Online and offline fast detection sample representativeness consistency evaluation method and device

Abstract

The invention relates to an online and offline fast detection sample representativeness consistency evaluation method and device, and the method comprises the steps: collecting online fast detection sample detection data and offline fast detection sample detection data of a target coal batch; performing high-dimensional feature extraction on the coal quality detection indexes of the online fast detection sample detection data and the offline fast detection sample detection data by using a multi-layer neural network to obtain an online fast detection sample feature vector set and an offline fast detection sample feature vector set; and respectively calculating the maximum average difference between the online fast detection sample feature vector set and the offline fast detection sample feature vector set by using a pre-constructed learnable kernel function, and obtaining an evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference. Therefore, the technical problems that the consistency difference between online and offline detection results cannot be effectively evaluated, the subjectivity is high, the difference cannot be quantified, and small samples are weak when the representativeness of the coal quality detection result is judged by adopting average value or rule sampling in the related technology are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal detection and intelligent analysis, and in particular relates to an online and offline fast detection sample representativeness consistency evaluation method and device. BACKGROUND

[0002] Coal, as the main energy and industrial raw material in the world, its quality detection is of great significance in the fields of energy security, environmental protection, industrial production and trade fairness.

[0003] In the related art, the coal quality detection adopts average value or rule sampling to judge the representativeness, which has strong subjectivity, cannot quantify the difference, and is powerless to small samples, and cannot effectively evaluate the consistency difference between online and offline detection results, and needs to be improved. SUMMARY

[0004] The present application provides an online and offline fast detection sample representativeness consistency evaluation method and device to solve the technical problems in the related art that the average value or rule sampling is used to judge the representativeness of the coal quality detection result, which cannot effectively evaluate the consistency difference between online and offline detection results, has strong subjectivity, cannot quantify the difference, and is powerless to small samples.

[0005] The first aspect embodiment of the present application provides an online and offline fast detection sample representativeness consistency evaluation method, comprising the following steps: collecting online fast detection sample detection data and offline fast detection sample detection data of a target coal batch; using a multi-layer neural network to perform high-dimensional feature extraction on coal quality detection indexes of the online fast detection sample detection data and the offline fast detection sample detection data respectively, to obtain an online fast detection sample feature vector set and an offline fast detection sample feature vector set; using a pre-constructed learnable kernel function to calculate maximum average differences of the online fast detection sample feature vector set and the offline fast detection sample feature vector set respectively, and obtaining an evaluation result of sample representativeness consistency of the target coal batch based on the maximum average differences.

[0006] Based on the above technical solution, the embodiments of the present application can use a pre-constructed learnable kernel function to calculate maximum average differences of the online fast detection sample feature vector set and the offline fast detection sample feature vector set respectively to cope with the difficulty of representativeness judgment caused by uneven sample quantity, and obtain an evaluation result of sample representativeness consistency of the target coal batch based on the maximum average differences, and complete the determination of whether the coal representativeness is consistent under two detection modes through the quantitative comparison of the consistency of online (multiple sub-samples) and offline (few sub-samples) fast detection results in statistical distribution.

[0007] Optionally, in an embodiment of the present application, before the maximum average discrepancy of the online rapid inspection sample feature vector set and the offline rapid inspection sample feature vector set is calculated respectively by using the pre-constructed learnable kernel function, the method further comprises: obtaining an online sample feature vector set and an offline sample feature vector set for training; inputting the online sample feature vector set and the offline sample feature vector set into an initial learnable kernel function, calculating the mapping of each sample to obtain a mapped online sample feature vector set and a mapped offline sample feature vector set; calculating the similarity of all sample pairs by using the mapped online sample feature vector set and the mapped offline sample feature vector set; calculating the loss value of the maximum average discrepancy by using the similarity, and optimizing the initial learnable kernel function by using the loss value to obtain the learnable kernel function.

[0008] Based on the above technical solution, the training mechanism of the embodiment of the present application can optimize feature mapping and similarity measurement through end-to-end, and solve the core pain points of complex coal quality data distribution and poor adaptability of traditional kernel functions.

