A rapid detection method for total potassium in coconut shell charcoal substrate based on spectral feature information combination

Abstract

The application discloses a kind of based on spectral feature information joint's coconut shell substrate total potassium rapid detection method.The method comprises the following steps: coconut shell substrate sample preparation and spectral data acquisition;Sample total potassium content physical and chemical reference value determination;Eliminate the dimensional difference of spectral data;Respectively from LIBS spectral curve and NIRS spectral curve, extract spectral feature, generate joint spectral feature;Construct the mapping function of coconut shell substrate total potassium rapid detection based on spectral feature information joint data;The total potassium content of coconut shell substrate to be measured is rapidly optically detected using the constructed detection mapping function.The rapid high-precision stable detection of the total potassium content of coconut shell substrate is realized, and the complexity of detection operation is greatly simplified, and the detection speed is faster.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of rapid optical detection of soilless culture substrate composition, and more particularly relates to a rapid detection method for total potassium in coconut coir substrate based on joint spectral characteristic information. BACKGROUND

[0002] Coconut coir substrate is a typical soilless culture substrate, which has the advantages of water saving, fertilizer saving and renewable, and is widely used in facility agriculture in China. At present, the water and fertilizer integration form is generally used in the soilless culture of facility agriculture to supply nutrients and water. Scientific and sufficient supply of water and fertilizer integration is the key to guarantee the yield and quality of coconut coir substrate cultivation. Therefore, rapid and accurate acquisition of nutrient and water content information of the cultivation substrate is of great significance for realizing accurate supply of nutrient elements in substrate cultivation and reducing agricultural non-point source pollution.

[0003] Potassium is one of the three most needed nutrients for plant growth, which plays a crucial role in crop growth. Potassium plays an important role in promoting photosynthesis, photosynthetic product transport and protein synthesis, and the abundance and deficiency of potassium directly affect the stress resistance, quality and yield of crops. The traditional chemical analysis method is used to measure the potassium element in coconut coir substrate, which has a complicated operation process, a time-consuming detection process and environmental pollution of detection reagents, and its real-time performance is difficult to meet the actual demand. At present, some researchers try to use advanced photoelectric detection technology to detect soil and substrate composition. For example, Laser-induced Breakdown Spectroscopy (LIBS) and Near Infrared Spectroscopy (NIRS) are two common spectral technologies applied to explore the rapid detection of soil composition.

[0004] Since NIRS spectrum is mainly used to detect substances containing C-H, N-H and O-H functional groups, it is difficult to directly and rapidly measure the potassium content in coconut coir substrate, but NIRS spectrum can well be used to characterize the substrate composition change information of coconut coir substrate sample. LIBS can theoretically realize the rapid measurement of the potassium content in coconut coir substrate, but its detection accuracy and stability are difficult to guarantee due to the influence of sample matrix effect and self-absorption effect. In summary, it is difficult to realize the rapid and accurate measurement of the potassium content in coconut coir substrate by using LIBS or NIRS spectrum alone. By combining LIBS and NIRS spectrum, the total potassium content in coconut coir substrate is characterized from the perspectives of atomic and molecular spectral characteristics, respectively, so as to fully utilize the complementary advantages of LIBS and NIRS spectrum, and realize the rapid and high-precision and stable measurement of the total potassium content in coconut coir substrate. SUMMARY

[0005] In view of defects of the related art, the present application aims to provide a coconut shell substrate total potassium rapid detection method based on spectral feature information combination, aiming to solve the problems of single LIBS spectral detection of coconut shell substrate total potassium content affected by matrix effect and self-absorption effect, resulting in poor detection precision and stability, and NIRS unable to directly realize coconut shell substrate total potassium measurement.

