Device and method for predicting ash content of coal slurry based on optical detection

Abstract

The invention relates to a coal slurry ash content prediction device and method based on optical detection, belongs to the technical field of coal slurry ash content prediction, and solves the problems of poor prediction accuracy and difficulty in meeting actual requirements of industrial production caused by high difficulty in data acquisition of an ash content prediction model in the prior art. The coal slurry treatment module is used for diluting coal slurry to be detected into low-concentration coal slurry and conveying the low-concentration coal slurry to the multifunctional infrared detection module; the multifunctional infrared detection module is used for detecting infrared light transmission light intensity and visible light transmission light intensity of the low-concentration coal slurry; the ash content prediction module is used for inputting the infrared light transmission light intensity and the visible light transmission light intensity into an improved MLP ash content prediction model to obtain the ash content of the current coal slurry to be detected; wherein the improved MLP ash content prediction model comprises an input layer, a first hidden layer, a first output layer, a second hidden layer and a second output layer which are connected in sequence. The technical scheme of predicting the ash content of the coal slurry by combining an optical method and a machine learning model method is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of coal slurry ash prediction, and particularly to a device and method for predicting coal slurry ash based on optical detection. Background Art

[0002] Ash is the non-combustible residue remaining after coal is completely burned at high temperature, mainly composed of minerals in coal (such as oxides of silicon, aluminum, iron, calcium, etc.). Ash is an important index of coal quality, and its content directly affects the calorific value, combustion efficiency of coal and the environmental protection of subsequent utilization.

[0003] In the prior art, traditional ash prediction methods mainly include: laboratory analysis method, empirical formula method, and prediction method based on physical properties. With the rapid development of machine learning technology, using artificial intelligence algorithms such as artificial neural networks to learn and train a large amount of coal slurry sample data (including various physical and chemical properties of coal slurry, production process parameters, etc.) to establish an ash prediction model has become the main research direction of ash prediction.

[0004] However, the method of predicting by using artificial intelligence algorithms has the following disadvantages: (1) The data collected in actual production is easily interfered by various factors, such as equipment sensor errors, changes in environmental temperature and humidity, accidental fluctuations in the production process, etc. These noises will affect the accuracy and reliability of the data and interfere with the learning of the artificial intelligence model; (2) During the production process, due to equipment failures, data transmission problems, etc., some data may be missing, or some key data may not be recorded, making the data incomplete and affecting the model's comprehensive capture and understanding of data features. As a potential alternative, optical methods have received extensive attention due to their advantages such as non-contact, rapidity, and non-destructiveness. However, past optical methods have often been limited to the utilization of a single optical property, such as only using one kind of transmitted light or backscattered light, and cannot comprehensively consider the complex properties of coal samples. In addition, too high a concentration of coal slurry will cause the absorption of infrared light by coal particles to increase, thus greatly reducing the intensity of the received infrared light, also resulting in insufficient prediction ability.

[0005] Therefore, it is necessary to provide a technical solution for predicting coal slurry ash by combining optical methods and machine learning model methods. Summary of the Invention

[0006] In view of the above analysis, the embodiments of the present invention aim to provide a device and method for predicting coal slurry ash based on optical detection, so as to solve the problems of high difficulty in obtaining data of the existing ash prediction model, resulting in poor prediction accuracy and difficulty in meeting the actual needs of industrial production.

[0007] The embodiments of the present invention provide a device for predicting coal slurry ash based on optical detection, and the device includes:

[0008] A coal slurry processing module for diluting the coal slurry to be measured into a low-concentration coal slurry and transporting the low-concentration coal slurry to a multi-functional infrared detection module;

[0009] A multi-functional infrared detection module for detecting the infrared light transmission intensity and visible light transmission intensity of the low-concentration coal slurry;

[0010] An ash content prediction module for inputting the infrared light transmission intensity and the visible light transmission intensity into an improved MLP ash content prediction model to obtain the ash content of the current coal slurry to be measured; wherein, the improved MLP ash content prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

[0011] Based on a further improvement of the above device, the coal slurry processing module includes a sample bucket, a clean water tank, a sampling pump, a dilution pump, a stirrer, and an ultrasonic disperser; wherein, the sample bucket is used to store the coal slurry to be measured, the clean water tank is used to store clean water, the sampling pump is used to obtain the coal slurry to be measured from the sample bucket and transport it to the stirrer, the dilution pump is used to obtain clean water from the clean water tank and transport it to the stirrer, the stirrer is used to mix and dilute the coal slurry to be measured, and the ultrasonic disperser is used to disperse the coal slurry particles to be measured.

