A water quality detection method, device and system based on cleaning waste liquid circulating purification

Abstract

This invention discloses a water quality detection method, equipment, and system based on the circulation and purification of cleaning wastewater, belonging to the field of water-damaged vehicle cleaning wastewater detection technology. It addresses the technical problem of the lack of real-time, accurate, and multi-parameter collaborative online detection methods in the existing water-damaged vehicle cleaning wastewater circulation and purification process, making it difficult to achieve full-parameter online self-inspection and judgment of water quality. The method includes: synchronously acquiring and processing multi-source sensor signals of relevant water quality indicators from the purified water body passing through the sensing detection chamber in a laminar flow state; concatenating the feature values ​​between the time-domain statistical features and the cross-correlation features between signals in the original water quality signal matrix; judging the state of the purified water body based on the similarity matching of the purified water quality feature map; performing reverse mapping processing of the water quality difference vector based on the dominant difference feature deviation degree of each feature; and performing a re-inspection cycle and state tracking of the secondary purified water body after the directional reflux strategy has been implemented.

Description

Technical Field

[0001] This application relates to the field of water-damaged vehicle cleaning wastewater testing, and in particular to a water quality testing method, equipment and system based on the recycling and purification of cleaning wastewater. Background Technology

[0002] Vehicle cleaning is an essential step in the repair of flood-damaged vehicles. Because these vehicles have been submerged in sewage for extended periods, their bodies and components are covered with large amounts of mud, oil, metal shavings, colloidal impurities, and decaying organic matter, making cleaning far more difficult than for ordinary vehicles. Statistics show that cleaning a single flood-damaged vehicle consumes up to 1.5 tons of water. Based on 1 million flood-damaged vehicles annually, this translates to an annual water consumption of 1.5 million tons, representing a severe waste of water resources.

[0003] Even more serious is the extremely complex composition of wastewater from water-damaged truck cleaning, which typically contains mud and sand particles, floating oil, metal shavings, colloidal impurities, and decaying organic matter, accompanied by a noticeable odor. However, the industry commonly sees wastewater being directly discharged into ditches, canals, or spilled onto the ground, posing a very high risk of environmental violations. Traditional wastewater treatment facilities are mostly enclosed, earthen-structured tanks, making it impossible for operators to visually observe water flow, sedimentation effects, and sludge accumulation. This makes dredging and maintenance difficult and easily leads to odor leaks and secondary pollution.

[0004] Currently, some car wash wastewater treatment technologies have been reported, such as AI-based car wash wastewater purification and control systems that monitor wastewater data through real-time water quality monitoring modules. However, existing technologies are mostly geared towards ordinary car wash scenarios, with single water quality testing indicators and a simple judgment logic of discharge if qualified and reprocessing if unqualified. They lack multi-parameter collaborative detection and graded recirculation control mechanisms for the complex characteristics of wastewater from water-damaged car washes. More importantly, existing technologies often use endpoint testing to determine the reuse of purified water, i.e., a one-time sampling and testing at the end of the treatment process. If the water fails to meet the standards, the entire batch is returned for reprocessing. This model is inefficient, has a slow response time, and cannot achieve true real-time online closed-loop control.

[0005] Therefore, there is an urgent need for an online water quality detection method in the recycling and purification of wastewater from water-damaged vehicle cleaning. This method should be able to achieve online self-inspection and judgment of all parameters of water quality, accurately identify indicators exceeding standards, and initiate targeted reflux re-inspection to ensure the recycling of purified water and zero discharge of wastewater. Summary of the Invention

[0006] This application provides a water quality testing method, equipment, and system based on the recycling and purification of cleaning wastewater, which solves the following technical problem: In the existing process of recycling and purifying wastewater from water-damaged vehicle cleaning, there is a lack of real-time, accurate, and multi-parameter collaborative online detection methods, making it difficult to achieve online self-inspection and judgment of all parameters of water quality.

[0007] The embodiments of this application adopt the following technical solutions: On one hand, this application provides a water quality detection method based on the recycling and purification of cleaning wastewater, including: according to the sensing detection flow path, simultaneously acquiring and processing multi-source sensor signals of relevant water quality indicators of the purified water body passing through the sensing detection chamber in a laminar flow state, and constructing a raw water quality signal matrix; splicing the feature values ​​between the time-domain statistical features and the cross-correlation features between signals in the raw water quality signal matrix to obtain a purified water quality feature map; and using a preset qualified water body benchmark feature map library, performing purified water body status analysis based on map similarity matching on the purified water quality feature map. The system first determines the state of the purified water body and obtains the water quality result. If the water quality result is abnormal, the system then performs a reverse mapping process on the dominant difference feature based on the deviation degree of each feature on the water quality difference vector to determine the abnormal state code in the current abnormal purified water body. Based on the abnormal state code, the system performs hierarchical directional backflow control on the corresponding current abnormal purified water body according to the relevant abnormal type code to obtain the directional backflow strategy for the current abnormal purified water body. The system then performs a re-inspection cycle and state tracking on the secondary purified water body after executing the directional backflow strategy to obtain the comprehensive water quality detection result.

[0008] This application's embodiments construct a 72-dimensional feature map containing time-domain statistical features and cross-correlation features between signals. Using the Mahalanobis distance between the overall map and the baseline map as the judgment criterion, it can capture situations where individual indicators meet standards but the overall state is abnormal. By extracting cross-correlation features between signals, it can keenly capture changes in the synergistic relationship between different water quality parameters. Furthermore, by weighting the natural fluctuation amplitude of each feature dimension using a covariance matrix, it avoids misjudgments caused by large natural fluctuations in certain features. Through difference vector analysis and dominant difference feature identification, it further subdivides abnormal water states into three categories: amplitude-type anomalies, fluctuation-type anomalies, and relational anomalies, each corresponding to different recirculation processing strategies, achieving precise recirculation and significantly improving recirculation processing efficiency. Simultaneously, by utilizing the recorded judgment results in the re-inspection cycle and tracking the changing trend of the Mahalanobis distance, it can automatically adjust the recirculation strategy (switching alternative paths) based on the changing trend, achieving adaptive optimization of recirculation processing and avoiding the failure of fixed recirculation paths under specific abnormal situations.

[0009] In one feasible implementation, according to the sensing and detection flow path, the purified water flowing through the sensing and detection cavity in a laminar flow state undergoes simultaneous multi-source sensor signal acquisition and processing for relevant water quality indicators, and a raw water quality signal matrix is ​​constructed. Specifically, this includes: the purified water entering the sensing and detection cavity in a laminar flow state through a honeycomb rectifier built into the inlet flow stabilization section of the sensing and detection flow path; upon detection of the purified water in the sensing and detection cavity, activating six sets of sensor detection probes; wherein the sensing and detection flow path includes: the inlet flow stabilization section, the sensing and detection cavity, and the outlet buffer section; and using turbidity sensing probes, conductivity sensing probes... The sensor, consisting of a redox potential sensor, a dissolved oxygen sensor, an ultraviolet absorption sensor, and an odor sensor, synchronously acquires signals from the purified water in the sensing chamber, obtaining multiple voltage signals. The odor sensor exhibits different sensitivities to sulfides, amines, hydrocarbons, and esters, and outputs four voltage signals. Based on the data window at each sampling time, the multiple voltage signals are vector-sorted to construct a raw water quality signal matrix for multiple consecutive sampling times. This raw water quality signal matrix represents the instantaneous response characteristics of the purified water across nine sensing dimensions.