[0006] To achieve the above-mentioned purpose, the present application provides a coconut shell substrate total potassium rapid detection method based on spectral feature information combination, comprising:

[0007] Training stage:

[0008] S1, preparing a plurality of coconut shell substrate test samples, and acquiring LIBS and NIRS original spectral data of the test samples;

[0009] S2, performing normalization processing on the LIBS original spectral data;

[0010] S3, extracting LIBS spectral features from the normalized LIBS spectral data, extracting NIRS spectral features from the NIRS original spectral data, and arranging the LIBS spectral features before the NIRS spectral features to generate joint spectral features;

[0011] S4, constructing a mapping relationship between the joint spectral features and determination data of total potassium content corresponding thereto;

[0012] Wherein, the determination data of total potassium content is obtained by chemical determination on the test samples;

[0013] Application stage:

[0014] Obtaining joint spectral features of a coconut shell substrate sample to be measured, and obtaining total potassium content of the coconut shell substrate sample to be measured according to the mapping relationship.

[0015] Optionally, step S3 comprises the following steps:

[0016] S31, extracting LIBS spectral features from the normalized LIBS spectral data by using a two-dimensional correlation spectral analysis method, the LIBS spectral features being fingerprint spectral line features of potassium element LIBS detection, the fingerprint spectral line features being two, and the corresponding wavelengths being 766.49 nm and 769.90 nm, respectively;

[0017] S32, extracting NIRS spectral features from the NIRS raw spectral data by using a variable screening algorithm, the NIRS spectral features being 7, and their corresponding wavelengths being 981.99 nm, 1034.07 nm, 1137.61 nm, 1183.37 nm, 1240.33 nm, 1426.47 nm and 1487.90 nm respectively;

[0018] S33, arranging the two LIBS spectral features obtained in S31 in the front, and arranging the seven NIRS spectral features obtained in S32 in turn behind the LIBS spectral features, to generate a matrix expressed joint spectral feature.

[0019] Optionally, after step S3, the following steps are further included:

[0020] S31', sorting the joint spectral feature data of all the test samples of the coconut husk matrix according to the size of the total potassium content of each test sample in the determination data;

[0021] S32', dividing the joint spectral feature data into calibration set data and validation set data according to the concentration gradient method at a ratio of 3:1.

[0022] Optionally, step S4 includes the following steps:

[0023] S41, constructing a mapping relationship between the calibration set data and the determination data of the corresponding total potassium content by using a partial least squares regression algorithm PLSR;

[0024] S42, verifying the mapping relationship by using the validation set data and the determination data of the corresponding total potassium content.

[0025] Optionally, the mapping relationship is:

[0026] y = 0.0723x 766.49 -0.0719x 769.90 -20.6190x 981.99 -26.8511x 1034.07

[0027] -8.7220x 1137.61 +23.0916x 1183.37 +9.3640x 1240.33

[0028] +30.6529x 1426.47 -6.4509x 1487.90 -0.0203

[0029] wherein y represents the total potassium content of the coconut husk matrix sample, x 766.49LIBS spectral data representing wavelength 766.49 nm, x 769.90 LIBS spectral data representing wavelength 769.90 nm, x 981.99 NIRS spectral data representing wavelength 981.99 nm, x 1034.07 NIRS spectral data representing wavelength 1034.07 nm, x 1137.61 NIRS spectral data representing wavelength 1137.61 nm, x 1183.37 NIRS spectral data representing wavelength 1183.37 nm, x 1240.33 NIRS spectral data representing wavelength 1240.33 nm, x 1426.47 NIRS spectral data representing wavelength 1426.47 nm, x 1487.90 NIRS spectral data representing wavelength 1487.90 nm.

[0030] Optionally, the step S1 comprises the following steps:

[0031] S11, drying, crushing, sieving and compression treatment are performed on the collected coconut husk matrix sample, and the prepared multiple coconut husk matrix test samples are unified through compression; wherein the compression sample preparation mode includes two forms of boric acid embedded bottom sample preparation and separate sample powder compression sample preparation;

[0032] S12, the LIBS atomic emission spectrum acquisition system and the NIRS diffuse reflectance spectrum acquisition system are used respectively to collect spectral data of the test samples, and LIBS original spectral data and NIRS original spectral data are obtained respectively.

[0033] Optionally, the total potassium content of all test samples is determined by a chemical method according to a national forestry industry standard.

[0034] Optionally, the variable screening algorithm is a continuous projection algorithm.