[0012] Based on a further improvement of the above device, the multi-functional infrared detection module includes:

[0013] A square groove with a feed port at one end for receiving the low-concentration coal slurry transported by the coal slurry processing module and a discharge port at the other end for discharging the low-concentration coal slurry;

[0014] A housing arranged outside the square groove for isolating external light sources;

[0015] An infrared laser device, which includes an infrared laser emitter and an infrared semiconductor sensor, and the infrared laser emitter and the infrared semiconductor sensor are arranged on opposite sides of the square groove;

[0016] A visible light device, which includes a visible light emitter and a visible light semiconductor sensor, and the visible light emitter and the visible light semiconductor sensor are arranged on opposite sides of the square groove;

[0017] And, the infrared laser device and the visible light device are arranged in sequence on the square groove.

[0018] Based on a further improvement of the above device, the device further includes:

[0019] A data processing module, configured to perform data conversion operations on the infrared light transmission intensity and the visible light transmission intensity obtained by the multifunctional infrared detection module, and send the obtained first transformed data and second transformed data to the ash prediction module for ash prediction of the coal slurry to be measured;

[0020] Among them, the data conversion operation includes:

[0021]

[0022] T1′ is the first transformed data, T1 is the first intensity data, T0 is the transmission light intensity in pure water, T2′ is the second transformed data, and T2 is the second intensity data.

[0023] Based on a further improvement of the above device, the first output result of the first output layer includes the grinding particle size of the coal slurry to be measured, the efficiency factor and volume concentration of the low-concentration coal slurry. The second hidden layer processes the first output result to obtain the second output result of the second output layer, and the second output result includes the coal slurry ash.

[0024] Based on a further improvement of the above device, the improved MLP ash prediction model is trained based on the first training sample set, and its loss function is:

[0025]

[0026] Among them, loss py is the loss caused by the physical properties of the coal slurry. d is the predicted grinding particle size output by the first output layer, d′ is the grinding particle size recorded in this sample in the first training sample set, φ is the predicted volume concentration output by the first output layer, φ′ is the volume concentration recorded in this sample in the first training sample set, ash is the predicted coal slurry ash output by the second output layer, ash′ is the coal slurry ash recorded in this sample in the first training sample set, and α1, α2, β1, β2 are weight coefficients.

[0027] Based on a further improvement of the above device, the device further includes:

[0028] A data storage module, configured to record the infrared light transmission intensity and the visible light transmission intensity collected by the multifunctional infrared detection module, as well as the grinding particle size and volume concentration output by the ash prediction module.

[0029] Based on a further improvement of the above device, the data processing module is further configured to perform filtering and noise reduction processing on the obtained infrared light transmission intensity and visible light transmission intensity to improve data accuracy.

[0030] Based on further improvements to the above device, the training samples in the first training sample set are obtained through the following methods:

[0031] Obtain coal sample specimens with multiple different ash contents, and perform the following operations on each coal sample specimen:

[0032] A1: Crush, grind, and dry the coal sample specimen, measure the true ash content value of the coal sample specimen through manual chemical analysis, select a coal sample specimen of a first preset mass after treatment as a coal test sample, place it in the sample bucket, add water of a second preset mass to the water tank, and transport the coal test sample and water to a stirrer for mixing and dilution;

[0033] A2: The diluted coal slurry enters the multi-functional infrared detection module through a hose;

[0034] A3: The data storage module records the infrared light transmission intensity and visible light transmission intensity at this volume concentration, and takes the grinding particle size of the coal sample specimen, this volume concentration, the infrared light transmission intensity and visible light transmission intensity at this volume concentration, and the true ash content value of the coal sample specimen as a piece of training sample data;

[0035] A4: Continue to add a coal sample specimen of a third preset mass to the sample bucket, transport the added coal sample specimen to the stirrer through the sampling pump, calculate the volume concentration at this time, and return to step A2 until the training sample data of the coal sample specimen at each volume concentration is obtained.