[0009] Optionally, in an embodiment of the present application, the optimizing the initial learnable kernel function by using the loss value to obtain the learnable kernel function comprises: calculating the gradient of the loss value with respect to the similarity; optimizing the initial learnable kernel function by using the gradient until a preset iteration stopping condition is reached to obtain the learnable kernel function.

[0010] Based on the above technical solution, the embodiment of the present application can optimize the function by using the loss value to improve the function accuracy and robustness, so that the final consistency evaluation result has high accuracy.

[0011] Optionally, in an embodiment of the present application, the method further comprises: based on the evaluation result, judging whether the online rapid inspection sample detection data and the offline rapid inspection sample detection data satisfy a preset representative deviation condition; if the preset representative deviation condition is not satisfied, generating a corresponding process adjustment suggestion based on the online rapid inspection sample detection data and the offline rapid inspection sample detection data.

[0012] Based on the above technical solution, the embodiment of the present application can generate a process adjustment suggestion according to the evaluation result, realize the active prevention and control of quality risks, and convert abstract data differences into specific engineering actions.

[0013] Optionally, in an embodiment of the present application, the coal quality detection index comprises at least one of an ash content index, a moisture content index, a calorific value index and a sulfur content index of the target coal batch.

[0014] The second aspect embodiment of the application provides an online and offline fast inspection sample representativeness consistency evaluation device, comprising: a collection module configured to collect online fast inspection sample detection data and offline fast inspection sample detection data of a target coal batch; an extraction module configured to perform high-dimensional feature extraction on coal quality detection indexes of the online fast inspection sample detection data and the offline fast inspection sample detection data by using a multi-layer neural network, to obtain an online fast inspection sample feature vector set and an offline fast inspection sample feature vector set; and an evaluation module configured to calculate maximum average differences of the online fast inspection sample feature vector set and the offline fast inspection sample feature vector set by using a pre-constructed learnable kernel function, and obtain an evaluation result of sample representativeness consistency of the target coal batch based on the maximum average differences.

[0015] Optionally, in an embodiment of the application, the device further comprises: an acquisition module configured to acquire an online sample feature vector set and an offline sample feature vector set for training; a first calculation module configured to input the online sample feature vector set and the offline sample feature vector set into an initial learnable kernel function, calculate a mapping of each sample, to obtain a mapped online sample feature vector set and a mapped offline sample feature vector set; a second calculation module configured to calculate similarities of all sample pairs by using the mapped online sample feature vector set and the mapped offline sample feature vector set; and an optimization module configured to calculate a loss value of the maximum average difference by using the similarities, and optimize the initial learnable kernel function by using the loss value, to obtain the learnable kernel function.

[0016] Optionally, in an embodiment of the application, the optimization module comprises: a calculation unit configured to calculate a gradient of the loss value with respect to the similarities; and an optimization unit configured to optimize the initial learnable kernel function by using the gradient until a preset iteration stopping condition is reached, to obtain the learnable kernel function.

[0017] Optionally, in an embodiment of the application, the device further comprises: a judgment module configured to judge whether a preset representativeness deviation condition is met between the online fast inspection sample detection data and the offline fast inspection sample detection data based on the evaluation result; and a generation module configured to generate a corresponding process adjustment suggestion based on the online fast inspection sample detection data and the offline fast inspection sample detection data in a case where the preset representativeness deviation condition is not met.

[0018] Optionally, in an embodiment of the application, the coal quality detection indexes comprise at least one of an ash content index, a moisture content index, a calorific value index, and a sulfur content index of the target coal batch.

[0019] An electronic device is provided in a third aspect of the present application, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the online and offline fast inspection sample representativeness consistency evaluation method as described in the above embodiments.

[0020] A computer readable storage medium is provided in a fourth aspect of the present application, which stores computer instructions for causing the computer to execute the online and offline fast inspection sample representativeness consistency evaluation method as described in the above embodiments.

[0021] A computer program product is provided in a fifth aspect of the present application, comprising a computer program, which, when executed, is used to implement the online and offline fast inspection sample representativeness consistency evaluation method as described above.