[0035] Optionally, the joint spectral feature of the to-be-tested coconut husk matrix sample is obtained, and the total potassium content of the to-be-tested coconut husk matrix sample is obtained according to the mapping relationship, comprising:

[0036] The to-be-tested coconut husk matrix is dried, crushed, sieved and compression sampled to obtain a to-be-tested coconut husk matrix sample;

[0037] The LIBS spectral data and the NIRS spectral data of the to-be-tested coconut husk matrix sample are obtained;

[0038] The obtained LIBS spectral data is normalized;

[0039] extract the LIBS spectral feature and the NIRS spectral feature of the coconut shell matrix sample, and generate a joint spectral feature of the coconut shell matrix sample;

[0040] The joint spectral feature of the coconut shell matrix sample is substituted into the mapping relationship, and the potassium element content of the coconut shell matrix sample is quickly obtained.

[0041] Compared with the prior art, the above technical scheme of the present application has the following beneficial effects:

[0042] 1. The coconut shell matrix total potassium rapid detection method based on joint spectral feature information provided by the present application uses the joint LIBS and NIRS spectral feature information to construct the mapping relationship between the spectral data and the coconut shell matrix total potassium content, and performs rapid optical detection of the total potassium content of the sample to be detected. The NIRS spectrum provides rich matrix component information of the sample, effectively reduces the interference of the coconut shell matrix sample matrix effect on the LIBS detection, and the joint LIBS and NIRS spectral feature information fully utilizes the advantages of the LIBS and NIRS spectral detection technologies, and has the advantages of fast detection speed, high detection precision, and stable detection performance.

[0043] 2. The coconut shell matrix total potassium rapid detection method based on joint spectral feature information provided by the present application can maximize the reduction of the influence of self-absorption effect on the coconut shell matrix total potassium content detection through the combination of LIBS and NIRS spectrum, thereby improving the robustness and universality of the coconut shell matrix total potassium content spectral detection method.

[0044] 3. The coconut shell matrix total potassium rapid detection method based on joint spectral feature information provided by the present application extracts the LIBS and NIRS spectral feature information of the coconut shell matrix sample, comprehensively characterizes the coconut shell matrix total potassium and sample matrix composition information from the atomic and molecular levels, and greatly improves the anti-interference ability of the method. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 is a flowchart of the coconut shell matrix total potassium rapid detection method based on joint spectral feature information provided by the present application;

[0046] Figure 2 is a schematic diagram of the coconut shell matrix LIBS spectral curve and total potassium feature provided by the present application;

[0047] Figure 3 is a schematic diagram of the coconut shell matrix NIRS spectral curve and total potassium feature provided by the present application. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0049] The following description, in conjunction with a preferred embodiment, illustrates the content involved in the above embodiments.

[0050] like Figure 1 As shown, a rapid method for detecting total potassium in coconut coir matrix based on combined spectral feature information includes:

[0051] Training phase:

[0052] S1. Prepare multiple test samples of coconut coir matrix and obtain the raw LIBS and NIRS spectral data of the test samples;

[0053] S2. Normalize the raw LIBS spectral data;

[0054] S3. Extract LIBS spectral features from the normalized LIBS spectral data, extract NIRS spectral features from the original NIRS spectral data, and arrange the LIBS spectral features before the NIRS spectral features to generate joint spectral features;

[0055] S4. Construct a mapping relationship between the joint spectral characteristics and the corresponding total potassium content determination data; wherein the total potassium content determination data is obtained by chemical determination of the test sample;

[0056] Application phase:

[0057] The joint spectral characteristics of the coconut coir matrix sample to be tested are obtained, and the total potassium content of the coconut coir matrix sample to be tested is obtained according to the mapping relationship.

[0058] Step S1 includes the following steps:

[0059] S11. The collected coconut coir matrix samples are dried, crushed, sieved and compressed. The compression process is used to make the test samples of multiple coconut coir matrices uniform. The compression sample preparation methods include two forms: boric acid embedding sample preparation and individual sample powder compression sample preparation.