[0036] On the other hand, an embodiment of the present invention provides a method for predicting the ash content of coal slurry based on optical detection, including:

[0037] Dilute the coal slurry to be measured into a low-concentration coal slurry, and obtain the infrared light transmission intensity and visible light transmission intensity of the low-concentration coal slurry;

[0038] Input the infrared light transmission intensity and the visible light transmission intensity into an improved MLP ash content prediction model to obtain the ash content of the currently measured coal slurry; wherein, the improved MLP ash content prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

[0039] Compared with the prior art, the present invention can at least achieve one of the following beneficial effects:

[0040] 1. The present invention provides an optical detection-based coal slurry ash prediction device and method. By diluting the coal slurry with the ash content to be detected, the matrix effect is reduced, and the particles in the coal slurry are more evenly dispersed in the solution. Moreover, the ash content is reduced to the measurable range of the instrument, expanding the measurable range of the coal slurry ash, so that the ash content can be accurately measured, further improving the accuracy of ash prediction. The dual information of infrared transmitted light and visible transmitted light is used to predict the clean coal ash content, comprehensively considering the complex characteristics of the coal slurry, and further improving the accuracy of the ash prediction result. The improved MLP ash prediction model is used to analyze and process complex non-linear relationships, more accurately reflecting the characteristics of the coal sample and improving the prediction accuracy and precision. In addition, using the transmitted light intensity as the input data for ash prediction is simpler and more convenient to obtain data compared with the method of using various physical and chemical properties and production process parameters of the coal slurry as input data in the prior art, and is more suitable for ash prediction in industrial production activities.

[0041] 2. The present invention provides an optical detection-based coal slurry ash prediction device and method, and constructs an MLP ash prediction model with an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer. First, the mapping relationships between the infrared light transmitted light intensity, the visible light transmitted light intensity, the grinding particle size of the coal sample to be measured, the efficiency factor and volume concentration of the low-concentration coal slurry are established. Then, the mapping relationships between the grinding particle size, the efficiency factor and the volume concentration and the coal slurry ash are established. In a prediction model, two mapping relationships are considered. Compared with the prior art where machine learning algorithms only establish the mapping relationship between input and output, the method proposed by the present invention further improves the accuracy of model prediction. In addition, during the training process, since the efficiency factor value is difficult to obtain in actual production, the loss function is adaptively improved, further improving the model training efficiency.

[0042] In the present invention, the above technical solutions can also be combined with each other to achieve more preferred combination schemes. Other features and advantages of the present invention will be described in the subsequent specification, and some advantages will be obvious from the specification or can be understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained from the content specifically pointed out in the specification and the drawings. Description of the Drawings

[0043] The drawings are only for the purpose of showing specific embodiments and are not considered to be a limitation of the present invention. Throughout the drawings, the same reference signs denote the same components.

[0044] Figure 1 It is an example diagram of an optical detection-based coal slurry ash prediction device in an embodiment of the present invention.

[0045] Figure 2 An exemplary diagram of the improved MLP ash prediction model in the embodiments of the present invention;

[0046] Figure 3 An exemplary diagram of the multi-functional infrared detection module in the embodiments of the present invention;

[0047] Figure 4 An exemplary diagram of a method for predicting coal slurry ash based on optical detection in the embodiments of the present invention;

[0048] Reference numerals:

[0049] 1 - Feed inlet; 2 - Discharge outlet; 3 - Infrared laser emitter; 4 - Infrared semiconductor sensor; 5 - Infrared laser emitter; 6 - Infrared semiconductor sensor; 7 - Infrared laser emitter; 8 - Infrared semiconductor sensor; 9 - Visible light emitter; 10 - Visible light semiconductor sensor; 11 - Square groove; 12 - Housing. Detailed implementation manners

[0050] The following will specifically describe the preferred embodiments of the present invention in conjunction with the accompanying drawings. The accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0051] A specific embodiment of the present invention discloses a device for predicting coal slurry ash based on optical detection, as Figure 1 shown. The device includes:

[0052] A coal slurry processing module, configured to dilute the coal slurry to be measured into a low-concentration coal slurry and transport the low-concentration coal slurry to the multi-functional infrared detection module.

[0053] A multi-functional infrared detection module, configured to detect the infrared light transmission intensity and the visible light transmission intensity of the low-concentration coal slurry.