[0010] In one feasible implementation, the time-domain statistical features and cross-correlation features in the original water quality signal matrix are concatenated to obtain a purified water quality feature map. Specifically, this includes: calculating the mean, standard deviation, peak factor, and waveform factor for each voltage signal sequence in the original water quality signal matrix, and statistically extracting time-domain statistical feature values; wherein the number of terms in the time-domain statistical feature values ​​is 36; calculating the Pearson correlation coefficient between each pair of voltage signals in the original water quality signal matrix to obtain a correlation coefficient matrix; wherein the correlation coefficient matrix represents the cooperative change relationship between responses of different sensing dimensions; extracting the upper triangular portion of the correlation coefficient matrix and determining the correlation coefficient values ​​in the correlation coefficient matrix as cross-correlation feature values; the number of terms in the cross-correlation feature values ​​is 36; and performing feature concatenation processing on the time-domain statistical feature values ​​and the cross-correlation feature values ​​to generate the purified water quality feature map; wherein the purified water quality feature map is a 72-dimensional feature vector.

[0011] In one feasible implementation, after concatenating the eigenvalues ​​of the time-domain statistical features and the cross-correlation features between signals in the original water quality signal matrix to obtain a purified water quality feature map, the method further includes: performing dimensionless processing on each eigenvalue in the purified water quality feature map based on a reference scale parameter to obtain dimensionless eigenvalues; performing interval mapping processing on the dimensionless eigenvalues ​​using a hyperbolic tangent function to obtain all eigenvalues ​​within a uniform amplitude range; and obtaining a standardized and normalized purified water quality feature map based on the dimensionless eigenvalues ​​and all eigenvalues ​​within the uniform amplitude range.

[0012] In one feasible implementation, a pre-set qualified water body benchmark feature map library is used to determine the state of the purified water body based on map similarity matching, thereby obtaining the purified water quality result. Specifically, this includes: constructing a historical standardized feature map based on continuously collected historical qualified water body sensor data; calculating the mean vector and covariance matrix of the historical standardized feature map; constructing the qualified water body benchmark feature map library using the benchmark feature map center of the qualified water body corresponding to the mean vector and the normal fluctuation range of the qualified water body feature map corresponding to the covariance matrix; weighting the fluctuation amplitude of each feature dimension based on the covariance matrix, and then comparing the standardized and normalized purified water quality feature map with the historical standardized feature map. The Mahalanobis distance is calculated between the mean vectors of the graphs to obtain the current Mahalanobis distance of the purified water quality feature graph. If the current Mahalanobis distance is less than or equal to a preset Mahalanobis distance, the similarity between the current purified water quality feature graph and the benchmark feature graph of qualified water is within the normal fluctuation range, and the purified water is determined to be of qualified quality. The preset Mahalanobis distance is the Mahalanobis distance distribution of qualified purified water samples. If the current Mahalanobis distance is greater than the preset Mahalanobis distance, the similarity between the current purified water quality feature graph and the benchmark feature graph of qualified water is significantly different, and the purified water is determined to be of abnormal quality. The purified water quality results include both abnormal and qualified results.

[0013] In one feasible implementation, if the purified water quality result is an abnormal result, the water quality difference vector is subjected to a back-mapping process based on the dominant difference feature deviation degree of each feature to determine the abnormal state code in the current abnormal purified water. Specifically, this includes: if the purified water quality result is an abnormal result, determining the difference vector based on the Mahalanobis distance between the mean vectors of the purified water quality feature map and the historical standardized feature map; calculating the deviation degree of each feature value in the difference vector based on the normal fluctuation amplitude to obtain the deviation degree of each feature; wherein the deviation degree is the degree to which each feature value in the difference vector deviates from the baseline value; sorting the deviation degrees of the 72-dimensional feature vector, and determining the feature vectors corresponding to several feature values ​​in the top-ranked interval as dominant difference features; wherein the dominant difference features are the main features in the 72-dimensional feature vector that cause the abnormal state of the purified water. The process involves identifying the leading factor characteristics. If the tracing path of the leading difference characteristic is the time-domain statistical characteristic of a single sensing voltage information, then the number of the leading difference characteristic is reverse-mapped to the corresponding single sensing voltage signal type. If the tracing path of the leading difference characteristic is the signal-to-signal cross-correlation characteristic between two voltage signals, then the number of the leading difference characteristic is reverse-mapped to sensing voltage signal types with a cooperative relationship. Based on the corresponding number of the sensing voltage signal type after reverse mapping, a signal code is generated for each abnormal sensing voltage signal type in the current abnormal purified water body. The abnormal type code and deviation level corresponding to each abnormal sensing voltage signal type are then formatted and encoded according to the framework of the signal code to obtain the abnormal state code in the current abnormal purified water body. The abnormal type code includes amplitude anomaly, fluctuation anomaly, and relational anomaly. The deviation level includes slight deviation, moderate deviation, and severe deviation.

[0014] In one feasible implementation, based on the abnormal state code, the corresponding current abnormal purified water body is subjected to hierarchical directional backflow control according to the relevant abnormal type code to obtain the directional backflow strategy of the current abnormal purified water body. Specifically, this includes: based on the abnormal type code in the abnormal state code, performing directional diversion processing on the sensor voltage signal type in the amplitude-type abnormal state for the current abnormal purified water body to determine the corresponding water quality purification unit allocation execution strategy; based on the sensor voltage signal type in the fluctuation-type abnormal state, sending the current abnormal purified water body to the buffer averaging unit and returning it to the sensor detection flow path for re-detection to obtain the water quality buffer execution strategy; and then... The specific sensor signal combinations involved in the abnormal state sensor voltage signal are used to automatically infer the failure of the water purification unit, and the current abnormal purified water is sent to the corresponding series deep processing combination unit to obtain the water quality abnormal combination execution strategy; when the abnormal state codes all include amplitude-type abnormal state, fluctuation-type abnormal state, and relational abnormal state, the current abnormal purified water is detected and controlled in a fixed priority order to obtain the water quality composite abnormal execution strategy; wherein, the directional backflow strategy includes: the water quality purification unit allocation execution strategy, the water quality buffer execution strategy, the water quality abnormal combination execution strategy, and the water quality composite abnormal execution strategy.

