CN-121978035-A - Full spectrum neural network water quality monitoring method based on flow-through self-cleaning

CN121978035ACN 121978035 ACN121978035 ACN 121978035ACN-121978035-A

Abstract

The invention discloses a full spectrum neural network water quality monitoring method based on flow-through self-cleaning, which belongs to the technical field of water quality detection, and comprises the steps of comparing light absorption line data of a sample to be detected with air standard light absorption line data, calculating the spectrum similarity to be detected through a similarity function, judging the existence of liquid by utilizing a spectrum similarity threshold value to finish automatic liquid detection, obtaining a liquid existence judging result, realizing targeted self-cleaning, avoiding dry brushing, greatly improving the service life of a cleaning motor, and collecting the spectrum of a target sample according to the liquid existence judging result and the cleaning condition of a flow-through cell so as to improve the accuracy of water quality monitoring.

Inventors

XIE JINQIANG
HU XIAOJING
Sheng Zhiquan
LU WENZHAO
ZHOU FEIYUN

Assignees

博瑞思数智科技(深圳)有限公司

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. The utility model provides a full spectrum neural network water quality monitoring method based on circulation formula self-cleaning which characterized in that includes: Acquiring preset air standard light absorption line data and a spectrum similarity threshold value, and acquiring a measurement trigger instruction to emit a wide spectrum light beam to a sample to be measured in a flow through tank, and performing spectrum acquisition of the sample to be measured to obtain the light absorption line data of the sample to be measured, wherein the air standard light absorption line data and the light absorption line data of the sample to be measured are a group of discrete vector sequences; obtaining a preset similarity function, calculating the to-be-detected spectrum similarity between the to-be-detected sample light absorption line data and the air standard light absorption line data by using the similarity function, and judging the presence of liquid by using the spectrum similarity threshold value to obtain a liquid presence judging result; Acquiring a preset self-cleaning period of the flow cell and a self-cleaning operation log of the flow cell, and executing the self-cleaning operation of the flow cell according to the self-cleaning period of the flow cell, the self-cleaning operation log of the flow cell and the liquid existence judging result to form a cleaned flow cell; And taking the sample to be detected in the cleaned flow cell as a target sample, emitting a wide spectrum light beam to the target sample in the cleaned flow cell, carrying out target sample spectrum acquisition to obtain target sample absorbance spectrum data of the target sample, inputting the target sample absorbance spectrum data into a pre-trained full spectrum feedforward neural network reasoning model, and calculating a target water quality parameter concentration value of the target sample to realize water quality monitoring.
2. The method for monitoring water quality of a full spectrum neural network based on flow-through self-cleaning according to claim 1, wherein obtaining preset air standard absorbance spectrum data and spectrum similarity threshold value, and obtaining measurement trigger instruction to emit wide spectrum light beam to a sample to be tested in a flow-through tank, performing spectrum collection of the sample to be tested, obtaining absorbance spectrum data of the sample to be tested, comprises: Acquiring preset air standard light absorption spectrum line data Wherein the air standard light absorption spectrum line data And (2) and Represent the first The number of spectral acquisition points is one, Represent the first Air standard absorbance collected at the individual spectral collection points, Representing the total number of spectrum acquisition points; Acquiring a measurement trigger instruction, and controlling the generation of a wide spectrum light beam according to the measurement trigger instruction so as to utilize The spectrum acquisition points acquire the synchronous spectrum of the sample to be detected in the flow cell to obtain the light absorption spectrum line data of the sample to be detected Wherein the sample to be measured is an air sample or a liquid sample in the flow cell, and the sample to be measured absorbs spectral line data And (2) and Represent the first Absorbance of a sample to be measured, which is collected at each spectrum collection point.