[0054] An ash prediction module, configured to input the infrared light transmission intensity and the visible light transmission intensity into the improved MLP ash prediction model to obtain the ash of the current coal slurry to be measured; wherein, as Figure 2 shown, the improved MLP ash prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

[0055] The device further includes:

[0056] A data processing module, configured to perform data conversion operations on the infrared light transmission intensity and the visible light transmission intensity obtained by the multi-functional infrared detection module, and send the obtained first transformed data and second transformed data to the ash prediction module for predicting the ash of the coal slurry to be measured. Wherein, the data conversion operation includes:

[0057]

[0058] T1' is the first transformed data, T1 is the first intensity data, T0 is the transmitted light intensity in pure water, T2' is the second transformed data, and T2 is the second intensity data. Considering that there may be instrument drift or signal instability during data acquisition, which may require correction of the background signal and other systematic errors. Therefore, the data processing module is also used to perform filtering and noise reduction processing on the acquired infrared light transmitted light intensity and the visible light transmitted light intensity to ensure the accuracy and stability of the final data.

[0059] The data storage module is used to record the infrared light transmitted light intensity and the visible light transmitted light intensity collected by the multifunctional infrared detection module, as well as the grinding particle size and volume concentration output by the ash content prediction module.

[0060] For the coal slurry processing module, it includes a sample bucket, a clean water tank, a sampling pump, a dilution pump, a stirrer, and an ultrasonic disperser. Among them, the sample bucket is used to store the coal slurry to be tested, the clean water tank is used to store clean water, the sampling pump is used to obtain the coal slurry to be tested from the sample bucket and transport it to the stirrer, the dilution pump is used to obtain clean water from the clean water tank and transport it to the stirrer, the stirrer is used to mix and dilute the coal slurry to be tested, and the ultrasonic disperser is used to disperse the coal slurry particles to be tested. Specifically, after the high-concentration coal slurry to be tested and clean water are input into the stirrer, the impeller in the stirrer starts to rotate, and at the same time the ultrasonic disperser also starts to work to complete the mixing and dilution of the coal slurry to be tested.

[0061] The coal slurry processing module also includes a discharge port for discharging the mixed and diluted low-concentration coal slurry, and its discharge port is connected to the feed port of the multifunctional infrared detection module; a mixing bucket for receiving the low-concentration coal slurry discharged from the discharge port of the multifunctional infrared detection module for use in supplementary experiments when the data deviation is large or incorrect. The low-concentration coal slurry in the mixing bucket is pumped into the stirrer via a circulation pump and circulated again until the test data meets the expectations.

[0062] For the multifunctional infrared detection module, it includes:

[0063] A square groove with a feed port at one end for receiving the low-concentration coal slurry transported by the coal slurry processing module and a discharge port at the other end for discharging the low-concentration coal slurry;

[0064] A housing arranged outside the square groove for isolating external light sources;

[0065] Infrared laser device, the infrared laser device includes an infrared laser emitter and an infrared semiconductor sensor, and the infrared laser emitter and the infrared semiconductor sensor are arranged on opposite sides of a square groove;

[0066] Visible light device, the visible light device includes a visible light emitter and a visible light semiconductor sensor, and the visible light emitter and the visible light semiconductor sensor are arranged on opposite sides of a square groove;

[0067] Moreover, the infrared laser device and the visible light device are arranged in sequence on the square groove.

[0068] Among them, the square groove can be made of a transparent material, for example, a quartz sample tube bonded by highly transparent infrared quartz; there is at least 1 infrared laser device and at least 1 visible light device, and the housing can be a black housing.

[0069] See Figure 3 , one end of the square groove 11 is provided with a feed port 1, the other end is provided with a discharge port 2, the housing 12 is arranged outside the square groove 11, there are 3 infrared laser devices, including infrared laser emitters 3, 5, 7 and infrared semiconductor sensors 4, 6, 8, and there is 1 visible light device, including a visible light emitter 9 and a visible light semiconductor sensor 10. Among them, the powers of the three infrared laser devices are different. This design is to adapt to the detection of different concentration ranges and coal slurry ash contents. The low-power infrared laser can be used to detect coal slurry with a relatively low ash content. Because the low-power laser has relatively weak energy and relatively mild interaction with the coal slurry, it will not cause overly strong reactions or changes in the components in the coal slurry, and can more accurately detect the characteristic signals of low-concentration ash, avoiding signal saturation or distortion caused by excessive energy, thereby improving the detection accuracy of low-ash coal slurry; for coal slurry with a relatively high ash content, the high-power infrared laser can provide sufficient energy to fully interact with various components in the coal slurry, making components such as minerals in the ash produce more obvious characteristic signals, such as specific infrared absorption or scattering signals, which is convenient for more accurately identifying and measuring the ash content. The appropriate power can be selected for detection according to the characteristics of different coal types. For example, for some coal types with hard texture and complex structure, a relatively high-power laser may be required to penetrate the coal slurry and obtain accurate information; while for coal types with soft texture and relatively simple composition, a relatively low-power laser may be sufficient to complete the detection, thereby expanding the detection range of coal slurry ash for different coal types.