[0015] In one feasible implementation, the secondary purified water body after the directional reflux strategy has been subjected to a re-inspection cycle and status tracking to obtain a comprehensive water quality test result. Specifically, this includes: performing a re-inspection cycle judgment on the water quality results of the secondary purified water body after the directional reflux strategy has been implemented; if the water quality result of the new purified water body is qualified, the comprehensive water quality test result is a water quality test pass result; if the water quality result of the new purified water body is unqualified, and the Mahalanobis distance between the mean vector of the new purified water quality feature map and the historical standardized feature map is a successively decreasing result, the comprehensive water quality test result is a result for continuing the re-inspection cycle; if the water quality result of the new purified water body is unqualified, and the Mahalanobis distance between the mean vector of the new purified water quality feature map and the historical standardized feature map is unchanged or increases, the comprehensive water quality test result is a water quality purification risk abnormal result, and a maintenance warning signal is sent. Secondly, embodiments of this application also provide a water quality testing device based on the circulation and purification of cleaning wastewater. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to execute a water quality testing method based on the circulation and purification of cleaning wastewater as described in any of the above embodiments.

[0016] Secondly, embodiments of this application also provide a water quality testing device based on the circulation and purification of cleaning wastewater. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to execute a water quality testing method based on the circulation and purification of cleaning wastewater as described in any of the above embodiments.

[0017] Thirdly, embodiments of this application also provide a water quality detection system based on the circulation and purification of cleaning waste liquid, so that the processor in the water quality detection system can execute a water quality detection method based on the circulation and purification of cleaning waste liquid as described in any of the above embodiments.

[0018] This application provides a water quality detection method, equipment, and system based on the recycling and purification of cleaning wastewater. Compared with the prior art, the embodiments of this application have the following beneficial technical effects: This application's embodiments construct a 72-dimensional feature map containing time-domain statistical features and cross-correlation features between signals. Using the Mahalanobis distance between the overall map and the baseline map as the judgment criterion, it can capture situations where individual indicators meet standards but the overall state is abnormal. By extracting cross-correlation features between signals, it can keenly capture changes in the synergistic relationship between different water quality parameters. Furthermore, by weighting the natural fluctuation amplitude of each feature dimension using a covariance matrix, it avoids misjudgments caused by large natural fluctuations in certain features. Through difference vector analysis and dominant difference feature identification, it further subdivides abnormal water states into three categories: amplitude-type anomalies, fluctuation-type anomalies, and relational anomalies, each corresponding to different recirculation processing strategies, achieving precise recirculation and significantly improving recirculation processing efficiency. Simultaneously, by utilizing the recorded judgment results in the re-inspection cycle and tracking the changing trend of the Mahalanobis distance, it can automatically adjust the recirculation strategy (switching alternative paths) based on the changing trend, achieving adaptive optimization of recirculation processing and avoiding the failure of fixed recirculation paths under specific abnormal situations. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart of a water quality detection method based on the recycling and purification of cleaning waste liquid is provided in this application embodiment; Figure 2 A schematic diagram of a water quality testing device based on the circulation and purification of cleaning waste liquid is provided in an embodiment of this application; Figure 3 This is a schematic diagram of a water quality detection system based on the recycling and purification of cleaning waste liquid, provided in an embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0021] This application provides a water quality detection method based on the recycling and purification of cleaning wastewater, such as... Figure 1 As shown, the water quality detection method based on the recycling and purification of cleaning wastewater specifically includes steps S101-S106: S101. According to the sensing and detection flow path, the purified water that passes through the sensing and detection chamber in a laminar flow state is simultaneously collected and processed by multi-source sensing signals of relevant water quality indicators, and a raw water quality signal matrix is ​​constructed.

[0022] Specifically, the purified water first enters the sensing chamber in a laminar flow state through the honeycomb rectifier built into the inlet flow stabilization section of the sensing detection path. Once the purified water in the sensing chamber is detected, six sets of sensor probes are activated. The sensing detection path includes: the inlet flow stabilization section, the sensing chamber, and the outlet buffer section.

[0023] In one embodiment, in the wastewater recycling and purification system for flooded trucks, a sensing and detection flow path is introduced via a delivery pipeline. This flow path, arranged sequentially along the water flow direction, includes an inlet flow stabilization section, a sensing and detection chamber, and an outlet buffer section. The inlet flow stabilization section incorporates a honeycomb rectifier to uniformly disperse the incoming purified water into several fine streams, eliminating water flow pulsations and turbulence interference caused by intermittent discharge from upstream purification equipment, allowing the water to enter the sensing and detection chamber in a laminar flow state. The water velocity decreases within the inlet flow stabilization section. The sensing and detection chamber is a rectangular sealed cavity, with six sets of sensor mounting points arranged sequentially along the water flow direction inside. An automatic air vent valve is located at the top of the sensing and detection chamber to expel air bubbles entrained in the water, preventing air bubbles from adhering to the sensor probe surface and affecting the stability of signal acquisition. A drain outlet is located at the bottom of the sensing and detection chamber for periodically discharging deposited microparticles. The outlet buffer section is a gradually expanding pipe structure, which is used to slow down the sudden change in water velocity when the water flows out, and to prevent the negative pressure caused by the sudden change in flow velocity from mixing external air into the water.

[0024] Furthermore, using turbidity, conductivity, oxidation-reduction potential, dissolved oxygen, ultraviolet absorption, and odor sensors pre-configured in the sensing chamber, signals from the purified water within the sensing chamber are synchronously acquired, resulting in multiple voltage signals. The odor sensor exhibits different sensitivities to sulfides, amines, hydrocarbons, and esters, and outputs four voltage signals.

[0025] In one embodiment, six sensor probes are used, including: 1) a turbidity sensing probe: a 90° scattered light turbidity sensor with an 860nm infrared LED as the light source, an optical path length of 2.5cm, a sampling frequency of 10Hz, and an output turbidity voltage signal V_T(t). 2) a conductivity sensing probe: a dual-electrode conductivity sensor with an electrode spacing of 5mm, an excitation frequency of 1kHz, a sampling frequency of 10Hz, and an output conductivity voltage signal V_C(t). 3) a redox potential sensing probe: a platinum electrode-silver / silver chloride reference electrode combination with a sampling frequency of 10Hz and an output ORP voltage signal V_R(t). 4) a dissolved oxygen sensing probe: a fluorescence quenching dissolved oxygen sensor with a sampling frequency of 5Hz and an output dissolved oxygen voltage signal V_O(t). 5) a UV absorption sensing probe: a 254nm deep UV LED as the light source with an optical path length of 1.0cm, a sampling frequency of 10Hz, and an output UV absorbance voltage signal V_U(t). 6) Odor sensing probe: It adopts a gas sensor array composed of four metal oxide semiconductor gas-sensitive elements with different sensitivity characteristics. The four elements have different response sensitivities to sulfides, amines, hydrocarbons and esters respectively. The sampling frequency is 5Hz and it outputs four odor voltage signals V_G1(t)~V_G4(t).

[0026] Furthermore, based on the data window at each sampling time, the multiple voltage signals are vector-sorted to construct a raw water quality signal matrix for multiple consecutive sampling times. This raw water quality signal matrix represents the instantaneous response characteristics of the purified water body across nine sensing dimensions.