[0060] S12. The spectral data of the test sample are acquired using the LIBS atomic emission spectroscopy acquisition system and the NIRS diffuse reflectance spectroscopy acquisition system, respectively, to obtain the raw LIBS spectral data and the raw NIRS spectral data.

[0061] In the embodiment, the spectral data of the test sample is acquired by a spectral acquisition system, which includes a LIBS atomic emission spectral acquisition system and a NIRS diffuse reflection spectral acquisition system. In step S11, the compression pressure and the pressure maintaining time of the compression molding are 28 MPa and 60 s respectively; in step S12, the wavelength acquisition range of the LIBS spectral acquisition system is 200-860 nm, and the wavelength acquisition range of the NIRS spectral acquisition system is 940-1650 nm; during the spectral acquisition, the two systems can be used to acquire the spectral data respectively, or the structure of the spectral acquisition system can be optimized to realize the synchronous acquisition of the two kinds of spectral data.

[0062] In step S2, the mathematical algorithm (normalization algorithm) is used to convert the peak intensity of all LIBS spectral curves into 0-1, eliminate the difference in the dimension of the two kinds of spectral data, and avoid the influence of the difference in the order of magnitude of the two kinds of spectral data on the spectral combination.

[0063] Optionally, step S3 includes the following steps:

[0064] S31, the LIBS spectral features are extracted from the normalized LIBS spectral data by using a two-dimensional correlation spectral analysis method, the LIBS spectral features are the fingerprint spectral line features of the potassium element LIBS detection, and the fingerprint spectral line features are two, and the corresponding wavelengths are 766.49 nm and 769.90 nm respectively;

[0065] S32, the NIRS spectral features are extracted from the NIRS original spectral data by using a variable screening algorithm, the NIRS spectral features are seven, and the corresponding wavelengths are 981.99 nm, 1034.07 nm, 1137.61 nm, 1183.37 nm, 1240.33 nm, 1426.47 nm and 1487.90 nm respectively;

[0066] S33, the two LIBS spectral features acquired in S31 are arranged in front, and the seven NIRS spectral features acquired in S32 are arranged in turn behind the LIBS spectral features, to generate the matrix expression of the combined spectral features.

[0067] Among them, the variable screening algorithm is a successive projection algorithm (SPA), which is a forward variable selection algorithm that minimizes the collinearity of vector space, can extract several characteristic wavelengths in the full wavelength range, eliminates the redundant information in the original spectral matrix, and can be used for screening of spectral characteristic wavelengths, and selects effective wavelengths when detecting the content of some important components in crops and food. When arranging the LIBS spectral features and the NIRS spectral features, they can be arranged in turn from small to large according to their corresponding wavelengths.

[0068] Optionally, step S3 further comprises the following steps:

[0069] S31', according to the size of the total potassium content of each test sample in the determination data, the combined spectral feature data of all test samples of the coconut coir substrate is sorted;

[0070] S32', according to the concentration gradient method, the combined spectral feature data is divided into calibration set data and validation set data according to the ratio of 3:1.

[0071] The total potassium content reference value of the coconut coir substrate sample is determined by the national forestry industry standard (Determination of potassium in forest soil: LY / T 1234-2015), and the total potassium content obtained is a more accurate value. In this embodiment, according to the total potassium content of each test sample in the determination data, the combined spectral feature data of all test samples of the coconut coir substrate is sorted according to the total potassium content value from large to small. The LIBS spectral curve is shown in Figure 2 , and the NIRS spectral curve is shown in Figure 3 with five-star marks.

[0072] Optionally, step S4 comprises the following steps:

[0073] S41, a partial least squares regression algorithm PLSR is used to construct the mapping relationship between the calibration set data and the determination data of the corresponding total potassium content;

[0074] S42, the validation set data and the determination data of the corresponding total potassium content are used for validation, and a mathematical mapping relationship for rapid detection of the total potassium of the coconut coir substrate is constructed.

[0075] The determination data of the total potassium content is obtained by chemical determination of the test samples, and the total potassium content of all test samples is chemically determined by the detection method of the national forestry industry standard (Determination of potassium in forest soil: LY / T 1234-2015).

