obtain the sulfur content in the fuel oil to be detected.

Further, the data preprocessing in S3 uses a dimension reduction process.

Further, the solution in the first sample cell in S1 is a mixed solution of tin dioxide quantum dots and the diluted fuel solution.

Further, the diluted fuel solution is diluted with an alcohol solution.

The detector and the detection method for the sulfur content in the ship have the characteristics of high detection speed, high accuracy, safety and environmental protection, and the sample does not need to be ignited in the use process, so that the pollution to the environment and the adverse components to a human body can be reduced, wherein the tin dioxide quantum dots used in the detection have the advantages of good chemical stability, no toxicity and low cost.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a schematic structural diagram of a sulfur content detector for marine fuel oil according to the present invention;

FIG. 2 is a schematic block diagram of sulfur measurement;

FIG. 3 is a flow chart of sulfur measurement;

FIG. 4 is a fluorescence spectrum of the mixed tin dioxide quantum dots and fuel solution;

FIG. 5 is a graph showing the change in fluorescence intensity measured at 310nm for various sulfur contents;

FIG. 6 is a general block diagram of an algorithm;

fig. 7 is a block diagram of a core multi-way convolutional neural network MCNN.

In the figure, the device comprises a power supply 1, a peristaltic pump 2, a data analysis module 4, a first sample cell 5, a spectrometer 6, an LED display screen 7, a detector body 8, a light source 9 and a second sample cell.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

As shown in fig. 1, the marine fuel sulfur content detector comprises a detector body 7 and a housing, wherein an accommodating cavity is arranged in the detector body 7, a light source 8, a first sample cell 4, a second sample cell 9, a spectrometer 5, an LED display screen 6, a data analysis module 3, a peristaltic pump 2 and a power supply 4 are arranged in the accommodating cavity, the power supply 1 is arranged on a bottom plate of the accommodating cavity, and the peristaltic pump 2 is arranged on the power supply 1; in this embodiment, preferably, the light source 8 is an LED light source, the data analysis module 3 is a sulfur content data analyzer for establishing a characteristic curve between the sulfur content and the fluorescence value, and the spectrometer 5 is a display panel for measuring the fluorescence intensity of the solution in the sample cell and displaying the sulfur content data.

One side of the peristaltic pump 2 is provided with the data analysis module 3, and the other side is provided with the second sample cell 9; further, the second sample cell 9 is a detachable structure. In this embodiment, the second sample cell 9 is detachable and can be detached from the detector body, which is convenient for cleaning.

The data analysis module 3 is provided with the light source 8, the light source 8 is provided with the first sample cell 4, and one side of the first sample cell 4 is provided with the spectrometer 5; the LED display screen 6 is arranged on the shell. In this embodiment, preferably, the first sample cell 4 is a four-way cuvette, the light source 8 is an LED excitation light source, the LED excitation light source irradiates the reagent through the cuvette from one side to generate fluorescence, and the generated fluorescence is collected through a spectrometer distributed at 90 degrees with the LED excitation light source through the cuvette.

Further, a liquid inlet of the peristaltic pump 2 is connected with the second sample cell 9, and a liquid outlet of the peristaltic pump 2 is connected with the first sample cell 4.

Further, the first sample cell 4 is connected to the light source 8 and the spectrometer 5, respectively.

Furthermore, an algorithm for analyzing the trace elements of the marine fuel is carried in the data analysis module 3.

As shown in fig. 2-7, the method for detecting the sulfur content in the marine fuel oil comprises the following steps:

s1: starting a light source to irradiate the solution in the first sample pool, starting a peristaltic pump to dropwise add a sulfur-containing calibration material to the solution in the first sample pool, and simultaneously starting a spectrometer to collect data in the first sample pool in real time; in this example, preferably, 10 sulfur-containing calibrators were added, and each drop of liquid was calculated to increase the sulfur content in the sample cell by o.1%, and the fluorescence intensity was measured 10 times, as shown in fig. 5, where the curve of n-butyl sulfide added several times was plotted and the features were identified by a convolutional neural network. 0.1% is the mass fraction of sulfur in the fuel, and we increased the overall sulfur content by increasing the mass fraction of sulfur in the test oil samples. The quality of oil added into the tin dioxide quantum dots after the test oil sample is diluted is fixed, and the content of sulfur in the fuel oil is increased instead of increasing the sulfur according to the quality of the sulfur. For example, the mass fraction of sulfur in 1g of fuel is 3%, and the addition of 0.1% sulfur is 0.1% of the addition of 1g of fuel.

S2: storing the data in the first sample pool collected in the spectrometer in S1 in real time through a data analysis module; in the present embodiment, the data acquisition module is configured to automatically save the fluorescence value data acquired by the spectrometer every 40 s. During actual measurement, a start key on the control panel is pressed, and the single chip microcomputer sends an external control signal and simultaneously triggers the peristaltic pump to automatically drop a sample and collect data of the spectrometer.