[0070] Exemplarily, two high-power (for example, 500 mw) infrared laser devices can be set, with emission wavelengths of 880 nm and 1500 nm respectively, one low-power infrared laser device with an emission wavelength of 1500 nm, and one visible light device with a power of 500 mw and an emission wavelength of 450 nm, so as to ensure that the rays emitted by the infrared laser device and the visible light device can penetrate the quartz sampling tube.

[0071] Exemplarily, the infrared semiconductor sensor is closely attached to the quartz sampling tube through a black silica gel gasket to detect the transmitted infrared light intensity; the photosensitive element of the visible light semiconductor sensor is at an angle of 180° with the laser path of the visible light emitter and is collinear.

[0072] For the ash prediction module, the present invention further improves the existing multi-layer perceptron MLP model. The input data is the infrared light transmission intensity and the visible light transmission intensity. The first output result of the first output layer includes the grinding particle size of the coal slurry to be measured, the efficiency factor and volume concentration of the low-concentration coal slurry. The second hidden layer processes the first output result to obtain the second output result of the second output layer, and the second output result includes the coal slurry ash. The improved MLP ash prediction model is trained based on the first training sample set, and its loss function is:

[0073]

[0074] where loss py is the loss caused by the physical properties of the coal slurry, d is the predicted grinding particle size output by the first output layer, d′ is the grinding particle size recorded in this sample in the first training sample set, φ is the predicted volume concentration output by the first output layer, φ′ is the volume concentration recorded in this sample in the first training sample set, ash is the predicted coal slurry ash output by the second output layer, ash′ is the coal slurry ash recorded in this sample in the first training sample set, and α1, α2, β1, β2 are weight coefficients. It should be noted that the parameter value of the efficiency factor is difficult to obtain. Therefore, when designing the loss function, the relationship between the predicted value and the calibrated value of the efficiency factor is not considered.

[0075] The training samples in the first training sample set are obtained by the following method:

[0076] Obtain multiple coal samples with different ash contents, and perform the following operations on each coal sample:

[0077] A1: Crush, grind, and dry the coal sample. Manually analyze and determine the true ash content value of the coal sample. Select a first preset mass of the processed coal sample as the coal test sample, put it into the sample bucket, add a second preset mass of water to the water tank, and transport the coal test sample and water to the stirrer for mixing and dilution.

[0078] Before the operating conditions of the coal slurry treatment module become stable, monitor and confirm the flow state of the coal slurry in the equipment. It is necessary to ensure that the coal slurry has no bubbles and flows smoothly, and judge the uniformity of the diluted coal slurry in the stirring pipe and the stability of the output signal of the transmitted light sensor. The stability means that when the state of the coal slurry remains unchanged, the floating range of the transmitted light intensity signal should be less than 0.5% of the signal range. The solution proposed by the present invention needs to ensure that the coal slurry flows evenly and smoothly inside the equipment and the output signal is stable, and then realize the acquisition of sample data.

[0079] Exemplarily, when the output signal of the transmitted light sensor is in a stable state, data acquisition is carried out for 3 minutes; the first preset mass is 20 g, and the second preset mass is 2 kg.

[0080] A2: The diluted coal slurry enters the multifunctional infrared detection module through a hose.

[0081] A3: The data storage module records the infrared light transmitted light intensity and visible light transmitted light intensity at this volume concentration, and takes the grinding particle size of the coal sample, this volume concentration, the infrared light transmitted light intensity and visible light transmitted light intensity at this volume concentration, and the true ash content value of the coal sample as a piece of training sample data.

[0082] A4: Continue to add a third preset mass of coal sample to the sample bucket, transport the added coal sample to the stirrer through the sampling pump, calculate the volume concentration at this time, and return to step A2 until the training sample data of the coal sample at each volume concentration is obtained.

[0083] After completing the data acquisition of the previous volume concentration group, continue to add coal samples with the same ash content to the sample bucket, continue to test the coal slurry under the new ash content, observe the change of the output signal, conduct data acquisition after the signal is stable, gradually increase the coal slurry concentration, conduct continuous experimental observations, until after measuring the coal samples with this ash content at each concentration within a preset range (for example, the preset range can be 0 - 180 g / L), replace the coal samples with different ash contents, and repeat the above steps for new measurements.