[0027] In one embodiment, the six sets of probes output a total of nine voltage signals (one for turbidity, one for conductivity, one for ORP, one for dissolved oxygen, one for UV absorption, and four for odor). Each signal is processed by a signal conditioning circuit (including pre-amplification, filtering, and analog-to-digital conversion) and then synchronously acquired by a data acquisition card at a sampling frequency of 100Hz. The data acquisition card then arranges the nine signal values ​​acquired at each sampling time t into a nine-dimensional original signal vector: X(t) = [V_T(t), V_C(t), V_R(t), V_O(t), V_U(t), V_G1(t), V_G2(t), V_G3(t), V_G4(t)] Then, N sampling times (N=50, corresponding to a 0.5-second data window) are continuously collected to form the original signal matrix M (9 rows × 50 columns): M=[X(t1), X(t2), X(t3), ..., X(t50)]. This original water quality signal matrix M is the original sensor response snapshot of the purified water body at that moment, reflecting the instantaneous response characteristics of the water body in 9 sensing dimensions.

[0028] S102. The time-domain statistical features and cross-correlation features between signals in the original water quality signal matrix are concatenated to obtain the purified water quality feature map.

[0029] Specifically, it is also necessary to calculate the mean, standard deviation, peak factor, and waveform factor of each voltage signal sequence in the original water quality signal matrix, and then statistically extract the time-domain statistical feature values. The number of time-domain statistical feature values ​​is 36.

[0030] In one embodiment, for the extraction of time-domain statistical features, the following calculations are required for each of the nine signal sequences (length 50): mean: μ_i = (1 / 50) × Σ_{k=1}^{50}V_i(t_k); standard deviation: σ_i = sqrt[(1 / 50) × Σ_{k=1}^{50}(V_i(t_k) - μ_i)²]; peak factor: CF_i = max(V_i(t_k)) / μ_i (reflecting the sharpness of signal fluctuations); waveform factor: WF_i = (1 / 50) × Σ_{k=1}^{50}|V_i(t_k)| / sqrt[(1 / 50) × Σ_{k=1}^{50}V_i(t_k)²] (reflecting the waveform shape of the signal); thus, a total of 36 time-domain statistical feature values ​​are extracted from the nine signals.

[0031] Furthermore, the Pearson correlation coefficients between each pair of voltage signals in the original water quality signal matrix are calculated, resulting in a correlation coefficient matrix. This correlation coefficient matrix represents the coordinated change relationship between responses from different sensing dimensions. Then, the upper triangular portion of the correlation coefficient matrix is ​​extracted, and the correlation coefficient values ​​in the matrix are determined as cross-correlation eigenvalues; the number of cross-correlation eigenvalues ​​is 36.

[0032] In one embodiment, for the extraction of cross-correlation features between signals, it is necessary to calculate the Pearson correlation coefficients between each pair of the nine signals, forming a 9×9 correlation coefficient matrix R. This matrix reflects the synergistic change relationship between responses of different sensing dimensions. For example, if the turbidity signal and the ultraviolet absorption signal are highly positively correlated, it indicates that suspended particles and organic pollutants in the water body have a synchronous change trend; if the correlation between two signals changes significantly during the purification process, it indicates that the internal composition of the water body has changed. Then, the upper triangular portion (excluding the diagonal) of the correlation coefficient matrix R is extracted, yielding C(9,2)=36 correlation coefficient values ​​as cross-correlation features.

[0033] Furthermore, the time-domain statistical feature values ​​and cross-correlation feature values ​​are subjected to feature splicing processing of the purified water body to generate a purified water quality feature map; wherein, the purified water quality feature map is a 72-dimensional feature vector.

[0034] In one embodiment, based on the time-domain statistical features and cross-correlation features obtained above, the 36 time-domain statistical feature values ​​and the 36 cross-correlation feature values ​​are concatenated to form a 72-dimensional feature vector F: F=[μ1, σ1, CF1, WF1, μ2, σ2, CF2, WF2, ..., μ9, σ9, CF9, WF9, r 12 r 13 , ..., r 89 ] The 72-dimensional feature vector F represents the feature map of the purified water at that moment. The feature map F is a highly condensed and abstracted representation of the original sensor signal. It does not rely on the instantaneous value of a single sample but is based on the overall statistical characteristics of the signal within a 0.5-second time window, exhibiting characteristics of strong noise resistance and high stability. More importantly, the cross-correlation feature in F captures the cooperative changes between different sensor dimensions.

[0035] Furthermore, after obtaining the purified water quality characteristic spectrum, each characteristic value in the spectrum is dimensionlessly processed based on a reference scale parameter to obtain dimensionless characteristic values. Then, the dimensionless characteristic values ​​are interval-mapped using a hyperbolic tangent function to obtain all characteristic values ​​within a uniform amplitude range. Finally, based on the dimensionless characteristic values ​​and all characteristic values ​​within the uniform amplitude range, a standardized and normalized purified water quality characteristic spectrum is obtained.

[0036] As a feasible implementation method, since the output signals of different sensor probes have different dimensions and amplitude ranges (turbidity signals are in the millivolt range, ORP signals are in the hundred millivolt range, etc.), directly using the original feature values ​​for similarity comparison will produce dimensional bias. Therefore, the data processing unit performs standardization and normalization operations on the feature vector F, including: (a) Dimensionless processing: For each feature value f_j in F, divide it by its corresponding reference scale parameter s_j (the reference scale parameter is obtained in advance through statistical analysis of the feature spectra of a large number of qualified water bodies, and is the standard deviation of each feature value), to obtain the dimensionless feature value f'_j=f_j / s_j. (b) Amplitude normalization: Map the dimensionless feature value f'_j to the interval [-1, 1] through the hyperbolic tangent function: f''_j=tanh(f'_j / 3), so that all feature values ​​have a uniform amplitude range, which is convenient for subsequent similarity calculation. Finally, after standardization and normalization, a standardized 72-dimensional feature map F_norm is obtained, which is the purified water quality feature map after standardization and normalization.

[0037] S103. Using a pre-set database of qualified water body reference feature maps, the purified water quality feature maps are used to determine the state of the purified water body based on map similarity matching, and the purified water quality results are obtained.

[0038] Specifically, based on historically continuously collected sensor data of qualified water bodies, a historical standardized feature map is first constructed. The mean vector and covariance matrix of the historical standardized feature map are then calculated. Next, a qualified water body baseline feature map library is constructed using the baseline feature map center of the qualified water body corresponding to the mean vector and the normal fluctuation range of the qualified water body feature map corresponding to the covariance matrix.

[0039] As a feasible implementation method, under the condition that the circulating purification system is operating stably and the purified water has been confirmed to meet all standards by laboratory testing, sensor data from 100 qualified water bodies are continuously collected. Following the aforementioned method for generating purified water quality characteristic maps, 100 standardized characteristic maps are generated, and the mean vector F_base and covariance matrix Σ of these 100 maps are calculated. F_base represents the baseline characteristic map center of the qualified water body, and Σ describes the normal fluctuation range of the qualified water body's characteristic map.