3. The method for monitoring water quality of a full spectrum neural network based on flow-through self-cleaning according to claim 2, wherein obtaining a preset similarity function, calculating a to-be-detected spectrum similarity between the to-be-detected sample light absorption spectrum line data and the air standard light absorption spectrum line data by using the similarity function, and determining the presence of liquid by using the spectrum similarity threshold value to obtain a liquid presence determination result, comprises: Obtaining a preset similarity function Wherein the similarity function Is Euclidean distance algorithm, inner product algorithm or cosine algorithm; the absorbance spectrum line data of the sample to be measured And the air standard absorbance line data Inputting the similarity function To calculate the light absorption spectrum line data of the sample to be measured And the air standard absorbance line data Spectral similarity to be measured between ; Using the spectral similarity threshold For the spectrum similarity to be measured Judging the existence of the liquid to obtain a judging result of the existence of the liquid; If the spectrum similarity to be measured Not lower than the spectral similarity threshold The liquid presence determination result is that no liquid is present; If the spectrum similarity to be measured Below the spectral similarity threshold The liquid presence determination result is that a liquid is present.
4. The flow-through self-cleaning based full spectrum neural network water quality monitoring method of claim 1, wherein obtaining a preset flow-through cell self-cleaning cycle and a self-cleaning operation log of the flow-through cell, performing flow-through cell self-cleaning operation according to the flow-through cell self-cleaning cycle, the flow-through cell self-cleaning operation log and the liquid presence determination result, forming a cleaned flow-through cell, comprising: When the liquid existence judging result is that the liquid does not exist, carrying out spectrum acquisition on a sample to be detected at the next moment to obtain light absorption spectrum data of the sample to be detected at the next moment, calculating the similarity of the spectrum to be detected between the light absorption spectrum data of the sample to be detected and the air standard light absorption spectrum data, and carrying out liquid existence judging on the similarity of the spectrum to be detected by utilizing the spectrum similarity threshold again until the liquid existence judging result is that the liquid exists; When the liquid existence judging result is that the liquid exists, acquiring a preset self-cleaning period of the flow cell and a self-cleaning operation log of the flow cell, and executing the self-cleaning operation of the flow cell based on the self-cleaning period of the flow cell and the self-cleaning operation log of the flow cell to form a cleaned flow cell.
5. The flow-through self-cleaning based full spectrum neural network water quality monitoring method of claim 4, wherein performing a flow-through self-cleaning operation based on the flow-through self-cleaning cycle and a self-cleaning operation log of the flow-through cell to form a cleaned flow-through cell comprises: Extracting a latest cleaning record based on the self-cleaning operation log of the flow cell, and extracting time information corresponding to the latest cleaning record as latest cleaning time; Acquiring time information of the current moment, and calculating a difference value according to the latest cleaning time and the time information of the current moment to obtain a difference value between the latest cleaning time and the time information of the current moment as the unwashed duration of the flow cell; judging whether the unwashed time length of the flow cell reaches or exceeds the self-cleaning period of the flow cell based on the self-cleaning period of the flow cell and the unwashed time length of the flow cell; if not, marking the flow cell at the current moment as a cleaned flow cell; If yes, sending a flow cell cleaning control signal to execute flow cell self-cleaning operation according to the flow cell cleaning control signal, marking the cleaned flow cell as a cleaned flow cell, recording time information at the current moment to form a current cleaning record, and recording the current cleaning record into a self-cleaning operation log of the flow cell.
6. The method for monitoring water quality of a full spectrum neural network based on flow-through self-cleaning according to claim 1, wherein a sample to be tested in the cleaned flow-through cell is used as a target sample, a wide spectrum light beam is emitted to the target sample in the cleaned flow-through cell, target sample spectrum acquisition is performed to obtain target sample light absorption spectrum data of the target sample, the target sample light absorption spectrum data is input into a pre-trained full spectrum feedforward neural network inference model, a target water quality parameter concentration value of the target sample is calculated, and water quality monitoring is achieved, and the method comprises the following steps: Taking a sample to be detected in the cleaned flow cell as a target sample, and emitting a wide spectrum light beam to the target sample in the cleaned flow cell so as to acquire synchronous spectrum, thereby obtaining target sample absorbance spectrum data of the target sample; And acquiring a pre-trained full-spectrum feedforward neural network reasoning model, inputting target sample extinction line data of the target sample into an input layer of the full-spectrum feedforward neural network reasoning model, and carrying out feature extraction and nonlinear mapping on the target sample extinction line data through the full-spectrum feedforward neural network reasoning model so as to output a target water quality parameter concentration value and finish water quality monitoring.