[0022] The embodiments of the present application can calculate the maximum average difference of the online fast inspection sample feature vector set and the offline fast inspection sample feature vector set by using the pre-constructed learnable kernel function, so as to cope with the difficulty in representativeness judgment caused by uneven sample quantity, and obtain the evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference, so as to complete the judgment of whether the coal representativeness is consistent under the two detection modes through the quantitative comparison of the consistency of the online (multi-subsample) and offline (few-subsample) fast inspection results in the statistical distribution. Thus, the technical problems in the related art that the average value or rule sampling is used to judge the representativeness of the coal quality detection result, the consistency difference between the online and offline detection results cannot be effectively evaluated, the subjectivity is strong, the difference cannot be quantified, and the small sample cannot be evaluated are solved.

[0023] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which: Figure 1 A flowchart of an online and offline fast inspection sample representativeness consistency evaluation method according to an embodiment of the present application is provided; Figure 2 A principle schematic diagram of an online and offline fast inspection sample representativeness consistency evaluation method according to an embodiment of the present application is provided; Figure 3 A structure schematic diagram of an online and offline fast inspection sample representativeness consistency evaluation device according to an embodiment of the present application is provided; Figure 4 A structure schematic diagram of an electronic device according to an embodiment of the present application is provided.

[0025] Wherein, 10 - online and offline fast inspection sample representational consistency evaluation device, 100 - acquisition module, 200 - extraction module, 300 - evaluation module; 1 - data input module, 2 - feature extraction module, 3 - kernel function learning module, 4 - consistency evaluation module, 5 - result analysis module; 401 - memory, 402 - processor, 403 - communication interface. DETAILED DESCRIPTION

[0026] The embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0027] The online and offline fast inspection sample representational consistency evaluation method and device of the embodiments of the present application are described below with reference to the accompanying drawings. In view of the technical problems in the related art mentioned in the background art, the average value or regular sampling is used to judge the representativeness of the coal quality detection result, which cannot effectively evaluate the consistency difference between the online and offline detection results, and has strong subjectivity, cannot quantify the difference, and is powerless for small samples. The present application provides an online and offline fast inspection sample representational consistency evaluation method. In the method, the maximum average difference of the online fast inspection sample feature vector set and the offline fast inspection sample feature vector set can be calculated by using a pre-constructed learnable kernel function to cope with the difficulty in representativeness judgment caused by uneven sample quantity. Based on the maximum average difference, the evaluation result of the sample representational consistency of the target coal batch is obtained. Through the quantitative comparison of the consistency of the online (multiple sub-samples) and offline (few sub-samples) fast inspection results in the statistical distribution, it is determined whether the coal representativeness is consistent under the two detection modes.

[0028] Specifically, Figure 1 A flowchart of an online and offline fast inspection sample representational consistency evaluation method provided by the embodiments of the present application is shown.

[0029] As Figure 1 shown, the online and offline fast inspection sample representational consistency evaluation method includes the following steps: In step S101, the online fast inspection sample detection data and the offline fast inspection sample detection data of a target coal batch are acquired.

[0030] In actual execution process, the detection data of the online fast inspection sample, such as ash content, moisture content, calorific value, sulfur content, etc., can be acquired in real time, and the detection data of the offline fast inspection sample of the same type of index can be acquired synchronously.

[0031] In step S102, the multi-layer neural network is used to perform high-dimensional feature extraction on the coal quality detection indexes of the online rapid inspection sample detection data and the offline rapid inspection sample detection data respectively, to obtain the online rapid inspection sample feature vector set and the offline rapid inspection sample feature vector set. The coal quality detection indexes include at least one of the ash content index, the moisture content index, the calorific value index, and the sulfur content index of the target coal batch.

[0032] The embodiment of the present application can merge the collected online rapid inspection sample detection data and offline rapid inspection sample detection data, and perform high-dimensional feature extraction through a multi-layer neural network.

[0033] For example, taking a fully connected network or an autoencoder as the multi-layer neural network, in the input layer, the original detection indexes (such as ash content, moisture content, etc.) of each sub-sample in the merged data are obtained, in the hidden layer, high-order features are extracted through a nonlinear activation function (such as ReLU), and in the output layer, a fixed-length high-dimensional feature vector, such as a 128-dimensional vector, is generated.