[0084] Exemplarily, the third preset mass is 1 g.

[0085] The present invention uses analysis of variance to conduct significance tests and evaluations on the improved MLP ash prediction model. If the adjusted regression coefficient is greater than 95% and the predicted regression coefficient is greater than 90%, it indicates that the model accuracy meets the requirements and the actual regression coefficient also meets the standard. If it fails to meet the standard, it is necessary to adjust the model parameters and re - train and fit.

[0086] Another specific embodiment of the present invention discloses an optical - detection - based coal slurry ash prediction method, as Figure 4 shown, including:

[0087] Dilute the coal slurry to be measured into a low-concentration coal slurry, and obtain the infrared light transmission intensity and visible light transmission intensity of the low-concentration coal slurry;

[0088] Input the infrared light transmission intensity and the visible light transmission intensity into an improved MLP ash prediction model to obtain the ash content of the current coal slurry to be measured; wherein, the improved MLP ash prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

[0089] Compared with the prior art, a coal slurry ash prediction device and method based on optical detection provided in this embodiment reduces the matrix effect by diluting the coal slurry with the ash content to be detected, makes the particles in the coal slurry more evenly dispersed in the solution, and reduces the ash content to the measurable range of the instrument, expanding the measurable range of the coal slurry ash, so as to accurately measure its ash content and further improve the accuracy of ash prediction; uses the dual information of infrared transmitted light and visible transmitted light to predict the clean coal ash content, comprehensively considers the complex characteristics of the coal slurry, and can further improve the accuracy of the ash prediction result; analyzes and processes the complex non-linear relationship through the improved MLP ash prediction model, more accurately reflects the characteristics of the coal sample, and improves the prediction accuracy and accuracy. In addition, using the transmitted light intensity as the input data for ash prediction is simpler and more convenient to obtain data compared with the method of using various physical and chemical properties and production process parameters of the coal slurry as the input data in the prior art, and is more suitable for ash prediction in industrial production activities. An MLP ash prediction model with an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer is constructed. First, the mapping relationships between the infrared light transmission intensity, visible light transmission intensity and the grinding particle size of the coal sample to be measured, the efficiency factor and volume concentration of the low-concentration coal slurry are established, and then the mapping relationships between the grinding particle size, efficiency factor and volume concentration and the coal slurry ash are established. In a prediction model, two mapping relationships are considered. Compared with the prior art where the machine learning algorithm only establishes the mapping relationship between input and output, the method proposed by the present invention further improves the accuracy of model prediction. In addition, during the training process, since the efficiency factor value is difficult to obtain in actual production, the loss function is adaptively improved, further improving the model training efficiency.

[0090] Those skilled in the art can understand that all or part of the processes for implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a disk, an optical disc, a read-only memory or a random access memory, etc.

[0091] As described above, it is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A coal slurry ash prediction device based on optical detection, characterized in that: The device comprises: A coal slurry processing module, used for diluting the coal slurry to be tested into a low-concentration coal slurry, and conveying the low-concentration coal slurry to the multifunctional infrared detection module; A multifunctional infrared detection module, used to detect the infrared light transmission intensity and the visible light transmission intensity of the low-concentration coal slurry; The ash content prediction module is used to input the infrared light transmission intensity and the visible light transmission intensity into the improved MLP ash content prediction model to obtain the ash content of the current coal slurry to be tested; wherein the improved MLP ash content prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

2. The device for predicting coal slurry ash content based on optical detection according to claim 1, characterized in that: The coal slurry processing module includes a sample barrel, a clean water tank, a sampling pump, a dilution pump, an agitator, and an ultrasonic disperser; wherein the sample barrel is used to store the coal slurry to be tested, the clean water tank is used to store clean water, the sampling pump is used to obtain the coal slurry to be tested from the sample barrel and transport it to the agitator, the dilution pump is used to obtain clean water from the clean water tank and transport it to the agitator, the agitator is used to achieve mixing and dilution of the coal slurry to be tested, and the ultrasonic disperser is used to disperse the coal slurry particles to be tested.