7. The full spectrum neural network water quality monitoring device based on flow-through self-cleaning is characterized by being applied to the full spectrum neural network water quality monitoring method based on flow-through self-cleaning as claimed in any one of claims 1-6, and comprising a control reasoning unit, a spectrum acquisition unit and a self-cleaning unit, wherein an instruction input end of the control reasoning unit is used for acquiring external instruction input, a control signal output end of the control reasoning unit is respectively and electrically connected with a spectrum acquisition control signal input end of the spectrum acquisition unit and a flow cell cleaning control signal input end of the self-cleaning unit, a spectrum data transmitting end of the spectrum acquisition unit is electrically connected with a spectrum data receiving end of the control reasoning unit, and a flow cell marking output end of the self-cleaning unit is electrically connected with a flow cell marking input end of the control reasoning unit; the control reasoning unit is used for acquiring a measurement trigger instruction to generate a first spectrum acquisition control signal, controlling the spectrum acquisition unit to acquire a spectrum of a sample to be detected, receiving light absorption spectrum data of the sample to be detected to judge the existence of liquid, and generating a flow cell cleaning control signal to control the self-cleaning unit to perform flow cell self-cleaning operation; the spectrum acquisition unit is used for receiving the first spectrum acquisition control signal to emit a wide spectrum light beam, carrying out spectrum acquisition on a sample to be detected, obtaining sample light absorption line data to be detected, and sending the sample light absorption line data to be detected to the control reasoning unit; The self-cleaning unit is used for receiving the flow cell cleaning control signal so as to execute the flow cell self-cleaning operation; The control reasoning unit is further used for identifying the cleaned flow cell, generating a second spectrum acquisition control signal, controlling the spectrum acquisition unit to acquire a target sample spectrum, and receiving target sample absorbance spectrum data so as to calculate a target water quality parameter concentration value of the target sample through a full spectrum feedforward neural network reasoning model deployed in the control reasoning unit; The spectrum acquisition unit is further used for receiving the second spectrum acquisition control signal to emit a wide spectrum light beam, performing spectrum acquisition on the target sample to obtain target sample light absorption spectrum line data, and sending the target sample light absorption spectrum line data to the control reasoning unit.
8. The full spectrum neural network water quality monitoring device based on flow-through self-cleaning according to claim 7, which is characterized by comprising a spectrum acquisition unit, a self-cleaning unit and a control unit, wherein the spectrum acquisition unit comprises a xenon lamp, a collimating mirror, a light guide column and a spectrometer, and the self-cleaning unit comprises a motor, a transmission shaft and a brush piece; The input end of the xenon lamp is used as a spectrum acquisition control signal input end of the spectrum acquisition unit and is electrically connected with a control signal output end of the spectrum acquisition unit, the emergent opening of the xenon lamp is opposite to the incident opening of the light guide column, the collimating lens is arranged between the emergent opening of the xenon lamp and the incident opening of the light guide column, the emergent opening of the light guide column is opposite to the spectrum acquisition end of the spectrometer, and the output end of the spectrometer is used as a spectrum data transmitting end of the spectrum acquisition unit and is electrically connected with a spectrum data receiving end of the control reasoning unit; the controlled end of the motor is used as a flow cell cleaning control signal input end of the self-cleaning unit and is electrically connected with a control signal output end of the system reasoning unit, an output shaft of the motor is fixedly connected with the transmission shaft in a coaxial mode, and the transmission shaft is fixedly connected with the fixed end of the brush piece.
9. An electronic device, comprising a memory, a processor and a transceiver which are connected in turn in communication, wherein the memory is used for storing a computer program, the transceiver is used for receiving and transmitting a message, and the processor is used for reading the computer program and executing the full spectrum neural network water quality monitoring method based on the flow-through self-cleaning as claimed in any one of claims 1 to 6.
10. A computer program product comprising a computer program or instructions which, when executed by a computer, implement a flow-through self-cleaning based full spectrum neural network water quality monitoring method as claimed in any one of claims 1 to 6.