[0034] The coal quality detection indexes correspond to the collected online rapid inspection sample detection data and offline rapid inspection sample detection data.

[0035] In the actual collection process, the embodiment of the present application can use the same device to obtain offline rapid inspection sample detection data and online rapid inspection sample detection data under the conditions of offline and online.

[0036] Among them, online rapid inspection is the process of continuously scanning the entire dynamic coal flow, which needs to collect a large amount of original data and auxiliary information to cope with complex environments, ensure real-time accuracy, and depict the continuous change of coal quality, thereby generating a huge amount of data; while offline rapid inspection is a single or a few measurements on static samples, which outputs highly condensed result information, and the data amount is naturally much smaller. Even if it is the same core detector, its working mode and supporting data acquisition and processing system are completely different in online and offline applications, resulting in a huge difference in data amount.

[0037] Therefore, in order to realize the consistent quantitative comparison of online (multi-sub-sample) and offline (few-sub-sample) rapid inspection results in statistical distribution, the embodiment of the present application can use a pre-constructed learnable kernel function to calculate the maximum average difference of the online rapid inspection sample feature vector set and the offline rapid inspection sample feature vector set respectively, to cope with the difficulty of representative judgment caused by uneven sample quantity, and obtain the evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference.

[0038] In step S103, the pre-constructed learnable kernel function is used to calculate the maximum average difference of the online rapid inspection sample feature vector set and the offline rapid inspection sample feature vector set respectively, and the evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference.

[0039] The embodiment of the present application can use the trained kernel function to calculate the maximum average difference. The smaller the maximum average difference is, the more consistent the online / offline sample distribution is. The larger the maximum average difference is, the larger the distribution difference is.

[0040] Optionally, in an embodiment of the present application, before the maximum average difference of the online fast detection sample feature vector set and the offline fast detection sample feature vector set is calculated respectively by using the pre-constructed learnable kernel function, the method further comprises: obtaining the online sample feature vector set and the offline sample feature vector set for training; inputting the online sample feature vector set and the offline sample feature vector set into the initial learnable kernel function, calculating the mapping of each sample to obtain the mapped online sample feature vector set and the mapped offline sample feature vector set; calculating the similarity of all sample pairs by using the mapped online sample feature vector set and the mapped offline sample feature vector set; calculating the loss value of the maximum average difference by using the similarity, and optimizing the initial learnable kernel function by using the loss value to obtain the learnable kernel function.

[0041] As a possible implementation manner, the embodiment of the present application can use the trainable parameter to replace the traditional Gaussian kernel, or use the neural network to generate the learnable kernel, so as to minimize the maximum average difference of the online / offline distribution, map the feature vector to the reproducing kernel Hilbert space, and implicitly model the data distribution.

[0042] The architecture of the initial learnable kernel function can include a feature mapping network and a differentiable kernel calculation layer.

[0043] The feature mapping network can map the original feature vectors in the online sample feature vector set and the offline sample feature vector set to the reproducing kernel Hilbert space; and the differentiable kernel calculation layer can calculate the similarity of all sample pairs by using the mapped online sample feature vector set and the mapped offline sample feature vector set, and calculate the loss value of the maximum average difference.

[0044] Optionally, in an embodiment of the present application, the initial learnable kernel function is optimized by using the loss value to obtain the learnable kernel function, comprising: calculating the gradient of the loss value with respect to the similarity; optimizing the initial learnable kernel function by using the gradient until a preset iteration stopping condition is reached to obtain the learnable kernel function.

[0045] During the training process, the forward propagation can calculate the mapping of each sample in the online sample feature vector set and the offline sample feature vector set, calculate the kernel matrix, and then calculate the loss value of the maximum average difference; the backward propagation can perform the gradient calculation of the loss value of the maximum average difference, so as to optimize the learnable kernel function by using the Adam optimizer, wherein, during the training process, overfitting needs to be avoided.

[0046] Optionally, in an embodiment of the present application, further comprising: based on the evaluation result, determining whether a preset representative deviation condition is met between the online rapid detection sample detection data and the offline rapid detection sample detection data; if the preset representative deviation condition is not met, generating a corresponding process adjustment suggestion based on the online rapid detection sample detection data and the offline rapid detection sample detection data.