[0076] The constructed LIBS and NIRS spectral prediction model is: a partial least squares regression algorithm (PLSR) is used to construct the mathematical mapping relationship between the combined spectral data of the sample calibration set and its total potassium physicochemical reference value, and the validation set spectral data is used for mapping relationship validation, thereby constructing the best mapping relationship, i.e. the coconut coir substrate total potassium rapid detection mapping function.

[0077] Optionally, the coconut coir substrate total potassium rapid detection mapping relationship is:

[0078] y = 0.0723x 766.49 -0.0719x 769.90-20.6190x 981.99 -26.8511x 1034.07

[0079] -8.7220x 1137.61 +23.0916x 1183.37 +9.3640x 1240.33

[0080] +30.6529x 1426.47 -6.4509x 1487.90 -0.0203

[0081] wherein y represents the total potassium content of the coconut coir substrate sample, x 766.49 represents the LIBS spectral data at a wavelength of 766.49 nm, x 769.90 represents the LIBS spectral data at a wavelength of 769.90 nm, x 981.99 represents the NIRS spectral data at a wavelength of 981.99 nm, x 1034.07 represents the NIRS spectral data at a wavelength of 1034.07 nm, x 1137.61 represents the NIRS spectral data at a wavelength of 1137.61 nm, x 1183.37 represents the NIRS spectral data at a wavelength of 1183.37 nm, x 1240.33 represents the NIRS spectral data at a wavelength of 1240.33 nm, x 1426.47 represents the NIRS spectral data at a wavelength of 1426.47 nm, x 1487.90 represents the NIRS spectral data at a wavelength of 1487.90 nm.

[0082] Optionally, the obtaining the joint spectral feature of the coconut coir substrate sample to be measured, and obtaining the total potassium content of the coconut coir substrate sample to be measured according to the mapping relationship, comprises:

[0083] drying, crushing, sieving and compressing the coconut coir substrate to be measured to obtain the coconut coir substrate sample to be measured;

[0084] obtaining the LIBS spectral data and the NIRS spectral data of the coconut coir substrate sample to be measured;

[0085] performing normalization processing on the obtained LIBS spectral data;

[0086] extracting the LIBS spectral feature and the NIRS spectral feature of the coconut coir substrate sample to be measured, and generating the joint spectral feature of the coconut coir substrate sample to be measured;

[0087] substituting the joint spectral feature of the coconut coir substrate sample to be measured into the mapping relationship to quickly obtain the total potassium content of the coconut coir substrate sample to be measured.

[0088] The application discloses a coconut shell matrix total potassium rapid detection method based on LIBS and NIRS spectral feature information jointing. The method comprises the following steps: drying and standard screen are used for pretreatment of collected samples, so that the influence of soil moisture content and particle difference on spectral detection is eliminated, and a tablet press is used for preparation of standardized samples for detection; LIBS and NIRS spectral acquisition systems are used for acquisition of original LIBS and NIRS spectral data of samples; meanwhile, a national forestry industry standard (determination of forest soil potassium: LY / T 1234-2015) is used for determination of reference values of total potassium content of soil samples; data normalization is used for processing of the acquired LIBS data, so that the influence of the dimension of original LIBS data on spectral feature information jointing is eliminated; algorithm is used for extraction of LIBS fingerprint spectrum features and NIRS spectral features of the coconut shell matrix total potassium detection, and the LIBS spectral features are arranged in front of the NIRS spectral features, so that the LIBS and NIRS spectral feature information jointing is realized; according to the LIBS and NIRS spectral feature information jointing strategy, LIBS and NIRS spectral feature information jointing data of all samples are acquired; all sample spectral jointing data are divided into a calibration set and a validation set according to the reference values of total potassium content of samples, a mapping relationship between the LIBS and NIRS spectral feature information jointing data and the total potassium content of the coconut shell matrix is established by using the spectral data of the calibration set, and the mapping relationship is verified by using the spectral data of the validation set; finally, the mapping function for rapid optical detection of the total potassium content of the coconut shell matrix is constructed by using the LIBS and NIRS spectral feature information jointing, and the mapping function has the advantages of fast detection speed, high detection precision and stable detection performance.