[0040] Furthermore, based on the covariance matrix, the fluctuation amplitude of each feature dimension is weighted, and the Mahalanobis distance is calculated between the mean vector of the purified water quality feature map after standardization and normalization and the historical standardized feature map to obtain the current Mahalanobis distance of the purified water quality feature map.

[0041] In one embodiment, in the determination of the state of purified water body based on map similarity matching, the Mahalanobis distance D between the purified water quality feature map F_norm and the mean vector F_base of the historical standardized feature map can be calculated first: D=sqrt[(F_norm-F_base)] ×Σ -1 [×(F_norm-F_base)]; where Mahalanobis distance not only considers the Euclidean distance between F_norm and F_base, but also weights the fluctuation amplitude of each feature dimension through the covariance matrix Σ. That is, feature dimensions with large fluctuations are given lower weights in distance calculation, and feature dimensions with small fluctuations are given higher weights, thereby avoiding misjudgment due to the natural large fluctuations of some features.

[0042] If the current Mahalanobis distance is less than or equal to the preset Mahalanobis distance, the similarity between the current purified water quality characteristic map and the benchmark characteristic map of qualified water bodies is within the normal fluctuation range, and the purified water quality is determined to be qualified. The preset Mahalanobis distance refers to the Mahalanobis distance distribution of qualified purified water samples. If the current Mahalanobis distance is greater than the preset Mahalanobis distance, the similarity between the current purified water quality characteristic map and the benchmark characteristic map of qualified water bodies is significantly different, and the purified water quality is determined to be abnormal. The purified water quality results include both abnormal and qualified results.

[0043] In one embodiment, a preset Mahalanobis distance threshold D0 is used (D0 is the Mahalanobis distance distribution obtained by statistically analyzing qualified water samples). When D≤D0, the similarity between the current water body's feature map and the benchmark feature map of qualified water bodies is within the normal fluctuation range, and the purified water body is determined to be of qualified quality, thus initiating a clean water circulation supply process. When D>D0, there is a significant difference between the current water body's feature map and the benchmark feature map of qualified water bodies, indicating an abnormal water quality, terminating direct reuse, and initiating a fine-grained water state classification and directional reflux process.

[0044] S104. If the water quality result of the purified water body is an abnormal result, then the water quality difference vector is subjected to the back mapping processing of the dominant difference feature based on the deviation degree of each feature to determine the abnormal state code in the current abnormal purified water body.

[0045] Specifically, if the water quality result of the purified water body is an abnormal result, the difference vector is determined based on the Mahalanobis distance between the mean vector of the purified water quality feature map and the historical standardized feature map.

[0046] In one embodiment, when D>D0, the data processing unit further calculates the difference vector Δ=F_norm-F_base and analyzes Δ to achieve fine classification of water body anomalies.

[0047] Furthermore, the deviation degree based on the normal fluctuation amplitude is calculated for each feature value in the difference vector to obtain the deviation degree of each feature. Here, the deviation degree is the degree to which each feature value in the difference vector deviates from the benchmark value.

[0048] In one embodiment, for feature dimension contribution analysis, it is necessary to first calculate the degree to which each feature value in the difference vector Δ deviates from the benchmark value. For the j-th feature, its deviation δ_j is defined as |Δ_j| / sqrt(Σ_jj), where Σ_jj is the j-th diagonal element of the covariance matrix Σ (i.e., the standard deviation of the j-th feature). δ_j reflects the degree of deviation of the j-th feature relative to its normal fluctuation range. Finally, the deviations of all 72 features are sorted from largest to smallest, and the top 10 features with the highest deviations are selected as the dominant difference features.

[0049] Furthermore, the deviations of the 72-dimensional feature vectors are sorted, and the feature vectors corresponding to several feature values ​​in the top-ranked interval are identified as dominant difference features. These dominant difference features are the leading factors among the 72-dimensional feature vectors that cause abnormal water purification states. If the tracing path of the dominant difference feature is the time-domain statistical characteristic of a single sensing voltage information, then the number of the dominant difference feature is mapped inversely to the corresponding single sensing voltage signal type. If the tracing path of the dominant difference feature is the cross-correlation characteristic between two voltage signals, then the number of the dominant difference feature is mapped inversely to the sensing voltage signal type with a cooperative relationship.

[0050] In one embodiment, the dominant difference feature also needs to be reverse-mapped. That is, the number of the dominant difference feature is reverse-mapped back to the original sensor signal type and feature type; if the dominant difference feature mainly comes from the time-domain statistical characteristics of a certain sensor signal (such as the mean μ1, standard deviation σ1, etc. of the turbidity signal), it indicates that the water quality parameter corresponding to that sensor probe has changed significantly. If the dominant difference feature mainly comes from the cross-correlation characteristics between two signals (such as the correlation coefficient r between turbidity and ultraviolet absorption), it indicates that the dominant difference feature has changed significantly. 12 If the result is negative, it indicates that the synergistic relationship between these two water quality parameters has changed, which means that new types of pollutants have appeared in the water body or the existing pollutants have changed in their form.

[0051] Furthermore, based on the corresponding numbers of the sensor voltage signal types after reverse mapping, a signal code for each abnormal sensor voltage signal type in the current abnormal water body needs to be generated. The abnormality type code and deviation level corresponding to each abnormal sensor voltage signal type are then encoded according to the signal code framework to obtain the abnormal state code in the current abnormal water body. The abnormality type codes include: amplitude anomaly, fluctuation anomaly, and relational anomaly. The deviation levels include: slight deviation, moderate deviation, and severe deviation.

[0052] In one embodiment, anomaly codes are first generated based on the distribution of dominant differential features. The coding format is: signal code + anomaly type code + deviation level. Signal codes include: turbidity (T), conductivity (C), ORP (ORP), dissolved oxygen (O), UV absorption (U), and odor (G); anomaly type codes include: amplitude anomaly (L: significant deviation of the mean), fluctuation anomaly (V: significant deviation of the standard deviation), and correlation anomaly (M: significant deviation of the cross-correlation feature); deviation levels are: 1 (slight deviation, δ_j∈[3,5)), 2 (moderate deviation, δ_j∈[5,8)), and 3 (severe deviation, δ_j≥8). That is, anomaly code TL-2 indicates a moderate deviation in the mean of the turbidity signal; TUM-1 indicates a slight deviation in the correlation coefficient between the turbidity signal and the UV absorption signal, also indicating a change in the synergistic relationship between suspended particles and organic matter.

[0053] S105. Based on the abnormal state code, perform hierarchical directional backflow control on the corresponding current abnormal purified water body according to the relevant abnormal type code to obtain the directional backflow strategy of the current abnormal purified water body.

[0054] Specifically, by using the abnormality type code in the abnormality state coding, the sensor voltage signal type in the amplitude-type abnormal state is used for the directional diversion processing of the current abnormal water body, and the corresponding water quality purification unit allocation execution strategy is determined.