[0047] In some embodiments, the threshold of the maximum average difference can be set according to historical data or industry standards, and the relationship between the evaluation result and the threshold is compared. If the maximum average difference of the evaluation result is greater than the threshold, it means that the representative deviation is too large, at which time a corresponding suggestion can be generated, such as a re-inspection instruction: re-collecting online samples and starting a new round of process. Process check prompt: prompting manual inspection of whether the offline sample preparation process (such as crushing and mixing) is standardized.

[0048] In combination Figure 2 As shown in the figure, the working principle of the online and offline rapid detection sample representative consistency evaluation method of the embodiment of the present application is described in detail.

[0049] The embodiment of the present application can solve the problem that the representative inconsistency between different rapid detection methods in current coal quality detection is difficult to quantify. By combining the deep feature extraction network with the trainable kernel function, the sub-sample data generated by the two detection methods are compared at the distribution level. The maximum average difference (MMD) is used to measure the distribution difference of the two types of samples in the high-dimensional feature space, and finally the representative deviation index is output to assist intelligent decision-making.

[0050] As Figure 2 shown, the structure involved in the embodiment of the present application can include: a data input module 1, a feature extraction module 2, a kernel function learning module 3, a consistency evaluation module 4, and a result analysis module 5.

[0051] The data input module 1 can be used to collect online and offline rapid detection sample coal quality index detection data and detection time.

[0052] The feature extraction module 2 can use a multi-layer neural network (such as a fully connected neural network MLP or a Transformer architecture) to extract high-dimensional features of historical coal quality detection indexes - to perform nonlinear mapping on the input feature vector and output a high-dimensional feature representation. The input is the detection index of multiple sub-samples (such as ash content, moisture content, calorific value, sulfur content, etc.), and the output is a high-dimensional representation vector of each sub-sample. If the time factor is considered to affect the change of coal quality, the detection time stamp can be encoded as a time feature (such as time interval) and input into the network with the coal quality feature vector. Or use LSTM, GRU, etc. Time series model to capture the time series dependency between sub-samples. Finally, each sub-sample is represented as a fixed-dimensional feature vector High-dimensional feature set of online fast inspection samples: , is the number of online sub-samples; high-dimensional feature set of offline fast inspection samples: , is the number of offline sub-samples.

[0053] The kernel function learning module 3 uses the high-dimensional feature representation of the historical online and offline samples of the feature extraction module 、 as the input of this module. Through an end-to-end training mechanism, a learnable kernel function can be introduced to replace the traditional Gaussian kernel, adapt to the distribution characteristics of coal quality data, and make the distributions of historical online and offline samples in the high-dimensional feature space as consistent as possible. The specific steps are as follows: Adopt the form of kernel function parameterized by neural network: .

[0054] where , : high-dimensional feature vectors of two coal samples; : mapping function parameterized by small neural network (such as MLP); : network parameters.

[0055] Construct kernel matrix, based on learnable kernel function, calculate the following three kernel matrices: Intra-sample kernel matrix of online samples: ; Intra-sample kernel matrix of offline samples: ; Cross-sample kernel matrix: .

[0056] End-to-end training mechanism: Considering the imbalance of online and offline sample sub-samples, use MMD 2 as the loss function, the goal is to minimize the distribution difference between historical online and offline samples:

[0057] where n, m are the number of online and offline sub-samples, respectively, and the parameters of the kernel function network can be updated by gradient descent method (such as Adam optimizer) to minimize the loss function.

[0058] Finally, output the optimal kernel function trained for subsequent consistency evaluation.

[0059] The consistency evaluation module 4 introduces the maximum mean difference (MMD 2As a consistency measure, the distribution difference of the representation vectors of the online and offline quick inspection sub-samples is calculated. Output MMD 2 The value can be used to measure the degree of representative consistency, and the smaller the value, the higher the representative consistency. The specific steps are as follows: (1) Input: high-dimensional feature representation of a new batch of online and offline samples from the feature extraction module , , and the trained kernel function ; (2) Calculate the kernel matrix: The kernel matrix of the online sample: ; The kernel matrix of the offline sample: ; The cross-sample kernel matrix: .