[0089] The following is described by taking the total potassium detection of 84 coconut shell matrix samples as an example, and the specific steps are as follows:

[0090] S1, an electric heating constant temperature air drying oven is used for drying treatment of the 84 coconut shell matrix samples, a small multifunctional crushing and grinding machine is used for crushing treatment of the 84 dried coconut shell matrix samples, and a full-automatic tablet press is used for compression sample preparation, and the compression pressure and pressure maintaining time of the full-automatic tablet press are set to 28 MPa and 60 s respectively.

[0091] LIBS and NIRS spectral acquisition systems are used respectively to acquire LIBS emission spectrum data and NIRS diffuse reflection spectrum data of the 84 compressed coconut shell matrix samples prepared above.

[0092] S2, a normalization mathematical algorithm is used for processing of the acquired LIBS spectral data, so that the peak intensity of all LIBS spectral curves is converted into 0-1, the dimension difference of the LIBS and NIRS spectral data is eliminated, and the influence of the order of magnitude difference of the two kinds of spectral data on spectral jointing is avoided.

[0093] S3、According to the atomic spectral database and two-dimensional correlation spectral analysis, the total potassium detection fingerprint line characteristics are extracted from the LIBS spectral data of the coconut shell matrix in the wavelength range of 200-860 nm, and two fingerprint line characteristics of 766.49 nm and 769.90 nm are obtained; the total potassium detection spectral characteristics are extracted from the NIRS spectral data of the coconut shell matrix in the wavelength range of 940-1650 nm by using SPA, and seven NIRS spectral characteristics are obtained, which are: 981.99 nm, 1034.07 nm, 1137.61 nm, 1183.37 nm, 1240.33 nm, 1426.47 nm, and 1487.90 nm.

[0094] The two LIBS fingerprint line characteristics obtained in the above step are arranged in the front, and the seven NIRS spectral characteristics of the coconut shell matrix extracted in the above step are arranged in the back, so as to realize the combination of the LIBS and NIRS spectral characteristic information of the coconut shell matrix.

[0095] S3', according to the spectral characteristic information combination method in step S3, the LIBS and NIRS spectral characteristic data of each coconut shell matrix sample are extracted, and the extracted LIBS and NIRS spectral characteristics are combined to form a combined spectral characteristic data matrix.

[0096] The chemical analysis method specified in the national forestry industry standard (Determination of forest soil potassium: LY / T 1234-2015) is used to determine the total potassium content reference value of the 84 coconut shell matrix samples from which the spectral data is obtained in step S1.

[0097] According to the measured total potassium physicochemical reference value content of the coconut shell matrix sample, the spectral characteristic information combined data of all coconut shell matrix samples are sorted, and the combined spectral data is divided into a calibration set and a validation set according to the concentration gradient method at a ratio of 3:1, wherein the calibration set contains 63 coconut shell matrix sample spectral data, and the validation set contains 21 coconut shell matrix sample spectral data.

[0098] S4, a PLSR is used to construct the mathematical mapping relationship between the spectral characteristic information combined data of the sample calibration set in step S3' and the measured total potassium physicochemical reference value of the calibration set sample, and the validation set spectral data and the validation set total potassium physicochemical reference value in step S3' are used for verification, so as to construct the best mapping relationship between the coconut shell matrix spectral data and the total potassium content. The constructed mapping relationship is: y = 0.0723x 766.49 -0.0719x 769.90 -20.6190x 981.99 -26.8511x 1034.07 -8.7220x 1137.61 +23.0916x 1183.37+ 9.3640x 1240.33 + 30.6529x 1426.47 - 6.4509x 1487.90 - 0.0203, the result of the optimal mapping relationship is shown in Table 1.