Description

Full spectrum neural network water quality monitoring method based on flow-through self-cleaning Technical Field The invention belongs to the technical field of water quality detection, and particularly relates to a full spectrum neural network water quality monitoring method based on flow-through self-cleaning. Background The water quality on-line monitoring technology is a core technology in the fields of environmental protection and water affairs. The traditional water quality monitoring mainly depends on a chemical analysis method, and a full-automatic online analyzer is adopted, but the problems of long analysis period, high cost and complex operation exist, and chemical reagents are needed to be used, so that secondary pollution is easy to cause. The full-spectrum water quality monitoring technology is rapidly developed due to the great potential of being rapid, nondestructive, free of chemical reagents and capable of detecting various parameters simultaneously, and becomes a hot spot for technical research. Compared with the traditional chemical method, the spectrum method can directly monitor the water body to be detected in situ or on line, greatly simplifies the flow, reduces the operation and maintenance cost and provides possibility for realizing wide-area and high-frequency water quality monitoring. The technology acquires the data rich ore containing a large amount of water quality information by collecting the continuous absorption spectrum of the water body in the ultraviolet, visible and near infrared wave bands (such as 200-900 nm), and the information quantity provided by a single or a plurality of wave length points is far exceeded. However, the conventional full-spectrum water quality monitoring device still uses absorption spectrums with limited frequency bands, and substitutes absorbance information of each frequency point to reversely calculate the concentration of a limited number of factors through a simple linear calibration or nonlinear fitting mode. When the water body structure of the simple model changes, the measured result has great deviation. In addition, the traditional full-spectrum water quality monitoring equipment is installed in situ and is directly put into water for use, when a water sample is turbid, measurement quality is seriously affected due to lack of necessary pretreatment, the equipment is usually deployed in the middle of a river and a lake, daily operation and maintenance are very troublesome, and the equipment is made into a shore extraction type, namely, the water is pumped up by a water pump, so that additional flow cells are needed, the cost is high, and the volume is large. At present, a flow type full-spectrum water quality detection device is arranged on the market, and a cleaning system is also arranged, but the cleaning period is scraped according to a preset fixed frequency, when a water intake is in water shortage (especially in a sewage pipe network occasion), the cleaning system still works according to a certain frequency, so that the service life of a motor is greatly shortened, and the cleaning system is particularly obvious in the use of a brush motor. From the foregoing, how to provide a full spectrum neural network water quality monitoring method based on flow-through self-cleaning, which can automatically perform liquid detection and self-cleaning to improve the service life of a brush motor, has become a problem to be solved in the art. Disclosure of Invention The invention aims to provide a full spectrum neural network water quality monitoring method based on flow-through self-cleaning, which is used for solving the problems in the prior art. In order to achieve the above purpose, the present invention adopts the following technical scheme: in a first aspect, the invention provides a method for monitoring water quality of a full spectrum neural network based on flow-through self-cleaning, which comprises the following steps: Acquiring preset air standard light absorption line data and a spectrum similarity threshold value, and acquiring a measurement trigger instruction to emit a wide spectrum light beam to a sample to be measured in a flow through tank, and performing spectrum acquisition of the sample to be measured to obtain the light absorption line data of the sample to be measured, wherein the air standard light absorption line data and the light absorption line data of the sample to be measured are a group of discrete vector sequences; obtaining a preset similarity function, calculating the to-be-detected spectrum similarity between the to-be-detected sample light absorption line data and the air standard light absorption line data by using the similarity function, and judging the presence of liquid by using the spectrum similarity threshold value to obtain a liquid presence judging result; Acquiring a preset self-cleaning period of the flow cell and a self-cleaning operation log of the flow cell, and executing the self-cleanin