[0060] (3) Calculate the distribution difference of the current batch of samples using the MMD 2 formula:

[0061] The representative deviation threshold of the result analysis module 5 can be set (which can be determined based on historical data or expert experience), if the MMD of the online and offline quick inspection samples exceeds the threshold, it can prompt that the representative deviation is too large, and automatically generate suggestions (such as re-inspection or attention to the manual sample preparation process). At the same time, it can also combine historical data to analyze the deviation source and determine whether it is an accidental deviation or a systematic deviation.

[0062] The representative consistency evaluation method of online and offline quick inspection samples according to the embodiments of the present application can use the pre-constructed learnable kernel function to calculate the maximum mean difference of the online quick inspection sample feature vector set and the offline quick inspection sample feature vector set, respectively, to cope with the difficulty of representative judgment caused by uneven sample quantity, and based on the maximum mean difference, the evaluation result of the sample representative consistency of the target coal batch is obtained. Through the quantitative comparison of the consistency of the online (multi-subsample) and offline (few-subsample) quick inspection results in the statistical distribution, the determination of whether the coal representative is consistent under the two detection modes is completed.

[0063] Second, the online and offline quick inspection sample representative consistency evaluation device according to the embodiments of the present application is described with reference to the accompanying drawings.

[0064] Figure 3 is a block schematic diagram of the online and offline quick inspection sample representative consistency evaluation device of the embodiments of the present application.

[0065] As Figure 3As shown, the online and offline fast inspection sample representativeness consistency evaluation device 10 comprises a collection module 100, an extraction module 200 and an evaluation module 300.

[0066] Specifically, the collection module 100 is configured to collect online fast inspection sample detection data and offline fast inspection sample detection data of a target coal batch.

[0067] The extraction module 200 is configured to perform high-dimensional feature extraction on coal quality detection indexes of the online fast inspection sample detection data and the offline fast inspection sample detection data respectively by using a multi-layer neural network, to obtain an online fast inspection sample feature vector set and an offline fast inspection sample feature vector set.

[0068] The evaluation module 300 is configured to calculate maximum average differences of the online fast inspection sample feature vector set and the offline fast inspection sample feature vector set respectively by using a pre-constructed learnable kernel function, and obtain an evaluation result of sample representativeness consistency of the target coal batch based on the maximum average differences.

[0069] Optionally, in an embodiment of the present application, the online and offline fast inspection sample representativeness consistency evaluation device 10 further comprises an acquisition module, a first calculation module, a second calculation module and an optimization module.

[0070] The acquisition module is configured to acquire an online sample feature vector set and an offline sample feature vector set for training.

[0071] The first calculation module is configured to input the online sample feature vector set and the offline sample feature vector set into an initial learnable kernel function, calculate a mapping of each sample, to obtain a mapped online sample feature vector set and a mapped offline sample feature vector set.

[0072] The second calculation module is configured to calculate similarities of all sample pairs by using the mapped online sample feature vector set and the mapped offline sample feature vector set.

[0073] The optimization module is configured to calculate a loss value of the maximum average difference by using the similarities, and optimize the initial learnable kernel function by using the loss value, to obtain the learnable kernel function.

[0074] Optionally, in an embodiment of the present application, the optimization module comprises a calculation unit and an optimization unit.

[0075] The calculation unit is configured to calculate a gradient of the loss value with respect to the similarities.

[0076] The optimization unit is configured to optimize the initial learnable kernel function by using the gradient until a preset iteration stop condition is reached, to obtain the learnable kernel function.

[0077] Optionally, in an embodiment of the present application, the online and offline fast inspection sample representativeness consistency evaluation device 10 further comprises a judgment module and a generation module.

[0078] The judgment module is configured to judge whether the preset representativeness deviation condition is met between the online fast inspection sample detection data and the offline fast inspection sample detection data based on the evaluation result.

[0079] The generation module is configured to generate a corresponding process adjustment suggestion based on the online fast inspection sample detection data and the offline fast inspection sample detection data in the case that the preset representativeness deviation condition is not met.