[0099] Table 1 result of the optimal mapping relationship of the total potassium detection of the coconut shell substrate based on the joint of the LIBS and NIRS spectral feature information

[0100]

[0101]

[0102] According to the relevant standards of evaluation, the mapping relationship of the spectral data with the optimal calculation result and the total potassium content should require as large as possible correlation coefficient (calibration set R c , validation set R v ) and as small as possible root mean square error (calibration set RMSEC, validation set RMSEV). As shown in Table 1, the two LIBS fingerprint spectral features and the seven NIRS spectral features extracted from the coconut shell substrate are combined for spectral feature information joint, and then the joint spectral feature information of LIBS and NIRS can be used to establish a mapping relationship between the joint spectral feature and the total potassium content with good detection performance.

[0103] After obtaining the mapping relationship between the joint spectral feature and the total potassium content, in the application stage:

[0104] According to step S1, the coconut shell substrate sample to be measured is pretreated by drying and crushing, and is compressed for sample preparation. The LIBS and NIRS spectral data of the sample to be measured are quickly obtained, and the LIBS data are normalized. The LIBS fingerprint spectral feature and the NIRS spectral feature of the total potassium of the coconut shell substrate are extracted according to step S3, respectively. The strategy in step S3 is used for the joint of the LIBS and NIRS spectral feature information. The joint spectral data are substituted into the mapping relationship constructed in step S4, so as to obtain the total potassium content of the coconut shell substrate to be measured, and the rapid and high-precision LIBS and NIRS joint detection of the total potassium content of the coconut shell substrate is realized.

[0105] This invention discloses a rapid detection method for total potassium in coconut coir matrix based on combined LIBS and NIRS spectral characteristic information. The method involves pre-treating the collected samples by drying and using a standard sieve to eliminate the influence of soil moisture content and particle size differences on spectral detection. Standardized test samples are then prepared using a tablet press for testing. The original LIBS and NIRS spectral data of the samples are acquired using LIBS and NIRS spectral acquisition systems, respectively. Simultaneously, the method adopts the national forestry industry standard (Determination of Potassium in Forest Soils: LY / T). (1234-2015) The reference values ​​of total potassium content in soil samples were determined; the collected LIBS data were processed by data normalization to eliminate the influence of the original LIBS data dimensions on the joint spectral feature information; LIBS fingerprint spectral features and NIRS spectral features for total potassium detection in coconut coir matrix were extracted separately using an algorithm, and the LIBS features were arranged before the NIRS features to achieve joint LIBS and NIRS spectral feature information; according to the joint LIBS and NIRS spectral feature information strategy, the joint LIBS and NIRS feature information data of all samples were obtained; the joint spectral data of all samples were divided into a calibration set and a validation set according to the reference values ​​of total potassium content in the samples; a fast optical detection mapping function for total potassium in coconut coir matrix based on the joint LIBS and NIRS spectral feature information was established using the calibration set spectral data; the performance of the established mapping function was verified using the validation set spectral data, and the correlation coefficient R of the calibration set was measured. c Correlation coefficient R with the validation set v The root mean square errors (RMSEC) of the calibration set and the root mean square error (RMSEV) of the validation set were 0.9909 and 0.9883, respectively, while the RMSEC and RMSEV of the validation set were 1.2068 g / kg and 1.3862 g / kg, respectively. This method utilizes LIBS and NIRS spectral feature information to jointly construct a detection mapping function for total potassium content in coconut coir matrix. This achieves rapid and high-precision detection of potassium in coconut coir matrix while significantly simplifying the complexity of the detection mapping function and improving the detection response speed.