[0055] In one embodiment, when the abnormal status code includes an L-type anomaly, it indicates that the absolute value of a certain water quality parameter has deviated significantly. The water purification unit allocation strategy can be as follows: If the turbidity amplitude is abnormal (TL-*), the water is directed to the precision filtration unit. If the ultraviolet absorption amplitude is abnormal (UL-*), the water is directed to the activated carbon adsorption unit. If the odor amplitude is abnormal (GL-*), the water is directed to the activated carbon adsorption unit. If the conductivity or ORP amplitude is abnormal (CL-* or RL-*), the water is directed to the pH adjustment unit.

[0056] Furthermore, based on the type of sensor voltage signal in a fluctuating abnormal state, the current abnormal purified water is sent to the buffer averaging unit and returned to the sensor detection flow path for re-detection to obtain the water quality buffering execution strategy.

[0057] In one embodiment, when the abnormal status code includes a V-shaped anomaly, it indicates that the stability of a certain water quality parameter has been compromised, meaning that the parameter fluctuates drastically within a short period of time. This may be due to unstable operation of the purification equipment or the intermittent release of pollutants in the water. In this case, the water quality buffering strategy is as follows: the water is directed to the buffer homogenization unit, and after the water is fully mixed and homogenized by extending the hydraulic residence time (residence time ≥ 10 minutes), it is returned to the detection flow path for re-detection.

[0058] Furthermore, it is also necessary to automatically infer the failure of the water purification unit by the specific sensor signal combination involved in the sensor voltage signal type that is in a relational abnormal state, and send the current abnormal purified water to the corresponding series deep processing combination unit to obtain the water quality abnormal combination execution strategy.

[0059] In one embodiment, when the abnormal status code includes an M-type anomaly, it indicates a change in the synergistic relationship between two water quality parameters. For example, under normal circumstances, turbidity and ultraviolet absorption are highly positively correlated (r≈0.85). If a test reveals that the correlation coefficient between the two suddenly drops to r≈0.3 (corresponding to an M-type anomaly), it indicates an abnormal combination of high turbidity and low organic matter or low turbidity and high organic matter in the water body, meaning that a certain treatment unit of the purification system may have experienced selective failure. In this case, the water quality abnormal combination execution strategy can be as follows: based on the specific signal combination involved in the M-type anomaly, automatically infer the treatment unit that may have failed: if it involves a turbidity-ultraviolet absorption relationship anomaly (TUM-*), it may be a synergistic failure of the filtration unit and the adsorption unit; if it involves a turbidity-odor relationship anomaly (TGM-*), it may be a synergistic failure of the filtration unit and the deodorization unit; if it involves a conductivity-ORP relationship anomaly (CRM-*), it may be a failure of the acid-base adjustment unit; finally, the water body is directed to the corresponding series deep treatment combination unit, and the treated water body is returned to the detection flow path for re-detection.

[0060] Furthermore, when the abnormal state codes all include amplitude-type abnormal states, fluctuation-type abnormal states, and relational-type abnormal states, the current abnormal water body is detected and controlled in a fixed priority order to obtain a water quality composite abnormality execution strategy.

[0061] In one embodiment, when the abnormal state code simultaneously includes L-type, V-type, and M-type anomalies, the system can organize a series processing flow according to the priority order of M-type → L-type → V-type; that is, it can first resolve relational anomalies (restoring the synergistic relationship between various parameters of the water body), then resolve amplitude anomalies (reducing the parameters exceeding the standard range), and finally resolve fluctuation anomalies through buffering and homogenization. The directional backflow strategy includes: a water purification unit allocation execution strategy, a water quality buffering execution strategy, a water quality anomaly combination execution strategy, and a water quality composite anomaly execution strategy.

[0062] S106. The secondary purified water body that has completed the directional reflux strategy will be re-inspected and its status tracked to obtain comprehensive water quality test results.

[0063] Specifically, the maximum number of re-inspection cycles, N_max, is set to 3. During each re-inspection, the system records not only the current judgment result (qualified / abnormal), but also the changing trend of the Mahalanobis distance D between the feature map F_norm and the baseline F_base. That is, a re-inspection cycle is needed to judge the water quality results of the water body after the secondary purification process following the directional reflux strategy. If the water quality result of the newly purified water body is qualified, then the overall water quality test result is a qualified water quality test result.

[0064] If the newly purified water quality result is unqualified, and the Mahalanobis distance between the mean vectors of the newly purified water quality characteristic spectrum and the historical standardized characteristic spectrum decreases successively, then the overall water quality test result is the result of continuing the retesting cycle. In other words, if D decreases successively, meaning the spectrum gradually approaches the baseline, it indicates that the reflux treatment is effective, and the retesting cycle continues.

[0065] If the newly purified water quality result is unqualified, and the Mahalanobis distance between the mean vector of the newly purified water quality feature map and the historical standardized feature map remains unchanged or increases, then the comprehensive water quality test result is an abnormal result of water quality purification risk, and a maintenance warning signal will be sent.

[0066] In one embodiment, 1) if D no longer decreases or even increases after a retest, it indicates that the current reflux treatment strategy is ineffective, and the system automatically switches to an alternative reflux path (e.g., changing the water originally sent to the precision filtration unit to the activated carbon adsorption unit). 2) If D is still greater than D0 after 3 retest cycles, the system issues a maintenance warning signal, prompting the operator to check the working status of the deep treatment unit.

[0067] As a feasible implementation method, the purified water that has passed the assessment can be transported to the clean water storage module through the water supply pipeline. The clean water storage module has a built-in pressure stabilizing tank to stabilize the pressure of the stored water at 0.2MPa~0.3MPa, and then it is evenly transported to each water usage point through the water distribution pipeline.

[0068] In addition, this application also provides a water quality testing device based on the circulation and purification of cleaning wastewater, such as... Figure 2 As shown, the water quality testing equipment 200 based on the recycling and purification of cleaning wastewater specifically includes: At least one processor 201; and a memory 202 communicatively connected to the at least one processor 201; wherein the memory 202 stores instructions executable by the at least one processor 201 to enable the at least one processor 201 to execute: According to the sensing and detection flow path, the purified water that passes through the sensing and detection chamber in a laminar flow state will be simultaneously collected and processed by multi-source sensing signals of relevant water quality indicators, and a raw water quality signal matrix will be constructed. The time-domain statistical features and cross-correlation features between signals in the original water quality signal matrix are concatenated to obtain the purified water quality feature map. By using a pre-set database of qualified water body benchmark feature maps, the state of purified water body is judged based on map similarity matching, and the water quality results of the purified water body are obtained. If the water quality result of the purified water body is an abnormal result, then the water quality difference vector is subjected to the back mapping of the dominant difference feature based on the deviation degree of each feature to determine the abnormal state code in the current abnormal purified water body. Based on the abnormal state code, the corresponding current abnormal purified water body is subjected to hierarchical directional backflow control with relevant abnormal type codes to obtain the directional backflow strategy of the current abnormal purified water body; The water body that has undergone secondary purification after the directional reflux strategy has been re-inspected and its status tracked to obtain comprehensive water quality test results.