[0080] Optionally, in an embodiment of the present application, the coal quality detection index comprises at least one of an ash content index, a moisture content index, a calorific value index and a sulfur content index of the target coal batch.

[0081] It should be noted that the foregoing explanation and description of the online and offline fast inspection sample representativeness consistency evaluation method embodiment are also applicable to the online and offline fast inspection sample representativeness consistency evaluation device of this embodiment, which will not be described here again.

[0082] The online and offline fast inspection sample representativeness consistency evaluation device according to the embodiment of the present application can calculate the maximum average difference of the online fast inspection sample feature vector set and the offline fast inspection sample feature vector set by using the pre-constructed learnable kernel function, so as to cope with the representativeness judgment difficulty caused by uneven sample quantity, and obtain the evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference, so as to complete the judgment of whether the coal representativeness is consistent under two detection modes through the quantitative comparison of the consistency of the online (multi-subsample) and offline (few-subsample) fast inspection results in the statistical distribution.

[0083] Figure 4 The structure schematic diagram of the electronic device provided by the embodiment of the present application is shown. The electronic device can comprise: The memory 401, the processor 402 and the computer program stored in the memory 401 and executable on the processor 402.

[0084] The processor 402 implements the online and offline fast inspection sample representativeness consistency evaluation method provided in the above embodiment when executing the program.

[0085] Further, the electronic device further comprises: The communication interface 403 is configured to communicate between the memory 401 and the processor 402.

[0086] The memory 401 is configured to store the computer program executable on the processor 402.

[0087] The memory 401 can include a high-speed RAM memory, and can also include a non-volatile memory, such as at least one disk memory.

[0088] If the memory 401, the processor 402 and the communication interface 403 are implemented independently, the communication interface 403, the memory 401 and the processor 402 can be connected to each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 4 In the figure, only one thick line is used to represent, but it does not mean that there is only one bus or one type of bus.

[0089] Optionally, in a specific implementation, if the memory 401, the processor 402 and the communication interface 403 are integrated on a chip, the memory 401, the processor 402 and the communication interface 403 can complete communication between each other through an internal interface.

[0090] The processor 402 can be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application.

[0091] The embodiment also provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the online and offline fast inspection sample representative consistency evaluation method.

[0092] The embodiment of the present application also provides a computer program product, which includes a computer program, and the computer program is executed by a processor to implement the online and offline fast inspection sample representative consistency evaluation method provided by the embodiment of the present application.

[0093] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. In addition, the usage of "N" means at least two, for example, two, three or the like, unless explicitly stated otherwise.

[0094] Furthermore, the terms "first", "second", or the like, are used merely as a designation of certain elements or features, and do not imply or connote relative importance or a specific order of categorization thereof. Accordingly, features described as "first" or "second" can be explicitly or implicitly included in at least one of the features. In the description of the application, the meaning of "N" is at least two, for example, two, three, etc., unless explicitly specified otherwise.

[0095] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments which can be managed as one or more modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps, and alternate implementations are possible. In some embodiments, the processes and methods described can be executed by one or more apparatuses or devices, either directly or after conversion to another language. Alternate implementations are possible.

[0096] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of executable instructions stored in a computer readable medium, which can be executed by an instruction execution system, apparatus or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or a combination thereof. For the purposes of this specification, a "computer readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer readable medium can be a computer readable storage medium or a computer readable signal medium. The computer readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a propagation medium. The computer readable signal medium can include, but is not limited to, a computer readable medium that facilitates transfer of the program from one place to another. A specific example of a computer readable medium is a non-transitory computer-readable storage medium. A specific example of a computer readable signal medium is a source or destination of the computer readable medium. Another specific example of a computer readable signal medium is a computer readable signal travelling through space. Thus, a computer readable medium can take many forms of hardware to carry out the program for use by or in connection with the instruction execution system, apparatus, or device.