[0106] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for rapid detection of total potassium in coconut shell based on spectral feature information combination, characterized in that, The application relates to a method for determining the total potassium content of coconut shell substrate. The method comprises the following steps: S1, preparing a plurality of test samples of coconut shell substrate and acquiring LIBS and NIRS original spectrum data of the test samples; S2, performing normalization processing on the LIBS original spectrum data; S3, extracting LIBS spectrum features from the normalized LIBS spectrum data, extracting NIRS spectrum features from the NIRS original spectrum data, arranging the LIBS spectrum features in front of the NIRS spectrum features, and generating joint spectrum features; S4, constructing a mapping relationship between the joint spectrum features and the determination data of the total potassium content corresponding to the joint spectrum features; The determination data of the total potassium content is obtained by performing chemical determination on the test samples; In the application stage, the joint spectrum features of a coconut shell substrate sample to be measured are acquired, and the total potassium content of the coconut shell substrate sample to be measured is obtained according to the mapping relationship. The step S1 comprises the following steps: S11, drying, crushing, sieving and compressing the collected coconut shell substrate sample, and unifying the plurality of test samples of coconut shell substrate prepared through compression; the compression sample preparation mode comprises two forms of boric acid embedded bottom sample preparation and separate sample powder compression sample preparation; S12, performing spectrum data acquisition on the test samples by using a LIBS atomic emission spectrum acquisition system and a NIRS diffuse reflection spectrum acquisition system respectively, and obtaining LIBS original spectrum data and NIRS original spectrum data respectively. The step S3 comprises the following steps:

2. The method of claim 1, wherein, S31, extracting LIBS spectrum features from the normalized LIBS spectrum data by using a two-dimensional correlation spectrum analysis method, wherein the LIBS spectrum features are fingerprint spectrum line features of potassium element LIBS detection, and the fingerprint spectrum line features are two, and the corresponding wavelengths are 766.49 nm and 769.90 nm respectively; S32, extracting NIRS spectrum features from the NIRS original spectrum data by using a variable screening algorithm, wherein the NIRS spectrum features are seven, and the corresponding wavelengths are 981.99 nm, 1034.07 nm, 1137.61 nm, 1183.37 nm, 1240.33 nm, 1426.47 nm and 1487.90 nm respectively; S33, arranging the two LIBS spectrum features obtained in S31 in front, arranging the seven NIRS spectrum features obtained in S32 in turn behind the LIBS spectrum features, and generating matrix expression joint spectrum features. The step S3 further comprises the following steps:

3. The method of claim 1 or 2, wherein, S31', sorting the joint spectrum feature data of all test samples of coconut shell substrate according to the total potassium content of each test sample in the determination data; S32', dividing the joint spectrum feature data into calibration set data and verification set data according to a concentration gradient method at a ratio of 3:

1. The step S4 comprises the following steps:

4. The method of claim 3, wherein, S41, constructing a mapping relationship between the calibration set data and the determination data of the total potassium content corresponding to the calibration set data by using a partial least squares regression algorithm PLSR; ​ S42, verifying the mapping relationship by using the verification set data and the determination data of the corresponding total potassium content.

5. The method of claim 1 or 2, wherein, The mapping relationship is: wherein, y represents the total potassium content of the coconut coir substrate sample, represents the LIBS spectral data at a wavelength of 766.49 nm, represents the LIBS spectral data at a wavelength of 769.90 nm, represents the NIRS spectral data at a wavelength of 981.99 nm, represents the NIRS spectral data at a wavelength of 1034.07 nm, represents the NIRS spectral data at a wavelength of 1137.61 nm, represents the NIRS spectral data at a wavelength of 1183.37 nm, represents the NIRS spectral data at a wavelength of 1240.33 nm, represents the NIRS spectral data at a wavelength of 1426.47 nm, represents the NIRS spectral data at a wavelength of 1487.90 nm.

6. The method of claim 1 or 2, wherein, The total potassium content of all test samples is chemically determined by using a detection method of a national forestry industry standard.

7. The method of claim 2, wherein, The variable screening algorithm is a continuous projection algorithm.

8. The method of claim 1, wherein, The method comprises the following steps: The coconut coir substrate to be tested is dried, crushed, sieved and compressed to obtain a coconut coir substrate sample; LIBS spectrum data and NIRS spectrum data of the coconut coir substrate sample to be tested are obtained; The obtained LIBS spectrum data are normalized; LIBS spectrum features and NIRS spectrum features of the coconut coir substrate sample to be tested are extracted, and joint spectrum features of the coconut coir substrate sample to be tested are generated; The joint spectrum features of the coconut coir substrate sample to be tested are substituted into the mapping relationship, and the potassium content of the coconut coir substrate sample to be tested is quickly obtained.

Images