[0069] This application also provides a water quality detection system 300 based on the recycling and purification of cleaning wastewater, such as... Figure 3 As shown, a water quality monitoring system based on the recycling and purification of cleaning wastewater is used for online self-inspection and control of water quality for all parameters, including: The purified water quality status judgment module 301 is used to synchronously collect and process multi-source sensor signals of relevant water quality indicators of the purified water passing through the sensor detection cavity in a laminar flow state according to the sensor detection flow path, and construct a raw water quality signal matrix; the time-domain statistical features and cross-correlation features between signals in the raw water quality signal matrix are spliced ​​to obtain a purified water quality feature map; and the purified water quality feature map is judged based on map similarity matching using a preset qualified water body benchmark feature map library to obtain the purified water quality result. The purification water body anomaly detection module 302 is used to perform reverse mapping processing of the dominant difference feature based on the deviation degree of each feature on the water quality difference vector if the purification water quality result is an abnormal result, to determine the abnormal state code in the current abnormal purification water body; according to the abnormal state code, the corresponding current abnormal purification water body is subjected to hierarchical directional backflow control related to the abnormal type code to obtain the directional backflow strategy of the current abnormal purification water body. The re-inspection and circulation detection module 303 is used to track the re-inspection and circulation of the secondary purified water body after the directional reflux strategy has been implemented, and to obtain comprehensive water quality detection results. The circulating water supply execution module 304 is used to control the circulating water supply of the current purified water body that is in a qualified state according to the comprehensive water quality test results.

[0070] This application's embodiments construct a 72-dimensional feature map containing time-domain statistical features and cross-correlation features between signals. Using the Mahalanobis distance between the overall map and the baseline map as the judgment criterion, it can capture situations where individual indicators meet standards but the overall state is abnormal. By extracting cross-correlation features between signals, it can keenly capture changes in the synergistic relationship between different water quality parameters. Furthermore, by weighting the natural fluctuation amplitude of each feature dimension using a covariance matrix, it avoids misjudgments caused by large natural fluctuations in certain features. Through difference vector analysis and dominant difference feature identification, it further subdivides abnormal water states into three categories: amplitude-type anomalies, fluctuation-type anomalies, and relational anomalies, each corresponding to different recirculation processing strategies, achieving precise recirculation and significantly improving recirculation processing efficiency. Simultaneously, by utilizing the recorded judgment results in the re-inspection cycle and tracking the changing trend of the Mahalanobis distance, it can automatically adjust the recirculation strategy (switching alternative paths) based on the changing trend, achieving adaptive optimization of recirculation processing and avoiding the failure of fixed recirculation paths under specific abnormal situations.

[0071] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0072] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0073] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0078] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0079] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0080] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0081] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of this specification.

Claims

1. A water quality detection method based on the recycling and purification of cleaning wastewater, characterized in that, The method includes: According to the sensing and detection flow path, the purified water that passes through the sensing and detection chamber in a laminar flow state will be simultaneously collected and processed by multi-source sensing signals of relevant water quality indicators, and a raw water quality signal matrix will be constructed. The time-domain statistical features and cross-correlation features between signals in the original water quality signal matrix are concatenated to obtain a purified water quality feature map. By using a pre-set qualified water body benchmark feature map library, the purified water quality feature map is used to determine the state of the purified water body based on map similarity matching, and the purified water quality result is obtained. If the water quality result of the purified water body is an abnormal result, then the water quality difference vector is subjected to the dominant difference feature reverse mapping processing based on the deviation degree of each feature to determine the abnormal state code in the current abnormal purified water body. Based on the abnormal state code, the corresponding current abnormal purified water body is subjected to hierarchical directional backflow control with relevant abnormal type codes to obtain the directional backflow strategy of the current abnormal purified water body; The water body that has undergone secondary purification after the directional reflux strategy has been processed will be re-inspected and its status tracked to obtain comprehensive water quality test results.

2. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, Based on the sensing flow path, the purified water flowing through the sensing chamber in a laminar flow state undergoes simultaneous acquisition and processing of multi-source sensor signals related to water quality indicators, and a raw water quality signal matrix is ​​constructed, specifically including: The purified water enters the sensing and detection chamber in a laminar flow state through the honeycomb rectifier built into the inlet flow stabilization section of the sensing and detection flow path; when the purified water in the sensing and detection chamber is detected, six sets of sensor detection probes are activated; wherein, the sensing and detection flow path includes: the inlet flow stabilization section, the sensing and detection chamber, and the outlet buffer section; The purified water in the sensing and detection chamber is synchronously acquired using a turbidity sensor, a conductivity sensor, a redox potential sensor, a dissolved oxygen sensor, an ultraviolet absorption sensor, and an odor sensor to obtain multiple voltage signals. The odor sensor has different response sensitivities to sulfides, amines, hydrocarbons, and esters, and outputs four voltage signals. Based on the data window at the sampling time, the multi-channel voltage signals are processed by vector sorting to construct the original water quality signal matrix at multiple consecutive sampling times; wherein, the original water quality signal matrix is ​​used to represent the instantaneous response characteristics of the purified water body in 9 sensing dimensions.

3. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, The time-domain statistical features and cross-correlation features between signals in the original water quality signal matrix are concatenated to obtain a purified water quality feature map, specifically including: The mean, standard deviation, peak factor, and waveform factor of each voltage signal sequence in the original water quality signal matrix are calculated, and time-domain statistical feature values ​​are statistically extracted; wherein, the number of terms in the time-domain statistical feature values ​​is 36. The Pearson correlation coefficient between each pair of voltage signals in the original water quality signal matrix is ​​calculated to obtain the correlation coefficient matrix; wherein, the correlation coefficient matrix represents the cooperative change relationship between responses of different sensing dimensions; Extract the upper triangular portion of the correlation coefficient matrix and determine the correlation coefficient values ​​in the correlation coefficient matrix as cross-correlation feature values; the number of terms in the cross-correlation feature values ​​is 36. The time-domain statistical feature values ​​and the cross-correlation feature values ​​are combined to generate the purified water quality feature map; wherein, the purified water quality feature map is a 72-dimensional feature vector.

4. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 3, characterized in that, After concatenating the eigenvalues ​​of the time-domain statistical features and the cross-correlation features between signals in the original water quality signal matrix to obtain the purified water quality feature map, the method further includes: Each feature value in the purified water quality feature spectrum is dimensionless based on the reference scale parameter to obtain a dimensionless feature value. By using the hyperbolic tangent function, the dimensionless eigenvalues ​​are subjected to interval mapping to obtain all eigenvalues ​​within a uniform amplitude range; Based on the dimensionless eigenvalues ​​and all eigenvalues ​​within a uniform amplitude range, a purified water quality characteristic spectrum is obtained after standardization and normalization.

5. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, Using a pre-defined database of qualified water body benchmark feature maps, the purified water quality feature maps are analyzed based on map similarity matching to determine the state of the purified water body, resulting in the purified water quality outcome, specifically including: Based on historically continuously collected qualified water body sensor data, a historical standardized feature map is constructed; and the mean vector and covariance matrix of the historical standardized feature map are calculated. The qualified water body benchmark feature map library is constructed by using the benchmark feature map center of the qualified water body corresponding to the mean vector and the normal fluctuation range of the qualified water body feature map corresponding to the covariance matrix. Based on the weighted processing of the fluctuation amplitude of each feature dimension by the covariance matrix, the Mahalanobis distance is obtained between the mean vector of the purified water quality feature map after standardization and normalization and the historical standardized feature map. If the current Mahalanobis distance is less than or equal to the preset Mahalanobis distance, then the similarity between the purified water quality feature map of the current purified water body and the benchmark feature map of qualified water body is within the normal fluctuation range, and the purified water body is determined to be qualified; wherein, the preset Mahalanobis distance is the Mahalanobis distance distribution of qualified purified water body samples. If the current Mahalanobis distance is greater than the preset Mahalanobis distance, then the similarity between the purified water quality feature map of the current purified water body and the benchmark feature map of qualified water body is significantly different, and it is determined that the purified water body is an abnormal water quality result. The water quality results of the purified water body include: abnormal water quality results and qualified water quality results.

6. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, If the water quality result of the purified water body is an abnormal result, then the water quality difference vector is subjected to back-mapping processing based on the dominant difference feature deviation degree of each feature to determine the abnormal state code in the current abnormal purified water body, specifically including: If the purified water quality result is an abnormal result, then the difference vector is determined based on the Mahalanobis distance between the mean vector of the purified water quality feature map and the historical standardized feature map. For each feature value in the difference vector, the deviation degree based on the normal fluctuation amplitude is calculated to obtain the deviation degree of each feature; wherein, the deviation degree is the degree to which each feature value in the difference vector deviates from the benchmark value; The deviation of the 72-dimensional feature vectors is sorted, and the feature vectors corresponding to several feature values ​​in the first interval of the sort are determined as the dominant difference features; wherein, the dominant difference features are the dominant factor features in the 72-dimensional feature vectors that lead to the abnormal state of water purification. If the tracing path of the dominant difference feature is the time-domain statistical feature of a single sensing voltage information, then the number of the dominant difference feature is mapped in reverse to the corresponding single sensing voltage signal type. If the tracing path of the dominant difference feature is the cross-correlation feature between two voltage signals, then the number of the dominant difference feature is mapped in reverse to the sensing voltage signal type with a cooperative relationship. Based on the corresponding number of the sensor voltage signal type after reverse mapping, a signal code is generated for each abnormal sensor voltage signal type in the current abnormal water body; and the abnormal type code and deviation level corresponding to each abnormal sensor voltage signal type are formatted and encoded according to the framework of the signal code to obtain the abnormal state code in the current abnormal water body; wherein, the abnormal type code includes: amplitude anomaly, fluctuation anomaly and relational anomaly; the deviation level includes: slight deviation, moderate deviation and severe deviation.

7. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, Based on the abnormal state code, the corresponding current abnormal purified water body is subjected to hierarchical directional backflow control according to the relevant abnormal type code, resulting in a directional backflow strategy for the current abnormal purified water body, specifically including: Based on the abnormality type code in the abnormality state code, the sensor voltage signal type in the amplitude abnormality state is subjected to directional diversion processing of the current abnormal water body, and the corresponding water quality purification unit allocation execution strategy is determined. Based on the sensor voltage signal type in a fluctuating abnormal state, the current abnormal purified water is sent to the buffer averaging unit and returned to the sensor detection flow path for re-detection to obtain the water quality buffering execution strategy. The specific sensor signal combination involved in the sensor voltage signal type that is in a relational abnormal state is automatically inferred and processed by the failed water purification unit, and the current abnormal purified water body is sent to the corresponding series deep processing combination unit to obtain the water quality abnormal combination execution strategy. When the abnormal state codes all include amplitude-type abnormal states, fluctuation-type abnormal states, and relational-type abnormal states, the current abnormal water body is detected and controlled in a fixed priority order to obtain a water quality composite abnormality execution strategy. The directional reflux strategy includes: the water purification unit allocation execution strategy, the water quality buffer execution strategy, the water quality anomaly combination execution strategy, and the water quality composite anomaly execution strategy.

8. The water quality detection method based on the recycling and purification of cleaning wastewater according to claim 1, characterized in that, The purified water body after the directional reflux strategy has been subjected to re-inspection, circulation, and status tracking to obtain comprehensive water quality test results, specifically including: The water quality results of the secondary purified water body after the directional reflux strategy has been implemented will be re-examined and cyclically judged. If the water quality result of the newly purified water body is a qualified result, then the comprehensive water quality test result is a water quality test pass result; If the water quality result of the newly purified water body is unqualified, and the Mahalanobis distance between the mean vector of the newly purified water quality feature spectrum and the historical standardized feature spectrum is a result of decreasing successively, then the comprehensive water quality test result is the result of continuing to perform the retest cycle. If the newly purified water quality result is unqualified, and the Mahalanobis distance between the mean vector of the newly purified water quality feature map and the historical standardized feature map remains unchanged or increases, then the comprehensive water quality test result is an abnormal result of water quality purification risk, and a maintenance warning signal is sent.

9. A water quality testing device based on the circulation and purification of cleaning wastewater, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, enabling the at least one processor to perform a water quality detection method based on the recycling and purification of cleaning wastewater according to any one of claims 1-8.

10. A water quality detection system based on the recycling and purification of cleaning wastewater, characterized in that, A water quality monitoring system based on the recycling and purification of cleaning wastewater is used for online self-inspection and control of water quality for all parameters, including: The purified water quality status judgment module is used to synchronously collect and process multi-source sensor signals of relevant water quality indicators of purified water passing through the sensor detection cavity in a laminar flow state according to the sensor detection flow path, and construct a raw water quality signal matrix; the module then performs feature value splicing between the time-domain statistical features and the cross-correlation features between signals in the raw water quality signal matrix to obtain a purified water quality feature map; and finally, the module performs a purified water quality status judgment based on map similarity matching on the purified water quality feature map using a preset qualified water body benchmark feature map library to obtain the purified water quality result. The purified water body anomaly detection module is used to determine the abnormal state code in the current abnormal purified water body by performing a back-mapping process on the water quality difference vector based on the dominant difference feature deviation degree of each feature if the water quality result of the purified water body is an abnormal result; and to perform hierarchical directional backflow control on the corresponding current abnormal purified water body according to the abnormal state code, so as to obtain the directional backflow strategy of the current abnormal purified water body. The re-inspection and cycle detection module is used to track the secondary purified water body that has completed the directional reflux strategy, perform re-inspection and cycle detection, and track its status to obtain comprehensive water quality detection results. The circulating water supply execution module is used to control the circulating water supply of the currently purified water body that is in a qualified state according to the comprehensive water quality test results.

Images