S3: data preprocessing is performed on the data stored in the data analysis module 3 in the step S2 to obtain preprocessed data; in this embodiment, the preprocessed data are imported into a trained convolutional neural network to process the data, and the data preprocessing adopts max-min normalization, PCA and discrete sampling methods to process 11 × 350 input data into 1 × 150 preprocessed data.

S4: classifying and testing the preprocessed data in the S3 by using a convolutional neural network to obtain classified and tested data; in the embodiment, the convolutional neural network part adopts the network structure and adopts the Softmax function as an activation function to perform multi-classification tasks in an output layer.

S5: outputting the data of the classification test in the S4 by using a convolutional neural network to obtain a classification probability value;

s6: and (5) weighting the probability value classified in the S5 and the real concentration value, and outputting the category serial number and the element content to obtain the sulfur content in the fuel oil to be detected. In this embodiment, the output part performs weighted sum on the class probability value output by the convolutional neural network and the real concentration value, and finally outputs the class number and the element content, and the convolutional neural network extracts different sulfur content data set characteristics through training of a large number of initial samples, thereby realizing measurement of the sulfur content in the fuel.

Further, the data preprocessing in S3 uses a dimension reduction process. In this embodiment, the algorithm first performs the dimension reduction processing and the GAMMA transformation on the collected data to better extract the data features.

Further, in the S1, the solution in the first sample cell 4 is a mixed solution of tin dioxide quantum dots and the diluted fuel solution.

Further, the diluted fuel solution is diluted with an alcohol solution. In the present embodiment, it is preferable that the fuel is diluted by an alcohol solution, and the amount of the alcohol solution used is 100 times the amount of the fuel.

In the embodiment, an excitation light source at the absorption wavelength of 300nm of the LED light source 8 is used to irradiate the tin dioxide quantum dots and the diluted fuel oil mixed solution in the first sample cell 4, so that the mixed solution generates fluorescence, and the spectrometer 5 collects a fluorescence signal and converts the optical signal into an electrical signal. Wherein the four-way cuvette pool is connected with the LED light source and the spectrometer, and the LED light source and the spectrometer are distributed at 90 degrees. The liquid inlet of the peristaltic pump is connected with the sample pool of the front shell, and the liquid outlet is communicated with the four-way cuvette pool. The second sample cell 9 is filled with n-butyl thioether solution, and the rotation speed of the peristaltic pump is automatically controlled by the singlechip to adjust the flow rate of the n-butyl thioether solution. Since the maximum absorption wavelength of the ultraviolet absorption spectrum of tin dioxide is about 310nm, the maximum absorption wavelength of an excitation light source of about 300nm is selected as the excitation wavelength of tin dioxide photoluminescence.

As shown in fig. 6 and 7, the programmed program is transferred by using the single chip as a man-machine interaction board, and the functions of starting and closing the peristaltic pump 2 and adjusting the rotating speed and the data acquisition function of the spectrometer 5 are realized by using the external control module. The spectrometer 5 converts the collected optical signals into electric signals and transmits the electric signals to the sulfur content data analyzer 3 for storage and processing, the sulfur content data analyzer serves as a data analysis module 3 and carries an algorithm special for a trace element analysis task of marine fuel oil, the algorithm firstly carries out dimensionality reduction processing and GAMMA conversion on the collected data to better extract data characteristics, then an innovative idea of a multi-path convolutional neural network and a non-symmetric convolutional kernel is adopted to classify and test the received data, the sulfur content data analyzer feeds back a test result to the single chip microcomputer, and a man-machine interaction interface of the single chip microcomputer displays the sulfur content and whether the sulfur content is qualified or not.

The algorithm takes a convolutional neural network as a core and simultaneously carries out the tasks of fuel oil classification and trace element content estimation. As shown in fig. 7, the overall structure of the algorithm includes: firstly, inputting spectral line data; preprocessing the data; then carrying out fuel classification and trace element content estimation by a convolutional neural network; and finally, outputting an analysis result. The algorithm structure of the existing MCNN multi-path convolution neural network is referred, and three convolution kernels with the sizes of 1 x 3, 1 x 7 and 1 x 9 are selected on the basis of the algorithm structure. And a pooling layer with a pooling kernel size of 1 x 2 was added after each convolutional layer. Firstly, the data preprocessing method adopts the methods of max-min normalization, PCA and discrete sampling to process 11 × 350 input data into 1 × 150 preprocessed data. Secondly, the convolutional neural network part adopts the network structure and adopts a Softmax function as an activation function at an output layer to perform multi-classification tasks. And finally, the output part performs weighted sum on the class probability value output by the convolutional neural network and the real concentration value, and finally outputs the class serial number and the element content.

Fig. 7 shows a structure diagram of a core multi-path convolutional neural network MCNN, which is the core of the algorithm and is used for performing the tasks of fuel classification and estimation of the content of trace elements. The device mainly comprises the following parts: firstly, an input layer is provided, and the scale of the input layer is the dimension of matrixing data; secondly, extracting the characteristics of the multi-path convolution layer; important characteristic information is reserved by the pooling layer, and the calculation complexity is reduced; then recombining the features through the planarization layer; then enters the full link layer through a Dropout mechanism; and finally, carrying out fuel classification and trace element content estimation.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.