[0097] It should be understood that aspects of the application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As such, if implemented in hardware and in another embodiment, any of the following technologies, known in the art, or their combinations can be used to implement the hardware: discrete logic circuitry having logic gates for implementing logic functions upon an application of data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0098] Those of skill in the art would understand that the steps carried out by the above-mentioned embodiments can be implemented by programs instructing the relevant hardware to complete all or part of the steps, and the programs can be stored in a computer readable storage medium. When the programs are executed, they include one of the steps of the method embodiments or a combination thereof.

[0099] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0100] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. An online and offline fast examination sample representativeness consistency evaluation method, characterized in that, Includes the following steps: Collect online and offline rapid test sample data of the target coal batch; High-dimensional feature extraction of coal quality detection indicators is performed on the online rapid detection sample data and the offline rapid detection sample data using a multi-layer neural network to obtain the feature vector set of the online rapid detection sample and the feature vector set of the offline rapid detection sample. The maximum average difference between the online rapid detection sample feature vector set and the offline rapid detection sample feature vector set is calculated using a pre-constructed learnable kernel function, and the evaluation result of the sample representativeness consistency of the target coal batch is obtained based on the maximum average difference.

2. The method of claim 1, wherein, Before calculating the maximum average difference between the online rapid detection sample feature vector set and the offline rapid detection sample feature vector set using a pre-built learnable kernel function, the method further includes: Obtain the online and offline sample feature vector sets for training; The online sample feature vector set and the offline sample feature vector set are input into the initial learnable kernel function to calculate the mapping for each sample, so as to obtain the mapped online sample feature vector set and the mapped offline sample feature vector set; The similarity of all sample pairs is calculated using the online sample feature vector set and the offline sample feature vector set. The loss value of the maximum average difference is calculated using the similarity, and the initial learnable kernel function is optimized using the loss value to obtain the learnable kernel function.

3. The method of claim 2, wherein, The step of optimizing the initial learnable kernel function using the loss value to obtain the learnable kernel function includes: Calculate the gradient of the loss value with respect to the similarity; The initial learnable kernel function is optimized using the gradient until a preset iteration stopping condition is met, thus obtaining the learnable kernel function.

4. The method of claim 1, wherein, Also includes: Based on the evaluation results, it is determined whether the online rapid test sample detection data and the offline rapid test sample detection data meet the preset representativeness deviation condition; If the preset representativeness deviation condition is not met, corresponding process adjustment suggestions are generated based on the online rapid test sample detection data and the offline rapid test sample detection data.

5. The method according to claim 1, characterized in that, The coal quality testing indicators include at least one of the following: ash content, moisture content, calorific value, and sulfur content of the target coal batch.

6. A device for evaluating the representativeness consistency of online and offline rapid test samples, characterized in that, include: The data acquisition module is used to collect online and offline rapid test sample data of the target coal batch; The extraction module is used to perform high-dimensional feature extraction on the coal quality detection indicators of the online rapid detection sample data and the offline rapid detection sample data using a multi-layer neural network, so as to obtain the feature vector set of the online rapid detection sample and the feature vector set of the offline rapid detection sample. The evaluation module is used to calculate the maximum average difference between the online rapid detection sample feature vector set and the offline rapid detection sample feature vector set using a pre-built learnable kernel function, and to obtain the evaluation result of the sample representativeness consistency of the target coal batch based on the maximum average difference.

7. The apparatus according to claim 6, characterized in that, Also includes: The acquisition module is used to acquire the online sample feature vector set and the offline sample feature vector set for training. The first calculation module is used to input the online sample feature vector set and the offline sample feature vector set into the initial learnable kernel function, calculate the mapping of each sample, and obtain the mapped online sample feature vector set and the mapped offline sample feature vector set. The second calculation module is used to calculate the similarity of all sample pairs using the mapping online sample feature vector set and the mapping offline sample feature vector set; An optimization module is used to calculate the loss value of the maximum average difference using the similarity, and to optimize the initial learnable kernel function using the loss value to obtain the learnable kernel function.

8. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, the processor executing the program to implement the online and offline rapid test sample representativeness consistency assessment method as described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the online and offline rapid test sample representativeness consistency assessment method as described in any one of claims 1-5.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed, it is used to implement the online and offline rapid test sample representativeness consistency assessment method as described in any one of claims 1-5.

Images