3. The device for predicting coal slurry ash content based on optical detection according to claim 2, characterized in that: The multifunctional infrared detection module comprises: A square trough, wherein one end of the square trough is provided with a feed port for receiving the low-concentration coal slurry delivered by the coal slurry processing module, and the other end is provided with a discharge port for discharging the low-concentration coal slurry; A cover shell is arranged outside the square slot and is used to isolate external light sources; An infrared laser device, the infrared laser device comprising an infrared laser emitter and an infrared semiconductor sensor, the infrared laser emitter and the infrared semiconductor sensor being arranged on opposite sides of the square groove; A visible light device, the visible light device comprising a visible light emitter and a visible light semiconductor sensor, wherein the visible light emitter and the visible light semiconductor sensor are arranged on opposite sides of the square groove; Furthermore, the infrared laser device and the visible light device are arranged in sequence on the square groove.

4. The device for predicting coal slurry ash content based on optical detection according to claim 3, characterized in that: The device also includes: A data processing module, used for performing data conversion operation on the infrared light transmission intensity and the visible light transmission intensity obtained by the multifunctional infrared detection module, and sending the obtained first transformation data and second transformation data to the ash content prediction module for ash content prediction of the coal slurry to be tested; The data conversion operation includes: T1′ is the first transformation data, T1 is the first intensity data, T0 is the transmitted light intensity in pure water, T2′ is the second transformation data, and T2 is the second intensity data.

5. The device for predicting coal slurry ash content based on optical detection according to claim 1, characterized in that: The first output result of the first output layer includes the grinding particle size of the coal slurry to be tested, the efficiency factor and volume concentration of the low-concentration coal slurry, and the second hidden layer processes the first output result to obtain the second output result of the second output layer, and the second output result includes the ash content of the coal slurry.

6. The device for predicting coal slurry ash content based on optical detection according to claim 5, characterized in that: The improved MLP ash content prediction model is trained based on the first training sample set, and its loss function is: Among them, loss py is the loss caused by the physical properties of coal slurry, d is the predicted grinding particle size output by the first output layer, d′ is the grinding particle size recorded in the sample in the first training sample set, φ is the predicted volume concentration output by the first output layer, φ′ is the volume concentration recorded in the sample in the first training sample set, ash is the predicted coal slurry ash output by the second output layer, ash′ is the coal slurry ash recorded in the sample in the first training sample set, α1, α2, β1, β 2 is the weight coefficient.

7. The device for predicting coal slurry ash content based on optical detection according to claim 6, characterized in that: The device also includes: The data storage module is used to record the infrared light transmission intensity and the visible light transmission intensity collected by the multifunctional infrared detection module, as well as the grinding particle size and volume concentration output by the ash prediction module.

8. The device for predicting coal slurry ash content based on optical detection according to claim 7, characterized in that: The data processing module is also used to perform filtering and noise reduction processing on the acquired infrared light transmission intensity and the visible light transmission intensity to improve data accuracy.

9. The device for predicting coal slurry ash content based on optical detection according to claim 8, characterized in that: The training samples in the first training sample set are obtained in the following manner: Obtain multiple coal samples with different ash contents, and perform the following operations on each coal sample: A1: crushing, grinding and drying the coal sample, determining the real ash value of the coal sample by manual testing, selecting a first preset mass of the processed coal sample as a coal test sample, putting the coal sample into the sample bucket, adding a second preset mass of water into the clean water tank, and conveying the coal test sample and the water to a stirrer for mixing and dilution; A2: The diluted coal slurry enters the multifunctional infrared detection module through a hose; A3: The data storage module records the infrared light transmission intensity and the visible light transmission intensity at the volume concentration, and uses the grinding particle size of the coal sample, the volume concentration, the infrared light transmission intensity and the visible light transmission intensity at the volume concentration, and the real ash value of the coal sample as a training sample data; A4: Continue to add a third preset mass of coal sample to the sample barrel, transport the added coal sample to the agitator through the sampling pump, calculate the volume concentration at this time, and return to step A2 until the training sample data of the coal sample at each volume concentration is obtained.

10. A method for predicting coal slurry ash content based on optical detection, characterized in that: include: Diluting the coal slurry to be tested into a low-concentration coal slurry, and obtaining the infrared light transmission intensity and the visible light transmission intensity of the low-concentration coal slurry; Used to input the infrared light transmission intensity and the visible light transmission intensity into the improved MLP ash prediction model to obtain the ash content of the current coal slurry to be tested; wherein the improved MLP ash prediction model includes an input layer, a first hidden layer, a first output layer, a second hidden layer, and a second output layer connected in sequence.

Images